INTRODUCTION TO

BIOINFORMATICS
INTRODUCTION TO
BIOINFORMATICS

Angshuman Bagchi

α
Alpha Science International Ltd.
Oxford, U.K.
Introduction to Bioinformatics
168 pgs.

Angshuman Bagchi
Department of Biochemistry and Biophysics
University of Kalyani
Nadia

Copyright © 2018
ALPHA SCIENCE INTERNATIONAL LTD.
7200 The Quorum, Oxford Business Park North
Garsington Road, Oxford OX4 2JZ, U.K.

www.alphasci.com

ISBN 978-1-78332-377-7

All rights reserved. No part of this publication may be reproduced, stored in

a retrieval system, or transmitted in any form or by any means, electronic,
mechanical, photocopying, recording or otherwise, without prior written
permission of the publisher.
 Preface

Bioinformatics is an interdisciplinary subject which uses the knowledge

of Biology, Physics, Chemistry, Statistics, Mathematics and Computer
Science. It deals with the building and development of Biological
Databases and Software tools. It uses the power of computations to analyze
and solve biological problems. The prerequisites for Bioinformatics study
are good quality computers with high processing speeds. However,
the analyses and interpretations of the Biological data require human
intervention. Bioinformatics is largely but not exclusively a computer
based study. Computers are needed to perform the repetitive tasks that are
at the core of many Bioinformatics analyses. Computers are also needed
for their abilities to solve problems. The bioinformatics study focuses
primarily on the following aspects:
• Database creation for storage and management of large biological
data
• Development of algorithms and statistics to analyze the
relationships between various types of biological data
• Use of the software tools to come up with new biological
hypothesis and to provide biological insights into the biological
phenomenon
The use of bioinformatics in analyzing the biological data became popular
in the early 1990s. However, previously bioinformatics study was confined
mainly to the management of biological sequence information. The
popularity of bioinformatics as an independent branch of science came
into reckoning after the advancement of sequencing technologies. This
led to a rapid outburst of biological sequence data especially nucleotide
vi Preface

sequence data which are subsequently being stored in GenBank.

Bioinformatics to a large extent depends on computational power.
However, other branches of science like Physics, Statistics, Mathematics,
and Chemistry play equally important roles in analyzing the biological
data by computational means. Thus, an attempt has been made to come up
with an overview of the important aspects of this interdisciplinary science.
This book is meant for the Bachelors and Masters students of
Bioinformatics. The second chapter of the book deals with the popular
and important biological databases. The third chapter focuses on the
biological sequence alignment methods. The fourth chapter presents
the molecular modelling techniques and the last chapter presents the
Phylogenetic analyses.
The details of some of the important web tools and web servers are also
presented. I incorporated some multiple choice questions for competitive
examination.
At the end I would like to wish the readers all the very best.
 Contents

Preface v
1. Introduction 1.1
1.1 History of Bioinformatics 1.2
1.2 Biological Databases 1.3
1.3 Biological Sequence Analysis 1.6
2. Biological Databases
2.1 Introduction 2.1
2.2 Some important Terminologies and Definitions 2.1
2.3 Primary Sequence Databases 2.3
2.4 The Nucleotide Sequence Databases 2.3
2.5 Final Words 2.9
2.6 The Protein Sequence Databases 2.9
2.7 The Protein Structure Database 2.15
2.8 Combination Databases 2.18
Some Important Questions and Answers 2.23
3. Biological Sequence Alignment
3.1 Introduction 3.1
3.2 Some Important Terms 3.1
3.3 Biological Importance of Sequence Alignment 3.3
viii Contents

3.4 Sequence Alignment Algorithms 3.4

3.5 Structural alignment 3.7
Some Important Questions and Answers 3.13
4. Molecular Modelling
4.1 Introduction 4.1
4.2 Experimental Methodologies 4.32
4.3 Solutions and Recommendations 4.39
Some Important Questions and Answers 4.41
5. Phylogenetic Analysis
5.1 Introduction 5.1
5.2 Phylogenetic Analysis 5.1
5.3 Some Important Terminologies and Definitions 5.4
5.4 Some Statistical Aspects in Phylogenetic s 5.7
5.5 Different Aspects of Phylogenetic Studies 5.9
5.6 Positing of Root Node in a Phylogenetic Tree 5.10
5.7 Steps for Construction of Phylogenetic Trees 5.11
5.8 The Reliability of a Phylogenetic Analysis 5.13
5.9 Determination of Evolutionary Rate and Time 5.15
Some Important Questions and Answers 5.18

Few More Important Questions and Answers QA.1

Further Reading F.1
Index I.1
 CHAPTER

1 Introduction

Bioinformatics is a multi-disciplinary science. The study of Bioinformatics

is related to solving problems of molecular biology and genetics with the
help of computational tools. In doing so, concepts from computer science,
mathematics and statistics, physics and chemistry are considered. The
basic framework of Bioinformatics lies on the use of computational tools
to solve biological problems which are data intensive and often incorporate
large-scale biological information. Bioinformatics takes the help of
the existing data that are available and tries to extract new information
from them. In general the following classes of problems are dealt in
Bioinformatics:
• Analyses of biological sequences to extract new information from
them
• Annotations of genome sequences
• Identification of phylogenetic lineages
• Modelling of the biological pathways
• Prediction of three-dimensional structures and functions of
biological molecules
• Simulations of biological molecules under different physiological
conditions
• Making repositories of collected biological information
The Bioinformatics analyzes the biological data in several steps.
However, the following steps can be considered as the general ones:
1.2 Introduction to Bioinformatics

• Collection of biological sequence information

• Building a computational model from the collected sequence
information
• Testing the generated model for real life biological data
In this chapter, the basic aspects of Bioinformatics will be discussed.
The interrelationships of Bioinformatics with other scientific fields will
be considered. The important terminologies in Bioinformatics will be
analyzed.

1.1 HISTORY OF BIOINFORMATICS

After the discovery of DNA structure and protein sequencing (1950-

70s), protein and nucleotide sequence storing and alignments were
being performed manually. A major advancement in the field of
molecular biology in the last two decades (1980-2000s) brought about
the exponential increase of biological data which need a time and cost
effective technology to manage all these. Along with this there was also
an increased ratio between performance and cost of computer hardware.
Hence, the field of Bioinformatics evolved and took a central role in
modern day biological science.

1.1.1 Computational System Biology

The better understanding of complex biological systems requires a
computational biology platform that applies the techniques of computer
science, applied mathematics and statistics to solve the biological
problems. After mid-1990s the Human Genome Project and rapid advances
in DNA sequencing technology culminated in explosive growth in the
field of computational biology. This necessitates the analyses of biological
data to produce meaningful information using algorithms from graph
theory, artificial intelligence, soft computing, data mining, and image
processing and computer simulation. The advent of databases and their
principles led to the concepts of sequence comparisons. Sequential
alignments led to three-dimensional model building, validation of models
and interaction studies.
 Introduction 1.3

Computational biology involves:

• Prediction of three-dimensional structures of proteins and nucleic
acids
• Prediction of binding interfaces of biological macro or micro
molecules
• Interaction study between biological molecules
• Functional domain/motif prediction
• Computational phylogentics
• Micro RNA reglation and post transcriptional modification
• Mutational analysis

1.2 BIOLOGICAL DATABASES

The rapid progress in gene and protein identification and sequencing led
to the collection of information related to genomics, transcriptomics,
proteomics, metabolomics and this required those data to be stored
which can be accessible and later updated and retrieved easily from all
over the world.

1.2.1 Primary Database

A database consisting of data derived directly from experiments such
as nucleotide or protein sequences and three-dimensional structures of
protein/nucleic acid are known as primary database. There are several
types of primary databases
A. Genome Database
1. Sequence Database – NCBI, GenBank, DDBJ, EMBL, TrEMBL
2. Structural Database – Nucleic Acid Database (NDB), RNABase
B. Protein Database
1. Sequence Database – Swiss-Prot
2. Structural Database – Protein Data Bank (PDB)
1.4 Introduction to Bioinformatics

Brief aspects of some important databases are presented in the

following section. The detailed analyses of the databases are given in
the respective chapters.
NCBI-The National Center for Biotechnology Information (NCBI):
It is the part of the United States National Library of Medicine (NLM), a
branch of the National Institutes of Health (NIH). NCBI maintains a series
of databases relevant to biomedicine and is an important resource for
bioinformatics tools and services. Major databases include GenBank for
DNA sequences and PubMed, a bibliographic database for the biomedical
literature.
PDB comprises of
• Protein Data Bank of Europe (PDBe)
• Protein Data Bank of Japan (PDBj)
• Research Collaboratory for Structural Bioinformatics in USA
(RCSB)
The three dimensional atomic coordinates of biological
macromolecules are deposited in PDB, which are obtained by X-Ray
crystallography, NMR spectroscopy or Cryo-Electron Microscopy (EM)
around the globe. The structure files may be viewed using one of several
free and open source computer programs as well as paid software tools
like Jmol, Pymol, Rasmol, VMD, UCSF Chimera, Swiss-PDB Viewer
and Discovery Studio.
Swiss-Prot: SWISS-PROT is an annotated protein sequence database
having an equal partnership between the European Molecular Biology
Laboratory (EMBL) and the Swiss Institute of Bioinformatics (SIB). 
The SWISS-PROT database distinguishes itself from other protein
sequence databases by three distinct criteria: (i) annotations, (ii) minimal
redundancy and (iii) integration with other databases.

1.2.2 Secondary Database

It is also known as curated database which includes the results of analysis
of primary databases and other significant data in the form of conserved
sequences, signature sequences, secondary structures, active site residues
 Introduction 1.5

of proteins etc. In the following section, a brief analysis of some of the

important secondary databases are presented.
RefSeq: The Reference Sequence (Refseq) database is an annotated,
non-redundant and curated collection of nucleotide sequences and their
protein products.
UniProtKB: It is the high quality annotated non-redundant protein
sequence database, generated by manual interventions, which brings
together experimental results, computed features and scientific conclusions
and reviewed section of the UniProt Knowledgebase (UniProtKB).
Prosite: A database of protein domains, families and functional sites
as well as associated patterns and profiles to identify them.
Pfam: Protein family database based on alignments and HMMs.
Pfam alignments are available from searching a variety of databases with
different accession numbers (e.g. all UniProt and NCBI).
InterPro: It provides a classification among protein families, motifs
and domains on functional basis to predict functional domain or site.
SCOP: The Structural Classification of Proteins (SCOP) database is
largely a manual classification of protein structural domains based on
similarities of their structures and amino acid sequences.
CATH: A database of Protein Structure Classification which provides
information on the evolutionary relationships of protein domains.
Dali: The Dali Database and server is based on all-against-all 3D
structure comparison and FSSP structural classification of protein
structures in the Protein Data Bank (PDB).

1.2.3 Composite Database

This type of database is a combination of many other databases.
The following examples are some of the most important composite
databases.
NRDB: It is a so-called non-redundant composite database of the
following sources: PDB, SWISS-PROT, SWISS-PROT update, PIR,
GenPept and GenPept update.
1.6 Introduction to Bioinformatics

OWL: It is a non-redundant composite database of 4 major publicly-

available primary sources: SWISS-PROT, PIR, GenBank (translation)
and NRL-3D.

1.3 BIOLOGICAL SEQUENCE ANALYSIS

Since the sequencing of bacteriophage ØX-174 in 1977, the nucleotide

sequences of thousands of organisms have constantly been decoded and
thereby getting stored in different biological databases. The nucleotide
sequences of the organisms are being analyzed by various scientific groups
to decode the underlying information. This is called genome annotations.
The most common annotations involve:
• identifications of protein coding genes
• identifications and analyses of genes that lead to RNA biogenesis
• detection of the presence of regulatory sequences, structural motifs
and repetitive sequences in the genes.
The nucleotide information of genes within a species or between
different species can be compared to identify the similarity between the
gene products, viz., the proteins, RNAs or the nucleotides themselves.
The databases contain large amounts of data. The genome information
of an organism is also very huge. With such large chunks of data,
it is simply impractical to analyze biological sequences manually.
Thus, Bioinformatics tools are routinely used to analyze the sequence
information. The most widely used tool for genome analyses is BLAST.
The method used for the purpose is called pair-wise sequence alignment.
However, there are other tools available that provide different sets
of results. A common practice in genome analyses is to compare the
characteristics of different genomes. For such purposes, the technique
that is used is called the multiple sequence alignment.

1.3.1 Pair-wise Sequence Alignment

Dayhoff et al., in 1978 studied the sequence alignments among
34 superfamilies of protein sequences and listed the accepted point
mutations i.e., replacement of amino acid by another residue by natural
 Introduction 1.7

selection. Based on accepted mutation and probabilitities of occurrence

of each amino acids they generated a mutation probability matrix
known as PAM. PAM1 matrix was generated for closely related protein
sequences having atleast 85% sequence identities. Later the PAM1 matrix
is multiplied by iteslf for several times to generate several other PAM
matrices. One of such derived matrices, the PAM250 matrix, is used for
sequence alignment of proteins with 20% amino acid sequence identitities.
In other words, the PAM250 matrix is used for sequnce comparison
of evolutionarily distantly related proteins. PAM mutation probability
matrix is then converted into a scoring matrix called log-odds matrix or
relatedness odds matrix. Another log-odds matrix developed by Henikoff
and Henikoff (1992, 1996) is called BLOcks SUbstitution Matrix
(BLOSUM) using BLOCKS database which consists of more than 500
groups of local alignments of distantly related protein sequences. High
value of BLOSUM (BLOSUM90) and low value PAM matrices (PAM30)
are used for conserved protein sequences whereas low BLOSUM and
high PAM numbers are used for distantly related proteins. BLOSUM62
matrix merged 62% or more identical sequences into one alignment;
thus proteins having less than 62% sequnence identities are analyzed
by this BLOSUM62 matrix and it is the default scoring matrix used by
most BLAST algorithms for searching sequence similarities between
sequences. Two types of pair-wise sequence alignments are :
(i) Global alignment technique: It is derived from the Needleman–
Wunsch algorithm (1970), which is based on dynamic
programming which attempts to align every residue in every query
sequence. It is most useful when the sequences in the query set
are similar and of roughly equal length.
(ii) Local alignment technique: It is more useful for dissimilar
sequences that are suspected to contain regions of similarity or
similar sequence motifs within their larger sequence contexts.
The Smith–Waterman algorithm (1981) is a general local
alignment method which is also based on dynamic programming
algorithm, for scoring the matrix and it uses trace-back procedure
to obtain the optimal alignment.
1.8 Introduction to Bioinformatics

Dynamic programming: Alignments of amino acid sequences of

proteins use a substitution matrix to assign scores to amino acid matches
or mismatches, and a gap penalty for matching an amino acid in one
sequence to a gap in the other. The nucleotide sequences of DNA and
RNA may be used in scoring the matrix. In practice, a simple scoring
scheme is generally used to score the sequence alignments; a positive
score is used for a match, a less positive or a negative score for mismatch
and a negative score for gap penalty.
Though the local alignment is accurate but it is relatively slow when
pairwise local alignment is used for a query sequence against an entire
database. The computational running space and time then become a
significant issue. Two other popular rapid, heuristic version of local
alignment algorithms have been developed based on identifying short
words or k-tuples. Such processes would involve 1/2 residues for protein
and upto 6 bases for DNA searching. The locally aligned regions are
called high-scoring segment pairs or HSPs. The expect (E) value is the
measure of statistical significance which represents the number of hits one
can expect to obtain by chance; that means the lower the E-value is the
more significant is the alignment. The E-value decreases exponentially
with high score (S) value corresponding to better alignments.
• Word methods: The heuristic methods are significantly more
efficient than dynamic programming. These methods are
especially useful in large-scale database searches and best known
for their implementation in the database search tools in a number
of web portals, such as EMBL FASTA and NCBI BLAST. NCBI
BLAST program is a collection of different tools like blastp and
PSI-BLAST.
• blastp: This program compares a protein query sequence against
a protein sequence database. In this software program the default
matrix is BLOSUM62 with a word size 6, gap existence 11 and
extension value 1.
• PSI-BLAST: This blast program is specilized, advanced and
more sensitive position-specific BLAST search which is used
for distantly related proteins sharing the same conserved three
 Introduction 1.9

dimensional structures but low sequence identities. An initial

blastp search is used to perform a multiple sequence alignment.
By analyzing this alignment it then creates a specialized position-
specific scoring matrix which again is used as query iteratively.
The orignial query is used as template for generating the PSSM
profile.

1.3.2 Multiple Sequence Alignment

The alignments of more than two protein or DNA sequences are very
useful to study the homologous group and it provides structure-function
and evolution of the protein or gene families. It is more sensitive than
pairwise alignment to detect homologs, to find conserved domain or
motifs or consensus regions. The most commonly used algorithm for
multiple sequence alignment is derived from Feng and Doolittle’s (1987,
1990) progressive alignment method. The basic strategy of this method
is calculaing pairwise alignment scores between all sequences and then
align two closely related sequences followed by addition of sequences
progressively to the alignment. The most popular web version of multiple
sequence alignment tool is ClustalW program. To generate a guide tree,
this program uses mostly distance matrix instead of similarity matrix.
The two main features of a guide tree are its branching order or topology
and branching length which is proportional to evolutionary distance. The
two main methods of tree construction are:
(a) unweighted pair group method of arithmatic averages (UPGMA)
and
(b) neighbour-joining method.
The multiple sequence alignment output arranges the sequences as
presented in the guide tree, i.e., two closely related sequences create a
pairwise alignment and then sequences are added one by one. The newer
web version of multiple sequence alignment is Clustal Omega which
uses seeded guide trees and HMM profile-profile techniques to generate
alignments.
1.10 Introduction to Bioinformatics

1.3.3 Molecular Phyllogeny

Previously the classification and phylogenetic studies were done
based upon morphological, anatomical, physiological, paleontological
characteristics. But after the huge progress in molecular biology the
evolutionary relationships among different species have been being
studied using their protein or nucleotide sequences. A phylogenetic tree
is a graphical representation of relatedness among sequences or ancestral
origin of the sequences. The node is the intersection or terminating point
of the tree which represents the taxonomic units (taxa or taxons, can be
internal node or external node). An operational taxonomic unit (OTU) is
the available sequence which is used for tree analysis. Branches define
the relatedness of the sequences and branch length sometimes are scaled
to represent the number of changes occurring between sequences. The
tree generated from scaling is called a phylogram. On the other hand,
the trees may be unscaled where the branch length is not proportional
to the number of changes. Such trees are called cladogram. A clade or
monophyletic group is a group of sequences that share a common ancestor.
A rooted tree represents the common ancestor whereas unrooted tree
does not define the evolutionary path leading to their common ancestors.
A phylogenetic tree generation using molecular sequence data involve
multiple sequence alignment using different approaches, viz., distance-
based methods or character based methods.
In distance based methods the tree is constructed by calculating the
distances between molecular sequences. In the construction of guide
tree in multiple sequence alignment, the distance based methods used
are unweighted pair group method of arithmatic mean (UPGMA) and
neighbor-joining (NJ). UPGMA is simple tree making algorithm where
the sequences make cluster based on distance matrix and it is always a
rooted tree.
The two main character based methods are maximun parsimony (MP)
and maximum likelihood (ML). Maximum parsimony method is applied
where the sequences are closely identical thereby building a tree with
shortest branch lengths.
 Introduction 1.11

Maximum likelihood method on the other hand uses the knowledge

of speciation to produce sequence clusters.
1.3.3.1 Neighbour Joining (NJ) Method

A star like tree is generated and then pairwise comparisons are made to
identify two closely related sequences, the neighbors, followed by joining
them until the topology of the full tree is completed.
1.3.3.2 Maximum Likelihood (ML) Method

It is the most computationally intensive method but the most flexible

method when large amounts of evolutionary changes are found in the
distantly related sequences. It uses the probability distribution statistical
method to generate the highest probable tree. The mutation of each
residue calculated is based on substitution model where probability of
each mutation is calculated and are then aligned.
1.3.3.3 Tree Evaluation and Bootstrapping

To evalute the tree, the most common method applied is randomizing

the test or bootstrapping analysis. Bootstrapping analysis describes the
robustness of the topology of a tree that means it checks the consistency
of the branching order. After developing the tree, the positions of the
aligned sequences are randomly sampled from the multiple sequence
alignment with replacements and then are assembled into new randomized
data sets, the so-called bootstrapped samples. A large number of bootsrap
replicates are generated. The bootsrap tree is compared with the original
tree and the bootstrap value denotes the frequencies of observed data to
obtain each clade.
Protein Domain Identification and Mutational Analysis: A
conserved protein family represents the functional aspect of the amino
acid sequences. Multiple alignment of the sequences obtained from the
blastp search would find the conserved functional domains. Multiple
sequence alignment (MSA) reveals some point mutations within protein
domains. Some mutations are   synonymous substitutions and semi-
conservative mutations in the MSA profile. Point mutations were identified
from the particular sequences and then short streches of sequences are
1.12 Introduction to Bioinformatics

again searched using blastp and new functional domains are found. The
relative mutational probabilities of different amino acid residues are
different according to their genetic codes. Some amino acid residues
such as asparagine and serine undergo substitutions very frequently,
while other residues (notably tryptophan and cysteine) are very rarely
mutated. Dayhoff et al calculated the relative mutabilities of amino acid
residues which describe how often each amino acid is likely to change
over a short evolutionary period. To calculate the relative mutability they
divided the number of times each amino acid was observed to mutate by
the overall frequency of occurrence of that amino acid. The less mutable
residues probably have important structural and functional roles in
proteins such that replacing them resulted in some significant changes.
The most common substitutions are Glu for Asp (both acidic), Ser for
Ala and Thr (both are hydroxylated), and Ile for Val (both hydrophobic
and of similar sizes). The relative mutabilities depend on to the physico-
chemical properties of the amino acids. With reference to the genetic code
it is also explained in such a way that common amino acid substitutions
tend to require only single nucleotide change and least mutable residues
are coded by only one or two codons. The low mutability of these amino
acids suggest that substitutions are not tolerated by natural selections.
Secondary Structure Prediction: Secondary structure prediction
techniques aim to predict the local secondary structures of proteins based
on the knowledge of their sequences. The first generation secondary
structure prediction methods in 1960-1970s were based on probabilities
of a particular amino acid for a particular secondary structure. The method
is referred to as Chou-Fasman method. The second generation methods
untill early 1990s consider the local environment of the adjacent residues
typically 3-51 segments. Modern protein secondary structure prediction
methods are based on exploiting evolutionary information contained
in multiple sequence alignments such as recognizing the patterns of
hydrophobicity, conservation, sequence edge effects etc. In relation to
the databases of known protein structures and modern machine learning
methods, such as neural networks, multiple linear regressions, k-nearest
neighborhood and support vector machines, the accuracy of secondary-
structure prediction is raised up to 80% for globular proteins. Further boost
 Introduction 1.13

up in prediction accuracy is highly limited by varying local conformations

under native conditions which are not reflected in crystal structures
due to packing constraints and local contacts. In addition, dramatic
conformational changes related to the protein’s function or environment
could complicate the situation a bit more. There are a number of
secondary structure prediction servers such as Network Protein Analysis
Server (NPS@). At NPS@, a blastp search of the sequence is performed
against the SWISS-PROT database. These results are filtered and then
aligned by CLUSTAL Omega. The resulting alignment is the input for
the following methods: GOR I, GOR III, GOR IV, DSC, DPM, PHD,
Predator, SOPM, HNN. GOR (Garnier, Osguthorpe and Robson) method
is one of the popular methods utilizing information theory that predict
locations of alpha-helix and beta-strand using amino acid sequence.
DSC (Discrimination of protein Secondary structure Class) is based on
dividing secondary structure prediction into the basic concepts and then
use of simple and linear statistical methods to combine the concepts for
prediction. DPM method has been called the ‘double prediction method’
and consists of a first prediction of the secondary structure from a new
algorithm which uses parameters of the type described by Chou and
Fasman, and the prediction of the class of the proteins from their amino
acid composition. PHD program achieved 70% accuracy level through
combining neural networks and evolutionary information. PREDATOR
method is based on recognition of potentially hydrogen-bonded residues in
a single amino acid sequence. Self-optimized prediction method (SOPM)
checks against an updated release of the Kabsch and Sander database,
‘DATABASE.DSSP’. The first step of the SOPM is to build sub-databases
of protein sequences and their known secondary structures drawn from
‘DATABASE.DSSP’. The second step is to submit each protein of the
sub-database to a secondary structure prediction tool using a predictive
algorithm based on sequence similarity. The third step is to iteratively
determine the predictive parameters that optimize the prediction quality
on the whole sub-database. The HNN (Hierarchical Neural Network)
prediction methods are made up of two networks: a sequence-to-structure
network and a structure-to-structure network. Kabsch and Sander in 1983,
developed the Dictionary of Protein Secondary Structure known as DSSP
1.14 Introduction to Bioinformatics

which is a set of simple and physically motivated criteria for secondary

structure prediction, programmed as a pattern-recognition process
of hydrogen-bonded and geometrical features extracted from X-Ray
crystallographic coordinates of the atoms of biomolecules. Cooperative
secondary structure is recognized as repeats of the elementary hydrogen-
bonding patterns “turn” and “bridge.” Repeating turns are “helices,”
repeating bridges are “ladders,” connected ladders are “sheets.” Geometric
structure is defined in terms of the concepts of torsion and curvature of
differential geometry. Local chain “chirality” is the torsional handedness
of four consecutive C-alpha positions and is positive for right-handed
helices and negative for ideal twisted beta-sheets. Curved pieces are
defined as “bends.”
Trans-membrane Helix Prediction: Experimental determination
of the 3D structure of transmembrane protein is a challenge as they do
not crystalize and are hardly tractable by NMR spectroscopy. Two major
classes of membrane proteins are known as:
(a) proteins which insert alpha helices into lipid bilayer (intergral)
and
(b) proteins that form pores by beta strand barrels (porins).
There is not much experimental information available regarding
porins like strand barrels. However, knowing the precise location of
transmembrane helics, the 3D structure can be predicted by exploring all
possible conformations. The basic properties of transmembrane helices
(TMHs) found are:
1. Trans-membrane helices are predominantly apolar and between
12 and 35 residues long.
2. Globular regions between membrane helices are typically shorter
than 60 residues.
3. Most proteins having trans-membrane helical regions have
a specific distribution of the positively charged amino acids
Arginine and Lysine coined the “positive-inside-rule”. Connecting
loop regions at the inside of the membrane have more positive
charges than loop regions at the outside.
 Introduction 1.15

4. Long globular regions (>60 residues) differ in their composition

from those globular regions subject to the “inside-out-rule”.
Most methods for predictions of the structures of trans-membrane
helices simply compile the hydrophobicity along the sequence and predict
a segment to be a trans-membrane helix if the respective hydrophobicity
exceeds some given threshold. Use of evolutionary information also
improves the prediction abilities of trans-membrane helices significantly
by carefully filtering the results from PSI-BLAST searches. The trans-
membrane topology was predicted from the amino acid sequences by
averaging the results from six different programs: DAS, HMMTOP,
TMHMM, TMPRED, TOPPRED II and GENEIOUSPRO. Dense
Alignment Surface (DAS) optimizes the use of hydrophobicity plots.
TMHMM is the most advanced and seemingly the most accurate current
method to predict membrane helices. It embeds a number of statistical
preferences and rules into a Hidden Markov Model to optimize the
prediction of the localization of membrane helices and their orientations.
Similar concepts are used for HMMTOP. TMPRED algorithm is based
on the statistical analysis of TMbase, a database of naturally occurring
trans-membrane proteins. The prediction is made using a combination
of several weight-matrices for scoring. TOPPRED II averages the GES-
scale of hydrophobicity using a trapezoid window.
Three Dimensional Modelling: Now-a-days in structural biology
field macromolecular structure development is necessary for the
interactions study. So the determination of three dimensional structures
of protein or nucleic acids using different experimental and theoretical
technique is mandatory for this purpose. Most widely used experimental
techniques for the three dimensional structure predictions are X-Ray
Crystallography, NMR and cryo-EM. Both NMR and X-Ray methods of
structure determination generate co-ordinate data that are usually deposited
in the Protein Data Bank (PDB). However, these techniques require huge
time, expertise and resource. With the progression of structural biology
research, molecular modelling is becoming an important method of choice
to obtain the 3-dimensional structures of biological macromolecules
such as Proteins, DNA etc relatively more quickly and easily. It helps to
1.16 Introduction to Bioinformatics

bridge the gap between the available sequence and structure information
by providing reliable and accurate models.
There are mainly three major methods for prediction of three
dimensional structure of protein.
1. Homology modelling or comparative modelling
2. Theading /fold recognition
3. Ab-initio structure prediction
The details of these methods are presented in Chapter 4.
Model Refinement/Optimizaion: The models of the biomolecules
obtained from the aforementioned techniques may have many errors.
Therefore, it is strongly recommended to check the stereo-chemical
qualities of the models. Such methods are called model optimizations.
Model optimization is applied for the whole protein not the isolated
loop regions. Energy minimization is done either by restraining the atom
positions and /or only for a few hundred steps. Molecular dynamics
simulation removes big errors but inefficient force-fields may introduce
atom clashes or small errors.
Validation of Models: Main sources of errors in Homology
Modelling are
1. Poor alignment: The target template sequence identity is very less.
When the target-template sequence identity is 90% and above then
model accuracy is comparable with crystal structures. At the 50-
90% identity range there are around 1.5Å rms deviations between
the structures of the target and the template with considerable
larger local errors but below 25% sequence identity the generated
model might have large errors.
2. Errors in template: If the template itself contains errors then the
generated model quality would be very low.
The structure of modeled protein must be validated through validation
server: Structure Analysis and Verification Server (SAVES) which
contains software programs for checking and validating protein structures
 Introduction 1.17

during and after model refinement. The following programs are generally
used for checking model qualities.
Procheck: The software tool checks the stereo-chemical qualities of
protein structures through Ramachandran Plot by analyzing overall and
residue by residue geometry. The plots include: Ramachandran plots,
both for the protein as a whole and for each type of amino acid; c1 – c2
plots (side chain torsion angles) for each amino acid type; main-chain
bond lengths and bond angles; secondary structure plot; deviations from
planarity of planar side chains and so on. Another important check is the
presence of bad contacts which is the distance between any pair of non-
bonded atoms being smaller than the sum of their van der Waals radii.
What_check: The tool provides an enormous number of checks
and detailed overall summary of the stereo-chemical parameters of the
residues in the model. Directional Atomic Contact Analysis implemented
in this program is used to analyze non-bonded contacts.
Verify3D: The tool examines the compatibility of an atomic model
(3D) with its own amino acid sequence (1D) by assigning a structural
class based on its location and environment (alpha, beta, loop, polar,
non-polar etc) and comparing the results to good structures.
ERRAT: The tool analyzes the statistics of non-bonded atom-atom
interactions with respet to a database of reliable high resolution structures.
Prove: The tool compares atomic volumes against a set of pre-
calculated standard values where atoms are treated like hard spheres
and calculates a statistical Z-score deviation for the model from highly
resolved and refined PDB-deposited structures.
Rampage: The tool provides stereo-chemical quality of protein
structure in Ramachandran Plot, where residues having proper y and
F main-chain torsion angles are clustered in allowed regions and the
percentage of residues in the disallowed regions are given.
In this chapter a brief overview of the different aspects of
Bioinformatics are provided. The details of these topics are presented in
the subsequent chapters.
1.18 Introduction to Bioinformatics

The analyses of biological data are heavily dependent on mathematics,

statistics and computer programming. Therefore, the basics of
mathematics, statistics and programming languages (Bio-Perl, Bio-Python
and Matlab) are discussed in the following sections.
Mathematics and Statistics in Bioinformatics: Mathematics is
routinely used in the analyses of Biological data. The rapid growth of
genomic data necessitates the use of Mathematics in Biology.
One of the main applications of Mathematics is the Chaos theory. The
Chaos theory is a special field of Mathematics which is used to analyze
the dynamical behaviors of Biological Systems. From time immemorial,
the biologists have been keeping track of populations of different
species by analyzing the population models. Most biological models
are continuous. However, recently scientists have started implementing
chaotic models in certain populations.  A study on models of Canadian
lynx showed that there was chaotic behavior in the population growth of
the species concerned. Chaos can also be found in ecological systems,
such as hydrology. Another important biological application of chaos
theory is found in cardiotocography.
Another important application of mathematics in solving Biological
problems is in the field of Evolutionary Biology. Evolutionary Biology
depends on extensive mathematical theorizing. The traditional approach in
this area which involves Mathematics and which includes complications
from genetics is population genetics. Most population geneticists propose
that the appearance of new alleles occurs by mutation; the appearance
of new genotypes occurs by genetic recombination; and changes in
the existing allele frequencies and genotypes occur at a small number
of gene loci. Considering infinitesimal effects at a large number of gene
loci together with the assumption of linkage equilibrium or quasi-linkage
equilibrium, the quantitative genetics method was devised.  Another
important avenue of population genetics is molecular phylogenetics. The
traditional population genetics model deal with information associated
with alleles and genotypes, and is frequently stochastic.
However, many population genetics models assume that population
sizes are constant. The variable sizes of populations, in the absence
 Introduction 1.19

of genetic variation information, are treated in the field of population

dynamics. The first principle of population dynamics is known as
the Malthusian growth model. Another important aspect of mathematical
Bioinformatics is the Lotka–Volterra predator-prey model. There are
various models of the spread of infections which have been proposed
and analyzed by Mathematical Biology. These theories provide
important results that may be applied to health policy decisions. Another
important field of Mathematical Bioinformatics is the evolutionary game
theory. This theory involves the selection of inherited phenotypes, without
genetic complications. The other avenues of Mathematical Biology are
as follows:
• Mechanics of biological tissues
• Theoretical enzymology and enzyme kinetics
• Cancer modelling and simulation
• Modelling the movement of interacting cell population
• Mathematical modelling of cell cycle
• Mathematical modelling of intra-cellular dynamics
Application of Mathematics in the analysis of cell cycle: One of
the important aspects of Bioinformatics is the analysis of eukaryotic cell
cycle. It is one of the most studied topics and a cell cycle mysregulation
leads to cancer. The analysis of cell cycle is performed by Mathematics
and involves simple calculus. There is a generic eukaryotic cell cycle
model representing a particular eukaryotic cell depending on the values
of the parameters like concentrations of the proteins in different cellular
organelles.
A set of ordinary differential equations are used to determine the
properties of the dynamical systems.
A few typical statistical concepts used in Bioinformatics are:
Hidden Markov Model (HMM): Hidden Markov Model (HMM) is
a statistical principle in which the biological system under consideration
is being modelled by assuming the system to have unobserved
1.20 Introduction to Bioinformatics

(i.e. hidden) states. In simpler Markov Models, the properties of the

biological systems are directly visible to the observer. This would help to
identify the transition parameters between the system variables. However,
in the Hidden Markov Model, the properties of the biological systems
are not directly visible; only the output becomes visible. In a simple
language, a Hidden Markov Model (HMM) is a statistical Model which
is used to describe the evolution of observable events. The observable
events depend on internal factors. The internal factors are not directly
observable. The observed events are called ‘symbol’ and the invisible
factors underlying the observation are called ‘state’. An HMM is made up
of two stochastic processes, viz., an invisible process of hidden states and
a visible process of observable symbols. HMMs have varied applications
in Bioinformatics. They are used in biological sequence analyses, gene
finding methods, and protein structure prediction methods to name a few.
Machine Learning: Machine learning is a subfield of computer
science. It involves the development of algorithms that learn how to
make predictions based on available data. The machine learning has a
number of emerging applications in the field of bioinformatics. Before
the advent of machine learning algorithms, bioinformatics algorithms
had to be explicitly programmed manually. However, this makes the
solution of the biological problems extremely difficult. Machine learning
techniques help the algorithm to consider the automatic feature learning
from the dataset. The machine learning algorithms can learn how to
combine multiple features of the input data into a more abstract set of
features which can further be manipulated to come up with a generalized
Model to solve the problem. Machine learning approaches are known to
be a very efficient technique to solve biological problems. However, the
most important pre-requisite of the process is that the dataset used for
the purpose should be large enough. Machine learning has been applied
to the following six main subfields of bioinformatics like: Genomics,
Proteomics, Microarrays, Systems Biology, Evolutionary Biology, and
Text Mining.
Application of Machine Learning in Genomics: Genomics is the
study of the genomes which is the complete DNA sequence of organisms.
 Introduction 1.21

After the advent of sophisticated genome sequencing techniques, the

number of available sequences is growing exponentially. Though the raw
data are constantly being deposited in the databases but it is a challenge
to annotate the raw data to extract relevant biological information from
the raw sequence data. Unfortunately, the biological annotations of the
raw sequence data are a daunting task and require a lot of involvements.
This necessitates the use of machine learning algorithms. The machine
learning methodologies are able to extract features to identify the specific
patterns in the biological sequences. This further would help to explain the
plausible functional annotations. One of such examples is the prediction of
the locations of the protein-coding genes within a given DNA sequence.
This is a problem known as computational gene prediction. The prediction
of genes is commonly performed through a combination of extrinsic and
intrinsic searches. In case of the extrinsic search, the input DNA sequence
is allowed to run through a large database of sequences. The database
contains a set of known genes for which sufficient information are already
available. The machine learning tools extract features from the known
gene sets and use those features to build a model. The new nucleotide
sequences are fed into the model generated from the available features as
mentioned before and are analysed. Machine learning algorithms are also
used for the problem of multiple sequence alignment. Multiple sequence
alignment involves aligning many DNA or amino acid sequences in
order to determine regions of similarity that could indicate a shared
evolutionary history. Multiple sequence alignment can also be used to
detect and visualize genome rearrangements.
Application of Machine Learning in Proteomics: Proteins are
polymers of amino acids. They are the main functional molecules in
the cellular systems. Proteins gain their functions from specific protein
folding mechanisms. Through the protein folding, the proteins conform
into a three-dimensional structure. This protein structure is composed of
a number of layers , viz., the primary structure (i.e. the sequence string
of amino acids), the secondary structure (alpha helices and beta sheets),
the tertiary structure (the three-dimensional structure of the protein), and
the quaternary structure (the relative orientations of the different chains
in the protein with respect to each other in the three dimension).
1.22 Introduction to Bioinformatics

Machine learning algorithms mainly aim to predict the secondary

structures of proteins from the amino acid sequences of the proteins.
After generating the secondary structures of the proteins, the tertiary
and quaternary structures of the proteins are subsequently built. This is
a very important avenue in structural bioinformatics as solving the true
structure of a protein in-vitro is an incredibly expensive and time-intensive
process. Such wet-lab experimental methods would require the Systems
that can accurately predict the structure of a protein by analyzing the
amino acid sequence directly. As in the case of genome analyses, prior
to the advent machine learning techniques, the structure prediction work
was performed manually. Today, through the use of automatic feature
learning, the best machine learning techniques are known to be able to
achieve a prediction accuracy of 82-84%. The current state-of-the-art
machine learning technique for prediction of protein structure is referred
to as DeepCNF (deep convolutional neural fields). This technique relies
on the machine learning model of artificial neural networks to achieve an
accuracy of approximately 84% when the method is applied to classify
the amino acid sequences of a protein into one of three structural classes
(helix, sheet, or coil). Machine learning has also been applied to other
proteomics problems such as protein side-chain prediction, protein loop
modelling, and protein contact map prediction.
Application of Machine Learning in Systems Biology: Systems
Biology is the study of the emergent behaviours obtained from complex
interactions of simple biological components in a system. Such
components are the biological macromolecules like, DNA, RNA, proteins,
and metabolites. Machine learning techniques have been used to aid in
the modelling of these complex interactions that occur in the biological
Systems in various different domains such as genetic networks, signal
transduction networks, and metabolic pathways. A specialized machine
learning technique, known as the Probabilistic Graphical Models is used
for determining the structure-function relationship between different
variables. This approach is one of the most commonly used methods for
modelling genetic networks. In addition to that, the machine learning
techniques have been applied to other Systems Biology problems such
as identification of transcription factor binding sites with the help of a
 Introduction 1.23

technique known as Markov chain optimization. Genetic algorithm is

another machine learning technique which is based on the natural process
of evolution. Genetic algorithm has constantly been used to model genetic
networks and regulatory structures.
Machine learning tools are also used for various other Systems Biology
applications which include the task of enzyme function prediction, high
throughput microarray data analysis, and analysis of genome-wide
association studies for a better understanding of the potential markers
of Multiple Sclerosis, protein function prediction, and identification of
NCR-sensitivity of genes in yeast.
Application of Machine Learning in text mining: Now-a-days,
there is a tremendous increase in the appearance of biological literature
in the form of biological publications. Therefore, it is very much a
necessity to extract relevant and important information from a huge
collection of available biological publications. This task is referred to as
knowledge extraction. The extraction of the relevant information from
available biological data is a daunting task. Such extraction process leads
to collection of biological data to be fed into a machine learning tool to
generate new biological knowledge. Machine learning techniques can be
used for this knowledge extraction task with the help of techniques such
as natural language processing to extract the useful information from
human-generated reports in a biological literature database. Another
variation of the text mining technique is called the text nailing which
has been applied to the search for novel drug targets. This technique
requires the examination of information stored in biological databases and
journals. Furthermore, it has been observed that annotations of proteins
in protein databases often require extraction of additional information
from biological literature. Machine learning techniques have been applied
to automatic annotation of the function of genes and proteins for the
determination of the subcellular localization of a protein, analysis of
DNA-expression arrays, large-scale protein interaction analysis, and
molecule interaction analysis.
Final Words: Bioinformatics is a multidisciplinary subject. It requires
knowledge from Physics, Chemistry, Biology, Mathematics, Statistics and
1.24 Introduction to Bioinformatics

Computer Science. Day-by-day it is becoming an important field of study.

There are various biological problems. The basic aim of the bioinformatics
study is to facilitate the experimental biologists. Bioinformatics
experiments are comparatively cheaper and reproducible. At the same
time, result from bioinformatics analyses, would provide detailed insights
into the problem in hand. The different aspects of bioinformatics will be
discussed in the subsequent chapters in a greater detail.
 CHAPTER

2 Biological Databases

2.1 INTRODUCTION

The basic aim of this chapter is to deal with vaious aspetcs of Biological
Databases emphasizing on the distinction between datatypes, data
organizations and usablity of the data. The chapter focuses on sequence
and structural databases. Necessary definitions are also provided. Web
addresses of the biological databases are also added.

2.2 SOME IMPORTANT TERMINOLOGIES AND DEFINITIONS

(a) Amino acids: Amino acids are biologically important molecules

bearing amino (-NH2) and carboxyl terminal ends (-COOH). The
amino acids differ in their side chains. The amino acids are broadly
classified as polar charged (acidic, basic), polar uncharged and
hydrophobic (aromatic, aliphatic) on the basis of the side chains.
Amino acids make peptide linkages among themselves to build
the protein chain.
(b) Representations of amino acids: Amino acids are represented
by single and three letter codes.
Name of the Single Three Side chain structure Type
amino acid letter letter
code code
Glycine G Gly H- Polar neutral
Alanine A Ala CH3- Hydrophobic aliphatic
Leucine L Leu (CH3)2-CH-CH2- Hydrophobic aliphatic
2.2 Introduction to Bioinformatics

Methionine M Met CH3-S-CH2-CH2- Hydrophobic aliphatic

Phenylalanine F Phe C6H5-CH2- Hydrophobic aromatic
Tryptophan W Trp Hydrophobic aliphatic

Lysine K Lys NH2-(CH2)3-CH2- Polar basic

Serine S Ser OH-CH2- Polar neutral
Asparagine N Asn H2N-CO-CH2- Polar neutral
Aspartic Acid D Asp HOOC-CH2- Polar acidic
Asparagine/ B Asx Polar neutral
Aspartate

Proline P Pro Hydrophobic aliphatic

Valine V Val (CH3)2-CH- Hydrophobic aliphatic

Isoleucine I Ile CH3-CH2-CH-CH3 Hydrophobic aliphatic
Cysteine C Cys SH-CH2- Polar neutral
Tyrosine Y Tyr Hydrophobic aromatic

Histidine H His Hydrophobic aromatic

Arginine R Arg Polar Neutral

Threonine T Thr Polar Neutral

Glutamine Q Gln H2N-CO-CH2-CH2- Polar Neutral

Glutamic Acid E Glu HOOC-CH2- CH2- Polar acidic
Glutamine/ Z Glx Polar Neutral
Glutamate
 Biological Databases 2.3

(c) Databases: Databases are collections of biological data. They

store a vast amount data. There are many different types of
databases based on the types of information being stored.

2.3 THE PRIMARY SEQUENCE DATABASES

With advent of sequencing techniques, more and more laboratories started

working on systematic organizations of available data. The endeavors
lead to the inventions of primary sequence databases. The biological
databases are built using nucleotide sequences, amino acid sequences.
The most important primary databases are:
For Nucleotides
(a) EMBL
(b) DDBJ
(c) NCBI-GenBank
For Proteins
(a) PIR
(b) SWISS-PROT
(c) UniProtKB/TrEMBL

2.4 THE NUCLEOTIDE SEQUENCE DATABASES

The most important DNA sequence databases are EMBL, GenBank

and DDBJ. These databases exchange data on a daily basis to ensure a
wholesome coverage of the sequence information among all of them.

2.4.1 EMBL-EBI (http://www.ebi.ac.uk/)

EMBL-EBI supplies the nucleotide sequence data from life science
experiments. These data are freely available. The database contains
sequences submitted directly by scientists individually from genome
sequencing groups as well as from the scientific literature and patents.
The database collaborates with DDBJ and GenBank and entries are
exchanged between these databases on a regular basis.
2.4 Introduction to Bioinformatics

There is exponential rate of growth of DNA sequence data. Till date

the EMBL database contains more than a billion of nucleotide sequence
entries.
EMBL can be searched by SRS (Sequence Retrieval System). This
is used to link the important DNA and protein sequence databases to
analyze the sequence motifs, structure etc. The databases are linked to
MEDLINE to search for the available literature references.
Data submission to EMBL

Many journals have made it mandatory for authors to submit the newly
found nucleotide sequence information to the EMBL, GenBank or DDBJ
database prior to publication in order to ensure its availability to scientists.
The data can be submitted via the webtool http://www.ebi.ac.uk/embl/
Submission/webin.html.
Database entries produced at the sequencing site can be deposited and
updated directly by the submitters using FTP or email. Groups producing
large volumes of genome sequence data over an extended period of time
are advised to contact the database at ku.ca.ibe@sbusatad.
Sequence manipulations

EMBL-Align is a public data set of both protein and nucleotide multiple

sequence alignments. This server can be queried from the EBI Sequence
Retrieval Server (SRS) server. It was developed in response to the need for
permanent electronic storage and standardized presentation of alignment
data from phylogenetic and population analyses. This is a dedicated
web-based tool for submission of multiple sequence alignments in all
common alignment formats which are available at http://www.ebi.ac.uk/
embl/Submission/align_top.html.
Sequence retrieval

The EBI Genomes server at http://www.ebi.ac.uk/genomes/can be used to

access the several thousand completely sequenced genomic components.
Proteome analysis information on all completely sequenced organisms
is available at http://www.ebi.ac.uk/proteome/.
 Biological Databases 2.5

Whole Genome Shotgun (WGS) data

WGS data are available at ftp://ftp.ebi.ac.uk/pub/databases/embl/wgs/.

The largest data set is that of Rattus norvegicus. While many of the WGS
data sets are not annotated, some biological features are present in some
of the sets. The WGS data set for Anopheles gambiae strain PEST is an
example with annotation. WGS data sets can now be searched using the
FASTA program.
Sequence Retrieval System (SRS)

The EMBL Nucleotide Sequence Database can be accessed via the EBI
SRS server at http://srs.ebi.ac.uk/. In SRS, the data are available in the
following libraries:
(i) EMBL: the database in its entirety by means of a virtual library
comprising EMBLRELEASE, EMBLNEW, EMBLTPA and
EMBLWGS;
(ii) EMBLRELEASE: library containing the latest official release of
the EMBL Nucleotide Sequence Database;
(iii) EMBLNEW: library containing updated and new entries created
since the last official release;
(iv) EMBLTPA: library containing TPA entries;
(v) EMBLWGS: library containing WGS entries;
(vi) EMBLCON: library containing CON entries.
Sequence searching

There is sequence analysis toolbox in EMBL available at http://www.

ebi.ac.uk/Tools/. Sequence similarity manipulations can be performed
interactively over the WWW as well as by email. Users can search
the EMBL Nucleotide Sequence Database as a whole or by individual
taxonomic division.
The most commonly used algorithms available are FASTA and WU-
BLAST, permitting comparisons between nucleotide query sequences and
the nucleotide or protein databases as well as searches of protein query
sequences against the nucleotide databases.
2.6 Introduction to Bioinformatics

The FASTA service for genomes and proteomes (http://www.ebi.

ac.uk/fasta33/genomes.html) enables users to search interactively
completed genomes and proteomes. The same searches can be performed
by email (ku.ca.ibe@atsafpg). User instructions are available by sending
an email with the word HELP in the body of the message to ku.ca.ibe@
atsafpg. WGS data sets are now available for searching.
Sequence analysis

One of the most important sequence manipulation techniques is the

multiple sequence alignment. This is performed by ClustalW. The results
from ClustalW can be used in phylogenetic analyses of the species
involved. The presence of specific domains and motifs in the amino
acid sequences can be identified using InterProScan and others. The
EBI also provides interactive sequence analysis resources based on the
European Molecular Biology Open Software Suite (EMBOSS) (http://
www.emboss.org/).

2.4.2 DDBJ (http://www.ddbj.nig.ac.jp/)

DDBJ stands for the DNA Data Bank of Japan. DDBJ first started in
1986 as a collaborative effort with EMBL and GenBank. The database
is maintained by the National Institute of Genetics, Japan. DDBJ accepts
nucleotide sequences from the entire world using the world wide web.
DDBJ has several tools for annotations of biological sequence data.
For database search the following tools are available:

getentry: To retrieve sequence data by accession numbers, etc.

ARSA: For all-round Retrieval of Sequence and Annotation
TXSearch: For retrieval of data from unified taxonomy database.
BLAST: For sequnce similarity search
VecScreen: It is a Vector Search to screen contamination in nucleic
acid sequences.
DRA Search: This tool is to search metadata by keywords and
retrieve data
 Biological Databases 2.7

For Phylogenetic s:

ClustalW: To perform multiple sequence alignment and to draw

phylogenetic trees
WABI (Web API for Biology): WABI is the Web API to use DDBJ‘s
data retrieval system without mediating web browsers. Following services
provide the Web APIs as well as web browser forms:
WABI BLAST(Instruction in Japanese only)
getentry WebAPI
ARSA WebAPI
For Next Generation Sequencing:

DDBJ Read Annotation Pipeline: This pipeline is used to analyze

the High-throughput data obtained from next generation sequencing
techniques.
But this tool requires registration.
For Genome Analysis:

MiGAP: This is a mechanical annotation tool for microbial genomes.

But this tool requires registration.
GTPS: This tool is for reannotation of bacterial genomes using a
new common protocol.
GTOP: This tool is to analyze genome to protein structure and
identification of protein function
AOE: Statistics and trends of gene expression data
Gendoo: Functional profiling of gene and disease features for omics
analysis
GGGenome: A ultrafast sequence search
GGRNA: A Google-like, ultrafast search engine for genes and
transcripts
RefEx: Reference Expression dataset for tissue transcriptome
2.8 Introduction to Bioinformatics

2.4.3 NCBI-GenBank (https://www.ncbi.nlm.nih.gov/genbank/)

GenBank is the repository of nucleic acid as well as amino acid sequence

information maintained by National Institute of Health (NIH), USA.
This database contains annotated sequence information of the available
sequences. In GenBank the sequence data are updated on a regular
basis. A new sequence in GenBank is released approximately in every
two months. All the data of the GenBank are available from the ftp
site. An important feature of a GenBank entry is that the entry contains
release notes. The release notes give the detailed aspect of the current
version of GenBank entry. However, the previous release notes for the
same GenBank entry also remain available. The GenBank entries can be
searched and disseminated freely over the World Wide Web.
GenBank accepts DNA and m-RNA sequences. It also accepts
un-annotated sequence information obtained from Next Generation
Sequencing (NGS) methods but such sequences are stored in Sequence
Read Archive. In no circumstances GenBank accepts the following data
types: Non-contiguous nucleotide or amino acid sequences, nucleic acid
sequences of primer regions, amino acid sequences of proteins having
missing nucleotide sequence submission, a mixture of mRNA and DNA
sequences.
Expressed sequence tags and sequences obtained from genomic
surveys are accepted in GenBank. GenBank also stores information about
the whole genome sequences of organisms.
Each sequence entry in GenBank is given an accession number which
serves as a unique identifier of the sequence.
There are various tools available in GenBank to deposit the sequence
from a user. The tools are: BankIt, Sequin, tbl2asn, Submission Portal,
Barcode Submission Tool.
A GenBank entry is linked to many other databases for cross
references. The most important aspect of a GenBank entry is that it is
fully annotated and all the annotations are self-explanatory.
 Biological Databases 2.9

2.5 FINAL WORDS

There are different nucleotide sequence databases available. The databases

are comprehensive and provide a large chunk of sequence information.
However, the basic problem of all these databases is the use of different
file formats. Thus, to maintain a uniformity between the different data
file formats, a new initiative has been adopted and it is called The
International Nucleotide Sequence Database Collaboration (INSDC).
It is available at http://www.insdc.org. INSDC is created on the basis
of its partnership with various other databases. The basic spirit of the
database is to come up with the proper establishment of specific formats
and protocols of collecting nucleotide sequence information. INSDC also
aims to incorporate the bibliographic information of the sequence and
most importantly a uniform accession number for all the entries in the
database which would make it easier to retrieve the sequence. INSDC
has collaborations with EMBL, DDBJ and NCBI-GenBank.

2.6 THE PROTEIN SEQUENCE DATABASES

Like nucleotide sequence databases, there are many different databases

that store information about the amino acid sequences of proteins.
The most important sequence databases are PIR, SWISS-PROT and
UniProtKB/TrEMBL.

2.6.1 PIR (http://pir.georgetown.edu/)

It is an amino acid sequence database. It is maintained by Georgetown
University Medical Center. This database is an integrated repository of
protein information. It supports researches in genomics and proteomics.
PIR stores the protein information in a database called the Protein
Sequence Database (PIR-PSD). PIR-PSD is a sub-database in the domain
of PIR and it has amino acid sequences of proteins and the amino acid
sequnces are properly annotated. The database generally covers the
amino acid sequences of nealy all the taxonomic groups. The amino acid
sequences are grouped into superfamilies, families and domains. These
annotations provide an in-depth knowledge of the protein product from
a gene. Not only the sequence level information about the proteins, the
2.10 Introduction to Bioinformatics

PIR database is also linked with litterature database to search for suitable
bibliographic references for the proteins and their annotations. As sub
databases PIR stores some other important information in terms of:
NREF: It is a non-redundant reference database. This database
maintains constant cross-talking with other existing databases like Swiss-
Prot, PDB etc. The NREF database can be used in searching for sequences
using various sequence analyses tools like BLAST.
iProClass: This is an integrated database of protein family, function,
structure information, pathways, interactions, genes and gene ontology
information and taxonomic results.
iPTMnet: This is a resource that provides information about the
various types of post-translational modifications in proteins. This resource
covers the Systems Biology aspects of proteins.
iProLINK (integrated Protein Literature, INformation and
Knowledge): This is a resource that links the litterature references of
proteins with the sequence information available in PIR. The resource is
aimed at providing a wholesum knowledge of the proteins.
PIRSF: This is a tool that is meant to gather information from other
resources, viz., UniProtKB in a comprehensive manner. This tool is also
used to derive evolutionary relationships from the amino acid sequences
present in the UniProtKB. One of the interesting features of the tool is to
analyze the protein relationships using the entire protein. This is unique
in the sense that other such resources perform the task using only the
specific domain information in the protein. The tool helps in proper
annotations of biological functions of proteins for which specific domain
information is not available.
PRO: This tool is called the protein ontology tool. This tool is
specific for analyzing the gene ontological details of the available
amino acid sequences of the proteins. The ontological information
provide information about the specific gene ontology terms like protein
complex clusters, protein othologous isoforms etc. The tool also provides
information about the protein product in taxon specific and taxon neutral
manner. There are sub ontologies of PRO. They are
 Biological Databases 2.11

ProEvo: This sub-tool of PRO provides the evolutionary information

of proteins on the basis of available evolutionary relationships.
ProForm: This sub-tool of PRO provides protein related information
from available gene locus.
ProComp: This sub-tool of PRO analyzes the protein complexes
and provides the information regarding the various aspects of the protein
complexes.
There are other resources available in PRO. They are as follows:
Text Search: This is to identify the details of a specific annotation.
This can be done for a single annotation or for a number of annotations
commonly referred to as the Batch search.
Sequence analysis: This is done by the tool Peptide Match. The tool
is used to match a specific peptide sequence with the existing sequences
present in the databases. The commonly used databases for searching
purposes are UniProtKB and UniRef100.
There are other tools available for sequence analysis. They are tools
for multiple sequence analysis and pairwise sequence analysis. The
multiple sequence analysis is commonly performed by ClustalW. For
pairwise alignment Search is used.
The theoretical molecular weight of a protein can be computed from
the amino acid sequence of a protein. Using this concept a tool was
developed and the tool is used in PRO.
The information regarding the same protein are stored in different
databases and these proteins bear different identifiers. The mapping of the
different protein identifiers can also be done with the tools available in PRO.

2.6.2 SWISS-PROT (http://web.expasy.org/docs/swiss-prot_guideline.html)

One of the vital protein sequence databases is SWISS-PROT. SWISS-
PROT is a database which contains the protein sequence information
alongwith all the necessary annotations. The database aims to provide
high-end sequence information about proteins, which include the
functional characterizations, domain information, post-translational
2.12 Introduction to Bioinformatics

modifications, information about protein variants and so on. SWISS-

PROT tries to minimize the data redundancy and is linked to many other
databases.
The organization of the database and the style of annotations in the
SWISS-PROT make it one of the most popular protein sequence databases.
The most important part of a SWISS-PROT entree is the CC and FT
lines. The CC lines provide the detailed comments regarding the protein.
It gives us the ideas about the protein function, its sub-cellular location,
the post-translational modifications associated with the protein, its tissue
specificity and so on. FT lines represent the structural details of the protein,
like the local secondary structure, the ligand binding sites, presence of
transmembrane domains. Most importantly, the SWISS-PROT provides
the three-dimensional structural details of proteins if available.

2.6.3 TrEMBL (http://web.expasy.org/docs/swiss-prot_guideline.html)

SWISS-PROT is a protein sequence database which provides a high
level of annotations. The database stores data pertaining to the functional
description of a protein, the presence of domains in the protein, post-
translational modification sites in the proteins and the different isoforms
of the protein. The database contains a minimal level of redundancy and
is linked to other databases. TrEMBL, on the other hand, is a computer-
annotated supplement of SWISS-PROT. This database contains all the
translations of EMBL nucleotide sequences which are not yet integrated
in SWISS-PROT database. TrEMBL data are divided into two main
sections: SP-TrEMBL and REM-TrEMBL. SP-TrEMBL (Swiss-Prot
TrEMBL) contains the data that will be eventually incorporated into
Swiss-Prot with Swiss-Prot accession numbers.
SP-TrEMBL data are arranged in subsections as follows:
arc.dat: This is for Archaea.
fun.dat: This is for Fungi
hum.dat: This is for Human
inv.dat: This is for Invertebrates
mam.dat: This is for Other Mammals
 Biological Databases 2.13

mhc.dat: This is for MHC proteins

org.dat: This is for Organelles
phg.dat: This is for Bacteriophages
pln.dat: This is for Plants
pro.dat: This is for Prokaryotes
rod.dat: This is for Rodents
unc.dat: This is for unclassified results
vrl.dat: This is for Viruses
vrt.dat: This is for Other Vertebrates
REM-TrEMBL is also referred to as REMaining TrEMBL database
which stores the data that are not ultimately included in Swiss-Prot.
Consequently REM-TrEMBL entries are not given any accession
numbers.

2.6.4 The Reference Sequence Database (RefSeq)

(https://www.ncbi.nlm.nih.gov/RefSeq)
The Reference Sequence (RefSeq) database comprises of data from
publicly available nucleotide sequences (DNA, RNA) and their protein
products. These data are already curated and annotated. This database is
hosted by National Center for Biotechnology Information (NCBI). The
unique feature of the database is that it stores only a single record for
each biomolecule (DNA, RNA or protein). The data are obtained from
a variety of organisms involving viruses to bacteria to eukaryotes. The
database contains non-redundant sequence data
The database provides separate and linked records for the genomic
DNA, the gene transcripts, and the protein products from the genes.
However, RefSeq captures data from a limited number of organisms for
which sufficient data are available.
Accession Description of the term
type
AC_ The entry having this accession code signifies complete genomic
molecule, usually alternate assembly
2.14 Introduction to Bioinformatics

NC_ The entry having this accession code signifies complete genomic
molecule, usually reference genome assembly. The reference genome
is selected as a representative from the set of best genomic sequences
available
NG_ The entry having this accession code signifies incomplete genomic
molecule
NT_ The entry having this accession code signifies genomic contigs obtained
from cloning experimentations. A sequence contig is a continuous stretch
of sequences bounded by gap or a run of 10 or more same nucleotides
NM_ The entry having this accession code signifies data pertaining to m-RNA
molecules
NR_ The entry having this accession code signifies data pertaining to non-
coding RNA molecules
XM_ The entry having this accession code signifies data pertaining to
predicted protein coding m-RNA molecules
XR_ The entry having this accession code signifies data pertaining to
predicted non-coding RNA molecules
AP_ The entry having this accession code signifies data pertaining to proteins
NP_ The entry having this accession code signifies data pertaining to proteins
obtained from genomes represented by AC_ or NC_

2.6.4 Interpro Database (http://www.ebi.ac.uk/interpro/)

InterPro database consists of data required for functional analysis of
protein sequences. These are done by classifying the protein sequences
into different protein families on the basis of the presence of functional
domains and other important sites. For classification purposes, the proteins
are classified using predictive models, called signature sequences. The
signature sequences were obtained from different other databases. The
database has a facility that allows a user to search for the signature
sequences in proteins using a tool called InterProScan. The most important
aspect of the database is that it collects information from various other
resources and combines them in a single comprehensive form. The
Interpro entry is classified into mainly four different categories. They are
Homologous Superfamily: This represents a group of proteins that
orginate from a common ancestor as reflected by similarity in their three-
dimensional structures. However, the members of the superfamily often
have very low similarity at the sequence level.
 Biological Databases 2.15

Family: A protein family in Interpro database consists of a group

of proteins that orginate from a common ancestor as reflected by their
functional similarities. The proteins belonging to the same family have
similarities in their amino acid sequences or in their tertiary structure.
Domain: A protein domain in Interpro database is a distinct
functional, structural or sequence unit. Such a unit may exist in a variety
of biological contexts.
Repeat: A repeat in an Interpro entry is a short stretch of sequence.
Such a sequence is found to be repeated within a protein.
Site: A site in an Interpro entry is another short stretch of sequence. It
contains one or more conserved residues and would reflect the presence
of active sites, binding sites, post-translational modification sites and
conserved sites.
The usefulness of Interpro database is manifold:
(a) It is comprehensive and non-redundant.
(b) It gathers information from various other resources and makes a
single searchable form.
(c) The database helps users to annotate their queries related to proteins.
The Interpro database can be used to analyse the presence of specific
domains or motifs in a protein. However, Interpro database can not
be used for genome annotations. The database also is not of much
help to perform structural alignments of proteins.

2.7 THE PROTEIN STRUCTURE DATABASE

One of the most important biological databases is the protein sequence-

structure database, known as the Protein Data Bank (PDB: https://www.
rcsb.org/pdb/home/home.do). PDB is a repository that stores the three-
dimensional coordinates of the atoms of the amino acid residues in
proteins, and protein-ligand complexes. The three-dimensional structures
of the proteins are determined by X-ray crystallography, NMR, Electron
Microscopy and a combinations of all these of structure determination
tools. The database stores information from all forms lives like bacteria,
2.16 Introduction to Bioinformatics

yeast, plants, flies, other animals and humans. The database not only
stores static information but contains several tools to process and analyze
the data. In general, a PDB entry contains the sequence and structural
information together. It also provides the experimental details of the
structure determination. The amino acid sequences in a PDB entry are
linked to their corresponding structural details. A PDB entry contains the
necessary annotations which provide information for the structure and
function of the protein. A PDB entry is linked to various other structural
databases like CATH and SCOP as well the litterature database Pubmed.
The tools available in PDB would help to find the proteins having similar
sequences and structures to an existing protein. The pdb data can be
freely downloaded for analysis purposes.

2.7. A. CATH DATABASE

The CATH or Class-Architecture-Topology-Homology-S.O.L.I.D

database is a freely available online repository for providing information
on the evolutionary relationships of protein domains. The protein domains
obtained from the protein structures deposited in the PDB are classified
within the CATH structural model as:
CLASS: At the Class (C) level, the protein domains are determined
according to the contents of their secondary structural elements. The
Class level in CATH has the following categories: all alpha, all beta, a
mixture of alpha and beta, or little secondary structure.
Architecture: At the Architecture (A) level, the protein domains are
classified on the structural arrangements of their secondary structural
elements in three-dimensional space.
Topology/Fold: At the Topology/Fold (T) level, the protein domains
are classified on the mode of connections between the different secondary
structural elements.
Homology: The protein domains are classified to be belonging to the
same three-dimensional space if they are related by evolution.
Sequence Family (S35): The protein domains are classified as
belonging to the same sequence family if they have >= 35% sequence
similarities among themselves.
 Biological Databases 2.17

Orthologous Family (O60): The protein domains are classified as

belonging to the same orthologous family if they have >= 60% sequence
similarities among themselves.
Like (S95): The protein domains are classified as belonging to having
the same domain structure if they have >= 95% sequence similarities
among themselves.
Identical (S100): The protein domains are classified as belonging to
having exactly the same domain structure if they have >= 100% sequence
similarities among themselves.
Domain Counter (D): The protein domains which are unique in
nature are classified as belonging to DOMAIN COUNTER.
If there are some sequence data available with no experimental
validations of their structural classes, the tool Gene3D , which is a part
of CATH database, is used to annotate the sequence data.

2.7. B. SCOP DATABASE

The SCOP database is for the structural classification of proteins. The

SCOP database has the following components: Protein types, Evolutionary
events, Structural classes and Protein relationships.
Protein Types: In this category the proteins are classified into the
following classes: soluble, membrane, fibrous and intrinsically disordered.
The protein type class is based on the characteristic sequence and
structural features of proteins.
Evolutionary Events: In this category, the various types of structural
aberrations in proteins belonging to the same class are considered.
Structural Class: The proteins are classified as per the presence of
different types of secondary structural elements in them. The structural
classes are all alpha, all beta, alpha beta and alpha/beta.
Family: Proteins are considered to be belonging to the same structural
families if they have the same structural domains.
2.18 Introduction to Bioinformatics

Superfamily: Proteins having the same structural domains but

belonging to different structural classes are considered to be in the same
superfamily.
Hyperfamily: Proteins belonging to different superfamilies but
having the same structural domains are considered to be present in the
same Hyperfamily.
The SCOP is a very useful database to analyze the structural features
of a protein.

2.7. C. FSSP DATABASE

The FSSP database is a collection of three-dimensional folds of proteins

obtained from the Protein Data Bank. The database stores the data based
on all-with-all structural comparisons of the proteins present in the Protein
Data Bank. The database contains different protein chains belonging
to different fold families. The mutual sequence identities between the
amino acid sequences of the proteins belonging to each fold fanily are
<25% among themselves. In the database, the data are stored as per the
three-dimensional coordinates of the atoms of the amino acid residues.

2.8 COMBINATION DATABASES

There are some other databases which provide information about all
types of biological systems. Some of those databases are Gene Ontology
Database (GO Database), DAVID (the Database for Annotation,
Visualization and Integrated Discovery), Pubmed,

2.8.1 G
 ene Ontology Database (GO Database) (http://www.
geneontology.org/page/go-database)
It is a major computational initiative to collect and present the different
features of genes and gene products from all species. The main aims
behind the development of the database are manifold.
1. To maintain and develop a controlled vocabulary of various
features of genes and gene products
 Biological Databases 2.19

2. To extract structural and functional features from the genes and

gene products and to present the data in unified manner
3. To provide different tools for easy access of all aspects of the
genetic data
The term Ontology is defined as the representation of things that can
be detected by means of experimentations. The GO project comes up
with such representations of the biological terms associated with gene
product. The GO covers mainly the three domains:
• Cellular component: It is the regions inside or outside of a cell
where the gene product exhibits its function
• Molecular function: It is the biological activity of the gene
product
• Biological process: It is the biological pathway where the gene
product exhibits its functionality.

2.8.2. D
 AVID (the Database for Annotation, Visualization and
Integrated Discovery) (https://david.ncifcrf.gov/home.jsp)
DAVID (the Database for Annotation, Visualization and Integrated
Discovery) is a repository developed by the Laboratory of
Immunopathogenesis and Bioinformatics (LIB). The tools present in
the DAVID Bioinformatics Resources help the user to provide functional
annotations of large lists of genes derived from genomic studies,
e.g. microarray and proteomics studies. The DAVID Bioinformatics
Repository comprises of the DAVID Knowledgebase and five other
integrated, web-based functional annotation tools:
• DAVID Gene Functional Classification Tool
• DAVID Functional Annotation Tool
• DAVID Gene ID Conversion Tool
• DAVID Gene Name Viewer
• DAVID NIAID Pathogen Genome Browser.
2.20 Introduction to Bioinformatics

The DAVID repository can be used in the following purposes:

• To identify enriched biological annotations of proteins
• To discover functional aspects of genes and gene products
• To analyze the cluster redundant annotation terms
• To visualize genes involved in metabolic pathways
• To search for functionally related genes
• To analyze a list of interacting proteins
• To explore gene names by batch processing
• To link a gene with diseases
• To identify protein functional domains and motifs
• To study literature references
• To convert gene identifiers from one type to another.

2.8.3. Pubmed (www.ncbi.nlm.nih.gov/pubmed/)

PubMed is a free search engine to access the MEDLINE database of
references and abstracts of literatures on life sciences and biomedical
topics. The United States National Library of Medicine (NLM) at the
National Institutes of Health holds the database as part of the Entrez
system for the retrieval of literature data.
MEDLINE database through Pubmed provides access to:
• Very old literature references from the print version of Index
Medicus, back to 1951 and even earlier
• Literature references to some specific journals before they were
indexed in Index Medicus and MEDLINE, for instance Science,
BMJ, and Annals of Surgery
• Very recent literature entries to records for an article before it is
indexed with Medical Subject Headings (MeSH) and added to
MEDLINE
• In addition to journal references MEDLINE provides access to a
collection of books for which full-texts may be available.
 Biological Databases 2.21

A sister database of Pubmed is PubMed Central (PMC). It is a free

digital storehouse that contains publicly accessible full-text scholarly
articles published within the biomedical and life sciences journal
literature. PubMed Central is much more than just a database of
documents. Submissions of any literature data into PMC would undergo
proper indexing and formatting resulting in a proper user-friendly
appearance of the literature.

FINAL WORDS
The biological sequences are not limited to nucleic acids and proteins
but there are carbohydrates as well. None-the-less, the advents of new
sequences are inreasing day-by-day. The databases are used to store the
large numbers of sequence information. The databases are constantly
being upgraded and there are cross-talks between the different databases.
However, not all the databases have the proper annotations of biological
data. The authenticity of the data may therefore be checked before
experimentations.
A few other important biological databases are mentioned in the
following sections:
Carbohydrate Structure Databases

1. EuroCarbDB: A database for both sequences and structures of

carbohydrates.
2. Carbohydrate Structure Database (CSDB): This database
stores sequence and structural information about naturally
occurring carbohydrates from bacteria, plants and fungi.
Signal Transduction Pathway Databases

1. Cancer Cell Map: This is a database containing a selected set

of browsable and searchable human cancer pathways.
2. Netpath: This database is a curated resource of signal transduction
pathways occurring in humans
3. NCI-Nature Pathway Interaction Database: This pathway
provides information regarding biomolecular interactions
2.22 Introduction to Bioinformatics

4. Reactome: This database provides the information regarding the

biological pathways that include metabolic processes to hormonal
signalling.
5. SignaLink Database: This database is an integrated repository
to analyze signaling pathways, cross-talks, transcription factors,
miRNAs and regulatory enzymes.
Protein-Protein and Other Molecular Interactions
1. BIND: This is a Biomolecular Interaction Network Database:
2. BioGRID: It is a database for Interaction Datasets
3. DIP: This is a Database of Interacting Proteins
4. IntAct molecular interaction database: This is a database for
analysis of protein–protein, protein–small molecule and protein–
nucleic acid interactions.
5. STRING: This database is for analyses of protein interaction
networks.
6. MINT: This database is for analysis of bio-molecular interaction
data.
7. Pfam: A protein familiy database
8. PROSITE: A database which stores protein motifs
9. CDD: A database which stores the information about conserved
domains in proteins
10. PRINTS: A database that contains information about the
fingerprint regions in protein sequences. The fingerprint regions
are particular sequence motifs in the proteins which confer them
with specific characteristics.
 Biological Databases 2.23

SOME IMPORTANT QUESTIONS AND ANSWERS

1. Margaret Dayhoff built the first protein sequence database called
(a) SWISS PROT
(b) PDB
(c) Atlas of protein sequence and structure
(d) Protein sequence databank
2. Each record in a database is known as
(a) entry (b) file
(c) record (d) ticket
3. Literature databases are
(a) MEDLINE and PubMED (b) MEDLINE and PDB
(c) PubMED and PDB (d) MEDLINE and PDS
4. FlyBase is a specialized database containing information about
(a) biodiversity (b) model organism
(c) literature (d) biomolecules
5. Which of the following is known as E. coli model organism database
(a) EcoGene (b) EcoBase
(c) EcoSeq (d) ColGene
6. Which of the following is a database of protein sequence
(a) DDBJ (b) EMBL
(c) GenBank (d) PIR
7. GenBnak, the nucleic acid sequence database, is housed and maintained
by
(a) Brookhaven laboratory
(b) DNA Database of Japan (DDBJ)
(c) European Molecular Biology Laboratory (EMBL)
(d) National Centre for Biotechnology Information (NCBI)
2.24 Introduction to Bioinformatics

8. Submission to GenBank is done via

(a) BankIt and Sequin

(b) BankIt and BankIn

(c) Sequin and BankIn

(d) Entrez

9. STAG is housed and maintained by

(a) Brookhaven laboratory

(b) DNA Database of Japan (DDBJ)

(c) European Molecular Biology Laboratory (EMBL)

(d) National Centre for Biotechnology Information (NCBI)

10. A database of human genetics and molecular biology is

(a) PDB (b) STAG

(c) OMIM (d) PSD

11. Find the odd one out

(a) PIR (b) PSD

(c) SWISS PROT (d) EMBL

12. The information retrieval tool of NCBI GenBank is known as

(a) Entrez (b) STAG

(c) SeqIn (d) text search

13. The first invented secondary database was

(a) PRINTS (b) PROSITE

(c) PDB (d) PIR

14. Find out the sequence alignment tool

(a) BLAST (b) PRINT

(c) PROSITE (d) PIR

 Biological Databases 2.25

15. MAtDB contains data for

(a) Mouse (b) Human
(c) E.coli (d) Arabidopsis

ANSWERS

1. (c) 2. (a) 3. (a) 4. (b) 5. (a)

6. (d) 7. (d) 8. (a) 9. (b) 10. (c)
11. (d) 12. (a) 13. (b) 14. (a) 15. (d)
 CHAPTER

3 Biological Sequence Alignment

3.1 INTRODUCTION

Biological Sequence Alignment is one of the most established and

effective protocols in Bioinformatics. The final goal of the sequence
alignment techniques is to find similar sequences to an existing unknown
sequence to come up with some idea about the possible structure-function
activity of the unknown sequence. However, the sequence alignment
techniques have manifold applications, viz.,
(a) Gene Finding
(b) Functional Prediction
(c) Assembly of Genome Sequences

3.2 SOME IMPORTANT TERMS:

The phenomenon of Biological Sequence Alignment often encompass

the following terms:
(a) Query Sequence: It is the unknown sequence. It is the sequence
which is given as an input to find suitable other sequences from
the databases.
(b) Subject: It is the sequence present in the database with which
the unknown sequence is compared.
(c) Pair-wise Sequence Alignment: It is the alignment of two
sequences together.
3.2 Introduction to Bioinformatics

(d) Multiple Sequence Alignment: It is the alignment between more

than two sequences.
(e) Global Sequence Alignment: It is the alignment between
sequences encompassing the entire lengths of the sequences. The
alignment may contain matches, mismatches and gaps.
Seq1: A T T T T G G G C A C G
Seq2: A T T T T G - - C T C G
In the above sequence alignment, there are 9 exact matches,
1 mismatch and two gaps. Gaps would represent deletion
mutations. In the above alignment there are two deletions in Seq2
corresponding to the regions of gaps. It can also be mentioned that
in Seq1 there are two insertions. The gaps are therefore referred
to as INDEL (Insertion Deletion). There is one mismatch where
there is an A in Seq1 and T in the corresponding place in Seq2.
(f) Local Sequence Alignment: It is the alignment of parts of
sequences. Local alignment is generally used to identify the
similarities in some specific regions of proteins, like the active
sites.
(g) Score of the Sequence Alignment: It is the measure of the
effectiveness of the sequence alignment. The total score of a
sequence alignment is the sum total score obtained from the
individual positions and includes matches, mismatches and gaps.
The scores are calculated using available substitution matrices.
(h) Substitution Matrix: A substitution matrix is a collection of
scores for aligning biological sequences (nucleotides or amino
acids) with one another. These scores generally reflect the relative
ease with which one nucleotide or amino acid may mutate into or
substitute for another nucleotide or amino acid. These scores are
used to measure similarity in sequence alignments. Two of the
most important substitution matrices are PAM and BLOSUM.
(i) PAM Matrix: PAM or Point Accepted Mutation is one of the
first amino acid substitution matrices. This matrix was built by
 Biological Sequence Alignment 3.3

calculating the differences in closely related proteins. PAM1

matrix would therefore reflect that there is 1% change in the
sequences of the amino acid residues in proteins belonging to the
same family. PAM1 matrix is used as the basis to calculate the
other matrices. The higher the number of the PAM matrix, the
lower is the sequence similarities between the sequences.
(j) BLOSUM Matrix: BLOSUM or Block Substitution Matrix
is obtained by calculating the differences in evolutionarily
divergent proteins. The substitution probabilities are computed
using the conserved blocks in the multiple sequence alignments
of evolutionarily divergent proteins. It is therefore an empirical
rule to use a higher numbered BLOSUM matrix for aligning two
closely related sequences and a lower number for more divergent
sequences.
(k) E-value: This is the Expectation value. It signifies the reliability
of a sequence alignment. The lesser the E-value is the higher is
the chance of having a significant sequence alignment.
(l) Low-Complexity region: A Low-Complexity region in a
sequence alignment represents a series of repeated amino acids or
nucleic acids which do not make any specific sense. Such Low-
Complexity regions must not be taken into consideration while
performing the sequence alignments.

3.3 BIOLOGICAL IMPORTANCE OF SEQUENCE ALIGNMENT

Sequence alignments are very much useful in bioinformatics to identify

the sequence similarity and to produce phylogenetic trees. The technique
is also used to build homology models of protein structures. However,
the biological significance and relevance of sequence alignments are
not always clear. Alignments are often assumed to reflect some degree
of evolutionary changes between sequences descended from a general
common ancestor. However, it is formally possible that convergent
evolution could occur to generate apparent similarity between proteins
that are evolutionarily unrelated but still are able to perform similar
functions having similar structures.
3.4 Introduction to Bioinformatics

3.4 SEQUENCE ALIGNMENT ALGORITHMS

There are different methods of biological sequence alignment techniques.

The techniques include:
Dot-Plot: This is probably the oldest way of comparing biological
sequences. A dot plot is a visual representation of the comparison between
two sequences. It looks like a rectangular array where each axis of the
array represents one of the two sequences to be compared. First, a window
length is fixed to compare the sequences. Whenever one window in one
sequence resembles another window in the other sequence, a dot or short
diagonal is put at the corresponding position of the rectangular array. Thus,
a diagonal line is drawn when two sequences share similarities over their
entire length from one corner of the dot plot to the diagonally opposite
corner. This might be considered as the global alignments between the
sequences.
Dynamic Programming Algorithm: In biological sequence
alignments, the algorithm proposed by Needleman–Wunsch, is considered
to be one of the most important useful applications of Dynamic
Programming Algorithm. The algorithm first breaks the entire sequence
into small pieces and compares the sequences to come up with a solution
which are then joined to build the total alignment. It is considered to be
the best option for global alignment. The algorithm can be explained by
the following example:
Step1: Two DNA sequences (S1 = GCATGCU and S2 = GATTACA)
are considered. A table is made with the DNA sequences as follows:
G C A T G C U

G
A
T
T
A
C
A
 Biological Sequence Alignment 3.5

Step 2: The best possible arrangement between S1 and S2 are

calculated as:
S1: GCATG-CU
S2: G-ATTACA
The table is then filled up as follows:
G C A T G C U
0 –1 –2 –3 –4 –5 –6 –7
G –1
A –2
T –3
T –4
A –5
C –6
A –7
In the next step, the scores to be used for filling up the table are to be
calculated either from top, left or top-left (diagonal). The scores calculated
from top and left would result in Indels. However, only a score calculated
from the diagonal would represent an exact match between the sequences.
We need to keep a track of the cell which is used to fill in the rows. As
there are no top or top left cells for the second row, -1, -2 etc. are used to
fill the row. The same is true for the second column as well. Following
this rule, the entire table is filled up as follows:
G C A T G C U
0 –1 –2 –3 –4 –5 –6 –7
G –1 1 0 –1 –2 –3 –4 –5
A –2 0 0 1 0 –1 –2 –3
T –3 –1 –1 0 2 1 0 –1
T –4 –2 –2 –1 1 1 0 –1
A –5 –3 –3 –1 0 0 0 –1
C –6 –4 –2 –2 –1 –1 1 0
A –7 –5 –3 –1 –2 –2 0 0
3.6 Introduction to Bioinformatics

Smith-Waterman Algorithm: The Smith-Waterman algorithm is

basically used for local sequence alignment. Local sequence alignment
aims to find local regions of similarities between nucleotide or protein
sequences. The algorithm is simple and analogous to the Needleman–
Wunsch algorithm. The difference lies in the fact that instead of using
negative numbers in the cells, the cells are filled with zero values. The
basic motivation behind the Smith-Waterman algorithm is to provide
reliable degrees of similarities between local regions of nucleotide or
protein sequences which are otherwise distantly related. Local alignment
techniques are used in FASTA and BLAST.
In database search techniques, such as BLAST, statistical methods
are used to evaluate the likelihood of a particular alignment between
regions of sequences arising by chance given the size and composition
of the database being used for searching. In particular, the probability of
finding a given alignment by chance increases if the database consists
only of sequences from the same organism as used in the query sequence.
Repetitive sequences in the database or query can also distort both
the search results and the assessment of statistical significance scores.
BLAST automatically filters such repetitive sequences in the query to
avoid apparent hits that are statistical artifacts. There are several versions
of BLAST alignments.
• blastp: This tool uses the amino acid sequence of a protein as the
query sequence to find its similar proteins in a protein sequence
database.
• blastn: This tool uses the nucleotide sequence as the query
sequence to find its similar nucleotides in a nucleotide sequence
database.
• blastx: This tool uses a nucleotide sequence as the query sequence
which is translated in all reading frames to generate a protein
which is then used as the query sequence to search for similar
sequences against a protein sequence database.
• tblastn: This tool uses a protein sequence as the query sequence
which is translated in all reading frames to generate a nucleotide
 Biological Sequence Alignment 3.7

sequence which is then used as the query sequence to search for

similar sequences against a nucleotide sequence database.
• tblastx: This tool uses the nucleotide query sequence with six
frame translation against the six-frame translations of a nucleotide
sequence database.
• BLAST2: This is a tool that allows the user to upload his or her
own sequence of interest to be compared with a sequence database
or with another sequence.

3.5 STRUCTURAL ALIGNMENT

Structural alignments are the techniques to determine the similarities

between the structures of proteins or RNAs. These methods use
information about the secondary and tertiary structure of the protein or
RNA molecule to identify the aligned regions in the molecules. Such
methods consider not only the structures but also the sequences of
the concerned molecules. The structural alignment methods typically
come up with a local sequence alignment between the biomolecules
under considerations. The local sequence alignment is produced as
a part of the structural alignment results. The methods are generally
used for comparison between two or more structures of biomolecules
and their corresponding sequences. The only limitation of the work is
that the alignment methodologies are dependent on the availability of
structural information such that they can only be used for the analysis of
sequences of biomolecules for which structural information are already
available. The structural information is generally obtained from X-ray
crystallography or NMR spectroscopy. It is, however, an important fact
that both the protein and RNA structures remain highly conserved during
evolution. Therefore, alignments involving structural information can be
considered to be more reliable in case of sequences which are obtained
from very distantly related organisms. Such sequences are known to be
diverged so extensively that sequence comparison cannot reliably detect
their similarities.
3.8 Introduction to Bioinformatics

The most important aspect of structure based sequence alignments

is that such alignments can be used as the “gold standard” for the
determination of protein structures by homology modelling techniques
for which sequence similarity between the target and template is the most
vital information. In case homology-based protein structure prediction
methodologies, the structurally similar regions in proteins should align
properly in their sequence parts. Such structurally conserved regions, if
found to be conserved in sequences as well, would provide additional
reliability of the structure prediction method. However, the structural
similarity scores cannot be used in homology based protein structure
prediction methods directly as the target sequence does not possess any
structure.
The most important methods for structural alignments are: DALI,
SSAP and Combinatorial Extensions.
DALI (http://ekhidna.biocenter.helsinki.fi/dali_lite/start)

DALI search method is one of the most important structural alignment

tools. This method generates a distance matrix by considering the amino
acid sequences extracted from the structures of the proteins provided
as inputs. This method generates the structural alignment between the
proteins and the method of identification of structural similarities depends
on the contact similarity patterns generated from the amino acid sequences
of the proteins. The entire protein sequences are divided into several hexa-
peptides and a similarity analysis is done between these hexa-peptides.
Then, the DALI program creates pair-wise sequence alignments between
the proteins and the process is repeated to find neighboring sequences
from the protein data bank. The tool then constructs a FSSP which is a fold
classification method using structural alignments between neighboring
protein structures.
SSAP (https://doi.org/10.1016/S0076-6879(96)66038-8)

SSAP is the Sequential Structure Alignment Program. It is based on

dynamic programming algorithm for alignments of structures. This
method is based on the analysis of atom-to-atom vectors in structure space
as comparison points. It has been extended since its original description
 Biological Sequence Alignment 3.9

to include multiple as well as pairwise alignments, and has been used in

the construction of the CATH (Class, Architecture, Topology, Homology)
hierarchical database classification of protein folds.The CATH database
can be accessed at CATH Protein Structure Classification.

FINAL WORDS
There are many methods to determine the biological sequence alignments.
All the methods are fairly accurate. The approximations used in the
algorithms are also quite good. The new methods that are being developed
are extensions of the previous methods. Not only the sequences of DNA
and proteins are analyzed but the sequences of RNA, such as expressed
sequence tags and full-length mRNAs, are also aligned to sequenced
genomes in order to find the locations of genes. This sequencing also
provides information about alternative splicing and RNA editing. The
biological sequence alignment techniques are also important parts of
genome assembly. In genome assembly techniques, the sequences are
aligned to the whole genomes to find overlapping portions. These would
provide information about contigs. The contigs are long stretches of
sequences. The analysis of nucleic acid sequences would also provide
information about the Single Nucleotide Polymorphism (SNP).
The techniques for analyses of the biological sequences have found
their applications in non-biological fields as well. The most important of
such applications is in natural language processing and in social sciences.
In these fields, the Needleman-Wunsch algorithm is usually used to
identify the optimal matching. The word selection methods in natural-
language processing were dependent on the process of multiple sequence
alignment techniques from bioinformatics to generate linguistic versions
of computer-generated mathematical proofs. On the other hand, in the
field of historical and comparative linguistics, sequence alignment tools
have been used to partially automate the comparison method by which
linguists traditionally reconstruct languages. Interestingly, business and
marketing professional also use multiple sequence alignment techniques
to analyze a series of purchases over time.
3.10 Introduction to Bioinformatics

Some important software tools

BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi): All purpose
sequence alignment tool.
CUDASW++ (http://cudasw.sourceforge.net/homepage.htm): Smith-
Waterman sequence alignment tool.
DIAMOND (http://ab.inf.uni-tuebingen.de/software/diamond/
welcome.html): Multi-purpose sequence alignment tool.
FASTA (https://fasta.bioch.virginia.edu/fasta_www2/): A tool for
sequence comparison.
HMMER (http://hmmer.org/): A sequence database search tool.
LAMBDA (http://www.seqan.de/apps/lambda/): A tool for local
sequence alignment.
USEARCH (http://drive5.com/usearch/): A tool for ultra-fast
sequence alignment.
PSISEARCH (https://www.ebi.ac.uk/Tools/sss/psisearch/): It is used
for searching of distant homologues.
ScalaBLAST (https://omics.pnl.gov/software/ScalaBLAST.php): A
multipurpose sequence alignment tool.
Sequilab (http://www.sequilab.org/): A multipurpose software suit.
SSEARCH (https://www.ebi.ac.uk/services): A tool for Smith-
Waterman sequence alignment. It is slower than FASTA but gives a
better result.
SWAPHI(http://swaphi.sourceforge.net/homepage.htm#latest): It is
used for searching on Intel Xeon PHI co-processor.
AlignMe(http://www.bioinfo.mpg.de/AlignMe): This is for the
alignment of membrane proteins.
SWIPE (http://dna.uio.no/swipe/): It is used for Smith-Waterman
sequence searching.
BioConductor (http://www.bioconductor.org/): This is a software
suite for analyses of high-throughput sequence data.
 Biological Sequence Alignment 3.11

Bio-Perl (http://bioperl.org/): Perl programming language for

Bioinformatics.
DNASTAR(https://www.dnastar.com/t-sub-solutions-molecular-
biology-sequence-alignment.aspx): A multipurpose software suite for
sequence analyses.
Genome Compiler (http://www.genomecompiler.com/): A tool for
multipurpose sequence analyses.
G-PAS (http://gpualign.cs.put.poznan.pl/project-gpu-pairAlign.html):
A tool for global sequence alignment.
JALIGNER (http://jaligner.sourceforge.net/): A software tool for
dynamic programming.
ROBETTA (https://web.archive.org/web/20150819163428/http://
www.robetta.org/): A protein- sequence to structure alignment tool
LALIGN (https://embnet.vital-it.ch/software/LALIGN_form.html):
A tool for local similarity search.
NW-align (https://zhanglab.ccmb.med.umich.edu/NW-align/): A tool
for Needleman-Wunsch dynamic programming.
Matcher (https://galaxy.pasteur.fr/?form=matcher): A tool for local
alignment.
Stretcher (https://galaxy.pasteur.fr/?form=matcher): A sequence
alignment tool.
MCALIGN (http://www.homepages.ed.ac.uk/pkeightl//mcalign/
mcinstructions.html): A tool for alignment of non-coding DNA sequences.
PATH (http://bioinfo.lifl.fr/path/path.php): A tool for searching of
distant protein homologues.
SEQALIGN (http://www-hto.usc.edu/software/seqaln/seqaln-query.
html): A tool for sequence alignment.
SWIFT (https://bibiserv.cebitec.uni-bielefeld.de/swift/): A tool for
local alignment.
UGENE (http://ugene.net/): A repository of sequence alignment tools.
3.12 Introduction to Bioinformatics

YASS (http://bioinfo.lifl.fr/yass/yass.php): A software tool to search

for genomic similarity.
ALE (http://www.red-bean.com/ale/): Software tool for multiple
sequence alignment.
BALI-PHY (http://bali-phy.org/): This tool is for multiple sequence
alignment. It has a downloadable version.
CHAOS-DIALIGN (http://dialign.gobics.de/chaos-dialign-
submission) : A tool for both pair-wise and multiple sequence alignment.
CLAUTALW2 (https://www.ebi.ac.uk/Tools/msa/clustalw2/): A tool
for multiple sequence alignment
CODONALIGNER (www.codoncode.com/aligner): A tool for
sequence assembly and alignment.
COMPASS (prodata.swmed.edu/compass/compass_advanced.php):
A tool for multiple alignments of protein sequences with a statistical
comparison of the alignments.
DECIPHER (http://decipher.cee.wisc.edu/): R programming language
platform for sequence analysis.
DIALIGN (http://dialign-tx.gobics.de/): A tool for segment based
multiple sequence alignment.
DNADyanamo (http://www.bluetractorsoftware.co.uk/): A tool for
analysis of DNA sequences.
DNASTAR (https://www.dnastar.com/t-sub-solutions-molecular-
biology-sequence-alignment.aspx): A tool for analysis of DNA sequences.
MULTALIN (http://multalin.toulouse.inra.fr/multalin/): A tool for
analysis of biological sequences.
LAGAN (http://genome.lbl.gov/vista/lagan/submit.shtml): A tool for
analysis of biological sequences.
MUSCLE (http://www.drive5.com/muscle/): A tool for sequence
alignment.
PMFastR (http://genome.ucf.edu/PMFastR/): A tool for analysis of
RNA sequences
 Biological Sequence Alignment 3.13

ePROBALIGN (http://probalign.njit.edu/probalign/login): A tool for

analysis of RNA and Protein sequences.
PROBCONS (http://probcons.stanford.edu/index.html): A
probabilistic approach to multiple sequence alignment of proteins.
PROMALS3D (http://prodata.swmed.edu/promals3d/promals3d.
php): A tool for multiple sequence alignment of protein sequences and
structures.
REVTRANS (http://www.cbs.dtu.dk/services/RevTrans/): A tool
that takes the DNA sequence and converts it to amino acid sequence of
proteins and then performs the sequence alignment.
T-COFFEE (http://www.tcoffee.org/): A tool for multiple sequence
alignment.
GENOME-VISTA (http://genome.lbl.gov/cgi-bin/GenomeVista): A
tool for whole genome analysis.
FLAK (http://www.flakbio.com/): A tool for comparison of whole
genomes.
PLASTRNA (http://plastrna.njit.edu/rnagenome/index): A tool for
the analysis of non-coding RNA.

SOME IMPORTANT QUESTIONS AND ANSWERS

1. The method of aligning two sequences by searching for similar sequence
patterns that is in the same order in the sequences is called
(a) sequence alignment
(b) pair wise alignment
(c) multiple sequence alignment
(d) all of these
2. A sequence alignment method that is suitable for aligning closely related
sequences is called
(a) multiple sequence alignment
(b) pair wise alignment
3.14 Introduction to Bioinformatics

(c) global alignment

(d) local alignment
3. A sequence alignment method that tries to align the entire sequence is
called
(a) multiple sequence alignment
(b) pair wise alignment
(c) global alignment
(d) local alignment
4. A sequence alignment method suitable for finding out conserved patterns
in DNA or protein sequences is known as
(a) multiple sequence alignment
(b) pair wise alignment
(c) global alignment
(d) local alignment
5. A sequence alignment method of aligning many sequences
simultaneously is known as
(a) multiple sequence alignment
(b) pair wise alignment
(c) global alignment
(d) local alignment
6. Sequence alignment methods help
(a) to identify evolutionary relationships
(b) to identify the functions of newly synthesized genes
(c) to propose new members of gene families
(d) all of these
7. A sequence alignment method that tries to align regions with high level
of matches without considering the alignment of rest of the sequences
is referred to as
(a) multiple sequence alignment
 Biological Sequence Alignment 3.15

(b) pair wise alignment

(c) global alignment
(d) local alignment
8. Find the odd one out
(a) Rasmol
(b) BLAST
(c) FASTA
(d) Clustal W
9. Which of the following servers is a sequence alignment tool provided
by NCBI
(a) Chime
(b) BLAST
(c) FASTA
(d) Clustal W
10. Which of the following is a multiple sequence alignment tool?
(a) Clustal W
(b) Chime
(c) Dismol
(d) PDB

ANSWERS

1. (b) 2. (c) 3. (c) 4. (d) 5. (a)

6. (d) 7. (d) 8. (a) 9. (b) 10. (a)
 CHAPTER

4 Molecular Modelling

4.1 INTRODUCTION

Molecular modelling techniques represent the combination of all

theoretical methods and computational techniques used to model or
mimic the behaviour of molecules. The techniques are used in various
different fields like computational chemistry, drug design, computational
biology and materials science for studying molecular systems ranging
from small chemical systems to large biological molecules and material
assemblies. The common feature of molecular modelling techniques is
the atom level description of the molecular systems. This may include
treating atoms as the smallest individual unit (the molecular mechanics
approach), or explicitly modelling electrons of each atom (the quantum
chemistry approach).
The molecular modelling techniques include the following:
(a) Homology Modelling
(b) Threading
(c) Ab Initio Modelling
(d) Molecular Docking
Before beginning a few words on protein structure are worth
mentioning.
Proteins are polymers made up of amino acids. The amino acids
form the protein by losing a water molecule and are linked by peptide
linkages (Figure 1).
4.2 Introduction to Bioinformatics

The R groups represent the side chains

Figure 1: Peptide linkage. The central Ca atom joins the two planes
together. The peptide linkage is the linkage between C=O and N-H
groups. The peptide linkage is planar and generally the O atom of the C=O
group and the H atom of the N-H group remain trans to each other. The
dihedral angle between the N-Ca is called the φ angle and that between the
Ca- C’ is called the ψ angle. The peptide linkages can assume only some
specific values of these angles and those values are called the allowed
values. The distributions of the values of φ and ψ angles are obtained
from Ramachandran plot.
By convention, a chain under 40 amino acid residues is often identified
as a peptide, rather than a protein. To be able to perform their biological
functions, proteins fold into one or more specific spatial conformations.
The underlying forces for the formation of the spatial conformations of
proteins are the various types of non-covalent interactions such as hydrogen
bonding, ionic interactions, Van der Waals forces, and hydrophobic
packing. The understanding of protein functions at the molecular level
requires the identification of their three-dimensional structures.
The protein structure is divided mainly into four categories:
(i) The primary structure: It is the linear sequence of the amino
acid residues in the protein.
(ii) The secondary structure: It is the highly regular local sub-
structure in a protein. The major secondary structures present in a
protein are the alpha helix, beta strand and turn. These secondary
structures have definite patterns of hydrogen bonds between the
 Molecular Modelling 4.3

main-chain peptide groups. They have a regular geometry and are

known to possess specific values of the dihedral angles ψ and φ
on the Ramachandran plot. The amino acids Ala, Glu, Leu, Met
are mostly observed in the helical regions of the amino acids.
On the other hand Pro, Gly, Tyr, Ser are not so common in the
helical regions. The hydrogen bonding pattern in alpha helix is n to
n + 4, where the carbonyl oxygen of the peptide bond of the nth
amino acid forms hydrogen bonding with the –NH portion of the
peptide bond of the n + 4 th amino acid. There are two other types
of helical structures in proteins. They are called the 310-helix and
p-helix. In the 310-helix the hydrogen bonding pattern is n to
n + 3. It is tighter than alpha helix. On the other hand, the p-helix
has n to n + 5 hydrogen bonding pattern. It is looser than alpha
helix. The secondary structures sometimes form super secondary
structures. The different beta strands combine together to form a
sheet like structure called beta sheet. In the beta sheets the different
beta strands remain hydrogen bonded. The beta sheets can be
of two different types: parallel and anti-parallel. In anti-parallel
beta sheets there are strands for which one sides remain buried
whereas the other sides remain exposed. Anti-parallel beta sheets
are more commonly observed and in them the hydrophobic and
hydrophilic amino acids alternate. In the parallel beta sheets the
strands remain mostly buried. The terminal amino acid residues
in parallel beta strands are hydrophilic whereas the central amino
acids are hydrophobic.
(iii) The super-secondary structure: A super-secondary structure
is defined as a compact three-dimensional protein structure of
several adjacent elements of secondary structure. These are also
referred to as motifs. The protein motifs also possess loops and
other unstructured regions.
(iv) The tertiary structure: Tertiary structure of a protein is the
three-dimensional structure of a monomeric or multimeric protein
molecule. The secondary structural elements, viz., alpha helix and
beta strands, are folded into compact shapes of tertiary structure.
4.4 Introduction to Bioinformatics

The folding of the secondary structural elements is facilitated

by non-specific hydrophobic interactions as well as salt bridges,
disulfide linkages and hydrogen bonds.
(v) The quaternary structure: The quaternary structure is the three-
dimensional structure of a multimeric protein. It represents the
relative orientations of the different protein chains with respect
to each other. The quaternary structure is also stabilized by
hydrophobic interactions as well as salt bridges, disulfide linkages
and hydrogen bonds.
(vi) Protein domains: A domain or a structural domain in a protein
molecule is an element of the protein’s overall structure that is
self-stabilizing. The domains have the capacity of being folded
independently of the rest of the protein chain. Domains take part
in the biological function of the protein they belong to.
(vii) Protein fold: It is the general protein architecture. A protein fold
may be defined by the way the secondary structural elements in
the protein are arranged relative to each other in space.
Proteins perform most if not all of the biological processes. Thus,
the basic aim of the molecular modelling techniques is to analyze the
structure function relationships of proteins. In most of the cases proteins
interact among themselves to form protein-protein complexes or proteins
interact with other different ligands (such as, macro-molecules like DNA,
RNA, carbohydrates, lipids; small organic or inorganic molecules or ions).
Molecular modelling techniques are used to build the protein-protein or
protein-ligand complexes. The bindings of proteins may be transient or
permanent. The interfaces of the interacting proteins have several different
characteristics. The identification of the interface residues may shed light
in many important aspects like drug development, analyses of molecular
pathways, generation of protein mimetics as well as understanding of
disease mechanisms and development of new docking methodologies
to build structural models of protein complexes. In the next few pages,
the basics of protein interactions are presented.
 Molecular Modelling 4.5

Amino acid conformations in proteins

There are 20 different amino acids. They are classified based on the
physico-chemical characteristics of their side chains. The various
physicochemical properties of side chains are:
Size, shape, hydrophobicity, charge, hydrogen bonding
The general properties of amino acids:

Glycine (Gly) (G): The amino acid G has the following characteristics:
• It is the smallest amino acid
• It has no side chain
• It is highly flexible and is commonly observed in turns
• It is generally present as the outlier in Ramachandran plot
• It is ambivalent. It can be located inside or outside of the protein
structure; buried or solvent exposed
Alanine (Ala) (A): The amino acid A has the following characteristics:
• It is small
• It is ambivalent. It can be located inside or outside of the protein
structure; buried or solvent exposed
Valine (Val) (V) : The amino acid V has the following characteristics:
• It has a large side chain
• It is hydrophobic
• It remains usually buried in the protein cor58e
Leucine (Leu) (L): The amino acid L has the following characteristics:
• It has a large side chain with an additional –CH2 group attached
to the side chain as compared to V. It is larger than V.
• It is hydrophobic
• It remains usually buried in the protein core
Isoleucine (Ile) (I): The amino acid I has the following characteristics:
4.6 Introduction to Bioinformatics

• It has a branched side chain

• It is hydrophobic
• It remains usually buried in the protein core
• It is important for protein stability and ligand binding
Phenylalanine (Phe) (F): The amino acid F has the following
characteristics:
• It has a nonpolar, hydrophobic side chain
• It is an aromatic amino acid containing an aromatic ring as the
side chain
• It remains usually buried in the protein core
• The p electrons of the aromatic ring of the side chain often stack
with other aromatic residues
• The p stacking increases protein stability
Methionine (Met) (M): The amino acid M has the following
characteristics:
• It contains sulfur-containing, non-polar flexible side chain
• It remains usually buried in the protein core
• The side chain is able to form S-mediated hydrogen bonds
Proline (Pro) (P): The amino acid P has the following characteristics
• It is a cyclic, chain imino acid
• The side chain is unable to form traditional hydrogen bonding
interactions like some other amino acids
• The amino acid is able to cause a bend of a-helical structure of
a protein
• The amino acid is very rigid and is generally present as cis-Pro
• This amino acid is important for protein folding
• It is generally present at the end of a-helix, turn, loops
 Molecular Modelling 4.7

Serine (Ser) (S): The amino acid S has the following characteristics:
• The amino acid contains a -OH group in the side chain
• The side chain of the amino acid remains uncharged at
physiological pH
Threonine (Thr) (T): The amino acid T has the following
characteristics:
• The amino acid has a small, hydrophilic side chain
• The amino acid takes part in post-translational modifications
• This amino acid is found to be present in many enzymes
• This amino acid is also involved in metal ion binding
Tyrosine (Tyr) (Y): The amino acid Y has the following characteristics:
• This amino acid has an aromatic side chain with a hydrophilic
–OH group attached to the aromatic ring
• This amino acid helps in identification of proteins by UV-Vis
spectroscopy
• It may remain buried in the protein core or exposed to the solvents
• This amino acid is found in be present in many enzymes
• The amino acid takes part in post-translational modifications
Tryptophan (Trp) (W): The amino acid W has the following
characteristics:
• It has the largest side chain
• The side chain of this amino acid is aromatic
• This amino acid helps in identification of proteins by UV-Vis
spectroscopy
• The nitrogen atom present in the side of the amino acid can form
hydrogen bonds
• This amino acid is found to be present in many enzymes
• The nitrogen atom of the side chain often remains exposed to
surface
4.8 Introduction to Bioinformatics

Cysteine (Cys) (C): The amino acid C has the following

characteristics:
• It contains sulfur-containing, non-polar flexible side chain
• The side chain of the amino acid is small
• The side of the amino acid can form disulfide linkages
• This amino acid is found to be present in many enzymes
• This amino acid is also involved in metal ion binding
AsparagiNe (Asn) (N): The amino acid N has the following
characteristics:
• The amino acid has uncharged polar side chain
• It may remain buried in the protein core or exposed to the solvents
• The amino acid takes part in post-translational modifications
Glutamine (Gln) (Q): The amino acid Q has the following
characteristics:
• The amino acid has uncharged polar side chain
• The side chain of this amino acid is longer than that of N
• It may remain buried in the protein core or exposed to the solvents
Histidine (His) (H): The amino acid H has the following characteristics:
• The amino acid has aromatic side chain
• The side chain of the amino acid remains positively charged when
both the nitrogen atoms are protonated
• This amino acid is found to be present in many enzymes
• This amino acid is also involved in metal ion binding
Aspartate (Asp) (D): The amino acid D has the following
characteristics:
• The amino acid has negatively charged side chain
 Molecular Modelling 4.9

• This amino acid is found to be present in many enzymes

• This amino acid is also involved in metal ion binding
Glutamate (Glu) (E): The amino acid E has the following
characteristics
• The amino acid has negatively charged side chain
• This amino acid is found to be present in many enzymes
• This amino acid is also involved in metal ion binding
Lysine (Lys) (K): The amino acid K has the following characteristics:
• The amino acid has positively charged side chain
• This amino acid is found to be present in many enzymes
• The amino acid has long flexible side chain
Arginine (Arg) (R): The amino acid R has the following characteristics:
• The amino acid has positively charged side chain
• This amino acid is found to be present in many enzymes
• The amino acid has long flexible side chain
The amino acid side chains can have different conformations. They
are called rotamers. A database is present where the three dimensional
coordinates of the atoms of the side chains of the amino acids in possible
different conformations are present. Such a library is called rotamer
library. In general the staggered conformation of the side chain is the
most favoured conformation.
Basic definitions:
• Monomer: A single unit of an assembly.
• Polymer: An assembly of single monomeric units i.e., Polymer
= (Monomer)n
If n = 2, the polymer is called dimer
If n = 3, the polymer is called trimer etc.
• Protomer: The monomeric constituent units of a protein having
two or more monomeric protein chains.
4.10 Introduction to Bioinformatics

• Homomer: A protein with the same monomeric constituents.

• Heteromer: A protein with different monomeric constituents.
• Obligate and non-obligate protein-protein complexes: A
protein-protein complex where the individual partners (protomers)
are not stable by themselves, such as, the Arc repressor dimer. If
the individual protomers can exist on their own, the complex is
then referred to as non-obligate. As for an example, the antibody-
antigen complexes are non-obligate ones.
• Transient and permanent complexes: If the associations
between the protomers in a protein-protein complex is weak and
are in a dynamic equilibrium in solution where it is broken and
formed continuously the complex is called transient complex,
such as, the non-obligate homo-dimer of sperm lysine.
On the other hand, if the associations between the protomers require
molecular switch to break, the complex is called permanent complex. The
hetero-trimeric G protein forms a permanent complex in presence of GDP.
It is to be noted that PPIs cannot be distinctly classified as transient
and permanent rather a continuum exists between them. The stabilities
of all these complexes depend upon physiological conditions and cellular
environments.
• Accessible surface area (ASA): It is the fraction of the total van
der Waal’s surface of an atom that can come in contact with other
atoms, specifically water. In case of proteins the accessible surface
area is calculated for each atom of each amino acid residue.
• Interface patch: The interface is the contact area of the proteins.
It involves those residues of the proteins, which have the ASA of
their side chains decreased by >1 Å2 on complex formation.
• Relative accessible surface area (RSA): It is defined as the ratio
of ASA to the maximal accessibility of each amino acid.
• Gap volume and Gap index: The gap volume is defined as the
volume enclosed between any two protein molecules delimiting
the boundary by defining a maximum allowed distance from both
 Molecular Modelling 4.11

the interfaces. The Gap index is calculated as Gap index = gap

volume / interface ASA
• Surface patch: Surface residues are the amino acid residues of a
protein with a relative accessible surface area of > 5%. A surface
patch is the central surface accessible residue and n nearest surface
accessible neighbors, where n is the size of the patch in terms of
the number of residues. A mean relative ASA for each patch is
calculated as
Patch ASA (Å2) = sum of the RSAs of the amino acid residues in the
patch / number of amino acid residues in the patch
• Solvation potential: This is the measure of the propensity of an
amino acid to get solvated. It is used to quantify the tendency of
a patch to be exposed to solvent or buried in the interface of a
protein-protein complex.
• Residue interface propensities (RIP): It represents the tendency
of the amino acid residues of a protein to be on the interface of
the protein-protein complex. The patch interface propensity (PIP)
is calculated as
PIP = sum of the natural logarithms of the RIPs of the amino acid
residues in the patch / number of amino acid residues in the patch
• Hydrophobicity: The surface patch hydrophobicities are defined
as
Patch Hydrophobicity = Hydrophobicity value of the amino acid

residues in a patch / number of amino acid residues in the patch
• Planarity: This quantity of the surface patch is evaluated by
calculating the root mean square (rms) deviation of all the atoms
present on the surface patch from the least square plane that passes
through the atoms.
• Protrusion index (PI): This quantity gives an idea of how much
a residue sticks out from the surface of a protein. The patch PI is
calculated as
4.12 Introduction to Bioinformatics

Patch PI = sum of the PIs of the amino acid residues in the patch /
number of amino acid residues in the patch
A few aspects of protein-ligand interactions
Ligand is a substance that binds to a biomolecule, mainly proteins, to
serve some specific purposes. In protein-ligand binding, the ligand is
usually a signaling molecule that binds to a specific site on a target protein.
In case of protein-ligand binding the binding occurs by intermolecular
forces, such as ionic bonds, hydrogen bonds and van der Waals forces.
However, association between the ligand and its receptor protein is
usually reversible. It is well established that an actual irreversible covalent
bonding between a ligand and its target molecule is rare in biological
systems.
A ligand binding to its receptor protein alters its chemical conformation
i.e., the three-dimensional shape of the receptor protein. It is known that
the conformational state of a receptor protein determines its functional
state. In biological systems, the ligands can be of different types. They
include substrates, inhibitors, activators, and neurotransmitters. The
tendency or strength of ligand binding to its receptor protein is called
affinity. The binding affinity of ligands is determined not only by direct
interactions, but also by solvent effects that are known to play a dominant
indirect role in driving non-covalent receptor-ligand bindings in solutions.
The interactions of most ligands with the binding sites of their
corresponding receptors can be characterized in terms of binding affinity.
In general, high-affinity ligand binding results from greater intermolecular
forces between the ligand and its receptors. On the other hand, low-
affinity ligand binding involves less intermolecular force between the
ligand and its receptor. Another important aspect of high affinity binding
is that high-affinity binding involves a longer residence time for the
ligand at its receptor binding site whereas the low-affinity binding leads
to comparatively lesser binding time. High-affinity binding of ligands to
their receptors is often physiologically important and in these cases some
of the binding energy can be used to cause a conformational change in
the receptor site. This results in some altered behavior of an associated
ion channel or enzyme.
 Molecular Modelling 4.13

A ligand is called the agonist for that particular receptor if the ligand
can bind to the receptor, alter the function of the receptor, and trigger
a physiological response. The agonist for a specific receptor can be
characterized on the basis of how much physiological response can be
triggered and in terms of the concentration of the agonist that is required
to produce the physiological response. For agonists binding to high-
affinity ligand binding sites in receptors, a relatively low concentration
of a ligand is adequate to maximally occupy a ligand-binding site and
trigger a physiological response. On the other hand, low-affinity binding
implies that a relatively high concentration of a ligand is required before
the binding site in the receptor is maximally occupied and the maximum
physiological response to the ligand is achieved.
On a similar note, an antagonistic ligand is the one that after binding
to its receptor fails to stimulate the receptor and brings about a change
in the receptor.
The ligands can be of the type selective and non-selective. Selective
ligands have their tendencies to bind to very limited types of receptor
proteins. On the other hand, non-selective ligands bind to several types
of receptor proteins. This has got a huge pharmaceutical importance.
Drugs that are non-selective tend to have more adverse effects, because
they bind to several other receptors in addition to the one generating the
desired effect. This leads to side effects.
There is another class of ligands referred to as bivalent ligands. The
bivalent ligands consist of two connected molecules as ligands, and are
used in scientific research to detect receptor dimers and to investigate
their properties. Bivalent ligands are usually large molecules and they
tend not to be ‘drug-like’, limiting their applicability in clinical settings.
Molecular modelling techniques utilize all the basic definitions of
protein structures and come up with plausible models of single proteins,
protein-protein and protein-ligand complexes. In the next few pages the
basic principles of molecular modelling techniques will be described. As
mentioned earlier molecular modelling techniques depend on molecular
mechanics and quantum chemistry principles.
4.14 Introduction to Bioinformatics

A brief overview of molecular mechanics principles

Molecular mechanics is one of the basic aspects of molecular modelling

techniques. The underlying principle of molecular mechanics is the use
of classical mechanics or Newtonian mechanics to describe the physical
basis behind the models. In molecular models the atoms are considered
to be a combination of the nucleus and electrons collectively and they
are depicted as point charges with an associated mass. The atoms are
considered to be joined by chemical bonds. The bonds are like springs
and they follow Hook’s law. There are non-bonded interactions in the
molecules which are known as van der Waals forces. The van der Waals
forces are described by Lennard-Jones potential. On the other hand, the
electrostatic interactions are computed based on Coulomb’s law. Atoms
are assigned coordinates in Cartesian space or internal coordinates. The
atoms can also be assigned velocities in dynamical simulations. The
atomic velocities are related to the temperature of the system which is a
macroscopic quantity. The collective mathematical expression of all the
interaction terms is known as a potential function and is related to the
system internal energy (U) which is a thermodynamic quantity equal to the
sum of potential and kinetic energies. Methods to minimize the potential
energy are known as energy minimization techniques (e.g., steepest
descent and conjugate gradient). On the other hand, the methods that
model the behaviour of the system with propagation of time are known
as molecular dynamics. The energy terms considered are presented as:
E = Ebonds + Eangle + Edihedral + Enon-bonded
Enon-bonded = Eelectrostatic + Evan der Waals
This function, referred to as a potential function, computes the
molecular potential energy as a sum of energy terms that describe the
deviation of bond lengths, bond angles and torsion angles away from
equilibrium values, plus terms for non-bonded pairs of atoms expressed
by van der Waals and electrostatic interactions. The set of parameters
consisting of equilibrium bond lengths, bond angles, partial charge values,
force constants and van der Waals parameters are collectively known as the
force field. However, different implementations of molecular mechanics
 Molecular Modelling 4.15

use different mathematical expressions and different parameters for the

potential function. The common force fields in use today have been
developed by using high level quantum calculations and/or fitting to
existing experimental data. The technique known as energy minimization
is used to find positions of zero gradients for all atoms. In other words, a
local energy minimum is generally found. Lower energy states are more
stable and are commonly investigated because of their role in chemical
and biological processes. A molecular dynamics simulation, on the other
hand, computes the behaviour of a system as a function of time. It involves
solving Newton’s laws of motion, principally the second law, F=m.a.
Where, F = Exerted force, m = Mass of the body, a = acceleration
Integration of Newton’s laws of motion, using different integration
algorithms, leads to atomic trajectories in space and time. The force
on an atom is defined as the negative gradient of the potential energy
function. The energy minimization technique is useful to obtain a static
picture of the molecular system to compare the states of similar systems.
On the other hand, molecular dynamics provides information about the
dynamic processes with the intrinsic inclusion of temperature effects. The
molecules can be modeled either in vacuum or in the presence of a solvent
such as water. The modeled structures may be simulated in vacuum or in
presence of a solvent such as water. The former type of simulation is called
the gas-phase simulations, while the latter which include the presence of
solvent molecules is referred to as explicit solvent simulations. However,
the inclusion of solvent molecule is very much time consuming. Thus,
the effect of solvent may be estimated using an empirical mathematical
expression. Such simulations are known as implicit solvation simulations.
The details of molecular mechanics force-fields are presented in the
following section.

Molecular Mechanics Force Field

Force-fields link a structure’s atomic coordinates to a potential energy.

Force fields for molecular systems can be interpreted in terms of a
relatively simple four component picture of the intra- and inter-molecular
4.16 Introduction to Bioinformatics

forces within the system. The molecular system’s potential energy (E) in
a given conformation is a sum of individual energy terms.
E= Ecovalent + E noncovalent
where, the components of the covalent and non covalent contributions
are given by the following summations:
Ecovalent = Ebond + Eangle + E dihedral
Enoncovalent = Evan der Waals +Eelectrostatic
One functional form of force field is
V (r ) = ∑ k (b − b ) + ∑ k (θ − θ )
bonds
b 0
2

angles
θ 0
2

+ ∑
dihedrals
kφ (1 + cos(nφ − φ0 )) ∑
impropers
kψ (ψ − ψ 0 ) 2

 σij  12  σij   qi q j
+ ∑ 4εj   −    + ∑
non-bonded  rij   rij   non-bonded ε D rij
pairs (i, j ) pairs (i, j )

V(r) denotes the potential energy, which is a function of the positions

(r) of N particles usually atoms. The first term denotes the interaction
between pairs of bonded atoms, a harmonic potential that gives the
increase in energy as the bond length b deviates from its reference
value b0. The second term is a summation over all valence angles in
the molecule and the third and fouth terms are torsional potential that
model how the energy changes as bond rotates. The non-bonded term is
usually modelled using a Coulomb potential term for electrostatic and a
Lennard-Jones potential for van der Waals interactions.This is calculated
 Molecular Modelling 4.17

between all pairs of atoms (i and j) that are in different molecules or in

the same molecule but separated by at least three bonds, i.e.; have a 1 n
relationship where n ≥ 4.
Common molecular nodelling force fields and software tools:
One of the widely used packages is AMBER (Assisted Model Building
with energy refinement). AMBER was developed with an aim to simulate
the biological macro molecule especially for protein and nucleic acid.
Later on the force field was modified for the simulation of carbohydrate.
The (Chemistry at Harvard Macromolecular Mechanics) CHARMM, is
a very important and highly capable force field and it is also a software
package used for molecular mechanics and dynamics simulations
of biological macromolecules such as proteins, nucleic acids and
carbohydrates. GROningen MOlecular Simulation (GROMOS) is another
versatile force field that comes as part of the GROMOS software suite.
This tool is a versatile package to perform molecular mechanics and
dynamics simulations. It is primarily designed for biological molecules
like proteins, lipids and nucleic acids that have a lot of complicated bonded
interactions. The package includes a fully automated topology builder
for proteins, even the multimeric structures of proteins.
CVFF (Consistent Valence Force Field) is also used broadly for small
molecules and macromolecules.
CFF (Consistent Force-Field) is a family of force-fields adapted to
a broad variety of organic compounds, and it includes force fields for
polymers, metals, etc.
Optimized Potential for Liquid Simulations (OPLS) force-field is
optimized to fit experimental properties of liquids, such as density and
heat of vaporization, in addition to fitting gas-phase torsional profiles.
Periodic Boundary Conditions
The proper treatment of boundaries and boundary effects is crucial to
simulation methods as it enables ‘macroscopic’ properties to be calculated
from simulations using relatively small numbers of particles. Periodic
boundary conditions (PBC) are particularly useful for simulating a part
4.18 Introduction to Bioinformatics

of a bulk system with no surfaces present, unless the behaviour near the
walls (surface) are of interest. The bulk is assumed to be composed of
the primary / unit cell surrounded by its exact replicas and thus forms an
infinite lattice in 3D space. The image cells not only have the same size
and shape as the primary one but also contain particles that are images of
the particles in the primary cell. Moreover the cells are space filling, i.e.,
separated by open boundaries so particles can freely enter or leave any
cell in coordination so that total number of particles remain conserved.
The number of image cells needed depends on the range of intermolecular
forces. When the forces are sufficiently short ranged (e.g. in truncated
Lennard-Jones model), only image cells that adjoin the primary cell are
needed (minimum image convention). For squares in two dimensions,
there are eight adjacent image cells whereas for cubes in three dimensions,
the number of adjacent images is twenty-six. In principle, any cell shape
can be used, provided it fills all of space by translation operations of the
central box in three dimensions. Five shapes satisfy this condition: the
cube (and its close relation, the parallelepiped), the hexagonal prism,
the truncated octahedron, the rhombic dodecahedron and the ‘elongated’
dodecahedron. The truncated octahedron and the rhombic dodecahedron
provide periodic cells that are approximately spherical, thus reduces
the number of solvent molecules needed to be added to solvate the
solute and saves CPU time as well. But it is often sensible to choose a
periodic cell that reflects the underlying geometry of the system like the
hexagonal prism is appropriate for solvated DNA. For some simulations,
it is inappropriate to use standard PBC in all directions. PBC may also
cause difficulties when simulation is performed in homogeneous systems
or systems that are not at equilibrium. For example, when studying the
adsorption of molecules onto a surface, it is clearly inappropriate to use the
usual PBC for motion perpendicular to the surface, although it is applied
to motion parallel to the surface. PBCs are not always used in simulation
in such systems like liquid droplets or van der Waals clusters as these
systems inherently contain a boundary. Moreover, to avoid interference
from non-bonded interactions, according to minimum image convention,
the periodic images must be separated by a distance equal to or larger
than a cut-off distance. In practice, a solute-box distance equal to the
 Molecular Modelling 4.19

cut-off (generally 1 nm) is used. For long-range electrostatic interactions,

lattice sum methods such as Ewald Sum, Particle Mesh Ewald (PME)
and Particle-Particle Particle-Mesh (P3M) algorithms are used.

4.2 Energy Minimisation

Energy minimisation is widely used in molecular modelling and is an
integral part of some techniques like conformational search procedures or
preparation of a system prior to other type of calculations like Monte Carlo
or molecular dynamics simulation in order to relieve any unfavourable
interactions in the initial configuration.
Potential energy of all except very simplest system is a complicated,
multidimensional function of the coordinates. For a system with N atoms
the energy is a function of 3N -6 internal or 3N Cartesian coordinates.
It is therefore impossible to visualize the entire energy surface except
for some simple cases where the energy is a function of just one or two
coordinates. In biomolecular modelling, minimum points on the energy
surface would correspond to stable states of the system. There might be a
very large number of minima on the energy surface. The minimum with
the very lowest energy is known as the global energy minimum. Other
minima are called local energy minima. To achieve the geometry of a
conformation with its minimum points of energy surface, minimisation
algorithm is applied. The highest point on the pathway between two
minima, the saddle point, and the transition structures are of special
interest. All these minima and saddle points are stationary points on the
energy surface, where the first derivative of the energy function is zero
and second derivative is positive with respect to all the coordinates.
Molecular mechanics minimisations are always performed by Cartesian
coordinates, where the energy is a function of 3N variables.
Most of the minimisation methods gradually change the coordinates
of the atoms as they move the system closer and closer to the minimum
on the potential energy surface of the molecule. In all these methods, a
numerical test is applied to the new geometry to decide if a minimum is
reached. For example, the slope may be tested to see whether it is zero
within some numerical tolerance or not. If the criterion is not met, then
4.20 Introduction to Bioinformatics

the formula is applied again to make another change in the geometry, until
the minimum is reached. In practice, it is rarely possible to identify the
exact location of minima and/or saddle point. Minimisations are usually
converged by tracking,
(a) energy difference between successive steps, or
(b) root mean square deviation (RMSD) between successive
configurations, or
(c) calculating root mean square gradients of the energy with respect
to the coordinates.
Derivative minimization algorithms: In this method, it is obvious
to calculate the derivatives of the energy wih respect to the variables,
i.e., the cartesian or internal coordinates. It is more useful as it provides
information about the shape of the energy surface and if used properly
it could significantly enhance the efficiency with which the minima are
located. The direction of the first derivative of the energy, the gradient,
indicates where the minimum lies, and the magnitude of the gradient
indicates the steepness of the local slope. The energy of the system can
be lowered by moving each atom in response to the force the acting on it.
The second derivatives indicate the curvature of the function, information
that can be used to predict where the function will change direction, i.e.
pass through a minimum or some other stationary point.
First-order derivative methods use the first derivatives, i.e., the
gradients, whereas second-order methods use both first and second
derivatives. The Newton-Raphson method is the simplest second-order
method.
First–order minimization methods: Two most frquently used
first-order minimisation algorithms are steepest descent and conjugate
gradients methods.
Steepest Descent (SD): Steepest descent method moves in the way
in which the geometry is first minimized in the direction parallel to
the net force, i.e., opposite to the direction in which the gradient is the
largest or, steepest at the initial point, hence its name is steepest. Once
a minimum in the first direction is reached, a second minimization is
 Molecular Modelling 4.21

carried out starting from that point and moving in the steepest remaining
direction. This process continues until a minimum has been reached in
all directions within a sufficient tolerance. The direction of the gradient
is determined by the largest inter-atomic forces. So SD is a good method
for relieving the highest-energy features in an initial configuration. The
method is generally robust even when the starting point is far from a
minimum, where the harmonic approximation to the energy surface is
often a poor assumption. However, it is known for slow progress near
minima. In SD method both the gradients and the direction of successive
steps are orthogonal and show oscillatory behaviour.
Conjugate gradient method (CG): Nonlinear conjugate gradient
methods form another popular type of energy minimization scheme for
large-scale problems where memory and computational performance are
important. In CG method, the first portion of the search takes place in
the direction parallel to the net force, just as in SD. However, to avoid
oscillatory behavior of SD as it moves toward the narrow minima, the
CG method uses the information from the previous direction in the next
search. This allows the method to move rapidly to the minimum in fewer
steps. In CG, the gradients at each point are orthogonal but the directions
are conjugate and hence the name. The line search method is usually
used, although an arbitrary step method is also possible. CG method is
usually attributed as the best choice for general use, even though time per
iteration is longer compared to steepest descents for complex systems.

The ideal minimization algorithm

There are many factors that must be taken into account when choosing the
most appropriate minimisation algorithm (or combination of algorithms)
for a given problem. The ideal minimisation algorithm is the one that
provides the answer as quickly as possible, using the least amount of
memory. Availability of analytical derivatives and the robustness of the
method are also required to be taken into account. No single minimisation
method has yet proved to be the best for all molecular modelling problems.
In particular, a method that works well with quantum mechanics (for fewer
atoms) may not be the most suitable for use with molecular mechanics due
to size of the system under consideration and differential computational
4.22 Introduction to Bioinformatics

requirement for calculating energy in both methods. For bio-molecular

systems, SD and CG methods are mostly used. Generally, SD performs
better than CG when the starting structure is far from the minimum.
However, CG takes over once the initial strain has been removed. In case
of small molecules Newton-Raphson method might be used after initial
minimisation by a more robust method.

Long range Electrostatic Calculations

Long-range interactions (that decay slower than r-n, where n is the
dimensionality of the system), form a complex part of molecular
simulations, as their range is often greater than half the box length. The
charge-charge interaction, which decays as r-1, is particularly problematic
like in case of charged species, molten salts and during calculation of
such properties like as dielectric constant. Moreover, an increase in the
cut-off distance as well in the box size within which all calculations are
required to be performed will increase computational costs unnecessarily.
A variety of methods are developed to deal with long-range electrostatics:
Ewald summation and related methods, reaction-field and image charge
method, cell multipole methods during ionization, and minimization,
equilibration and production phases of simulation.
Ewald summation method: Ewald summation, named after Paul
Peter Ewald, was developed to study the energetics of ionic crystals,
where a particle interacts with all the other particles in the simulation
box and with all of their images in an infinite array of periodic cells.
It has been extensively used in simulations involving highly charged
systems (such as ionic melts, solid state materials) and where electrostatic
effects are important (lipid bilayers, proteins and DNA). The Ewald
summation method is computaionally quite expensive to implement. A
popular approach is the particle-mesh method of Hockney and Eastwood,
which uses the nearest 27 points in three dimensions. From this grided
charge density it is possible to calculate (through Fast Fourier Transform
algorithm) the potential due to the Gaussian distributions at the grid points,
which by interpolation gives rise to the desired potential at each of the
particles. A number of variants of Ewald summation was later developed
and all of which use Fast Fourier Transform algorithm but differ in other
 Molecular Modelling 4.23

aspects of implementation. These are particle-mesh Ewald method and

particle-particle-particle-mesh approach.
An early application of the particle-mesh Ewald (PME) method was
the MD simulation of a crystal of the bovine pancreatic trypsin inhibitor.
Ewald summation were in close agreement with those derived from the
crystallographic temperature factors unlike the non-Ewald mehod. The
highly charged nature of DNA simulation using PME are often much more
stable and trajectories remain closer to the experimental structures. PME
algorithm is almost linear as it scales as N log(N), which is substantially
faster and more accurate than the ordinary Ewald summation on medium
to large systems. The method is based on interpolation of the reciprocal
space Ewald sums and evaluation of the resulting convolutions using Fast
Fourier Transform algorithms. Timings and accuracies are presented for
three large crystalline ionic systems. However on very small systems,
the simple Ewald method gets a better preference to avoid the overhead
in setting up grids and transforms.

Molecular dynamics with continuous potential

The first simulation using continuous potential was performed by
Rahman, with Argon in 1964 and molecular liquid (water) in 1971.
Under the influence of continuous potential the motions of all the
particles are coupled together, giving rise to a many-body problem that
cannot be solved analytically. However, for biological molecules, mainly
constrained dynamics is performed.

Constrained dynamics
The conformational behavior of a flexible molecule is usually a complex
superposition of different motions. The high frequency motions (e.g. bond
vibrations) are usually of less interest than the lower frequency modes,
which often correspond to major conformational changes. Unfortunately,
the time step of a molecular dynamics simulation is dictated by the
highest frequency motion present in the system. It would therefore
be of considerable benefit to be able to increase the time step without
prejudicing the accuracy of the simulation. Constraint dynamics enables
individual internal coordinates or combinations of specified coordinates
4.24 Introduction to Bioinformatics

to be constrained, or ‘fixed’ during the simulation without affecting the

other internal degrees of freedom. A constraint is a requirement that the
system is forced to satisfy, i.e. bonds or angles are forced to adopt specific
values throughout a simulation.
Shake: The most widely used algorithm for performing constrained
dynamics for large molecules is SHAKE. SHAKE is an iterative method;
it resets bonds and angles to prescribed values by moving the bonded
particles parallel to the old bond directions. Sequentially SHAKE changes
a set of unconstrained coordinates, all bonds, to a set of coordinates that
fulfill a list of distance constraints, using a set of references. As the bonds
are coupled, this procedure has to be repeated until the desired accuracy is
reached. The number of iterations can be decreased using over-relaxation.
SHAKE is simple and numerically stable because it resets all constraints
within a prescribed tolerance. SHAKE exhibits the problem with large
displacements – it fails to provide any solution as the coupled bonds are
handled sequentially. Thus, correcting one bond may tilt a coupled bond
so far that the method does not converge. Due to its iterative nature it is
difficult to parallelize.
Lincs: LINear Constraint Solver (LINCS) is an algorithm that resets
bonds to their correct lengths after an unconstrained update. The method is
non-iterative, as it always uses two steps. In the first step, the projections
of the new bonds on the old bonds are set to zero. In the second step,
a correction is applied for the lengthening of the bonds due to rotation.
The algorithm is inherently stable, as the constraints themselves are
reset instead of derivatives of the constraints. In this way, the algorithm
eliminates drift. The constraint problem reduces to a linear matrix
equation as the second derivatives of the constraint equations are set to
zero. However, in a finite discretization scheme corrections are necessary
to achieve accuracy and stability. LINCS does this by implementing
an efficient solver for the matrix equation, a velocity correction that
prevents rotational lengthening and a length correction that improves
accuracy and stability. Although the derivation of the algorithm is based
on matrices, no matrix-matrix multiplications are needed and only the
nonzero matrix elements have to be stored, making the method useful for
 Molecular Modelling 4.25

very large molecules. LINCS has broader convergence conditions; the

sparse constrain coupling matrix is inverted using a power series, which
converges rapidly because the constraining influence is relatively local;
therefore, parallelisation of the algorithm is straightforward. The latter
is of major importance for simulations of large molecules on parallel
computers. Moreover, the LINCS algorithm is three to four times faster
than the SHAKE algorithm, at the same accuracy. Thus use of LINCS
enables time step to be guided by physical conditions only, independent
of constrained algorithm.

Overviews of biological modelling techniques:

(a) Homology Modelling

Homology modelling, also known as comparative modelling of protein, is

the method to build an atomic-resolution model of a protein from its amino
acid sequence on the basis of an experimental three-dimensional structure
of a related homologous protein. The protein for which the model is being
built is called the target protein. On the other hand, the other homologous
protein on which the model of the target protein is built is called the
template. Homology modelling technique relies on the identification of
one or more known protein structures which are likely to resemble the
structure of the target sequence. The amino acid sequence of the target
protein is used as a query sequence to search structural databases in order
to find the structures of suitable homologous proteins having sufficient
sequence identities with the amino acid sequence of the target protein.
The underlying principle of the method is based on the assumptions that
protein structures are more conserved than protein sequences. However,
the typical cut-off value for sequence identity to reliably predict the
structures of proteins from the sequence alignments of homologous
proteins is 20%. Evolutionarily related proteins have somewhat similar
sequences and naturally occurring homologous proteins have similar
protein structures. It has been observed that three-dimensional protein
structure is evolutionarily more conserved than would be expected on the
basis of sequence conservation alone. The sequence alignment between
the target and template protein sequences and template structure are
4.26 Introduction to Bioinformatics

then used to produce a structural model of the target. Because protein

structures are more conserved than protein sequences, detectable levels
of sequence similarity usually imply significant structural similarity
between the proteins.
The quality of the homology model is dependent on the accuracy
of the sequence alignment and the quality of the template structure.
However, the homology modelling techniques can be complicated by the
presence of alignment gaps (commonly called INDELS) that indicate a
structural region present in the target but not in the template. Sometimes
the template structures also have missing amino acid residues arising
out of poor resolution in the experimental procedure (usually X-ray
crystallography) used to solve the structure. The quality of a homology
model declines with decreasing sequence identity. A typical good quality
homology model built from the alignment of a template and a target
protein sequences sharing 70% sequence identity should have ~1–2 Å
root mean square deviation (RMSD) between the matched Cα atoms.
However, with decreasing sequence identities (below 25% level) the
RMSD is only 2–4 Å . However, the errors are significantly higher in the
loop regions, where the amino acid sequences of the target and template
proteins may be completely different.
Regions of the model that are constructed without a template are
generally much less accurate than the rest of the model; the technique
is called loop modelling. Errors in side chain packing and position also
increase with decreasing sequence identities between the target and the
template. Variations in these packing configurations have been suggested
as a major reason for poor model quality at low identity. Taken together,
these various atomic-position errors are significant and limit the use of
homology models for purposes that require atomic level resolution of
structural data, such as drug design and predictions of protein–protein
interaction. It is also known that even the quaternary structure of a protein
may be difficult to predict from homology models of its subunit(s).
Nevertheless, homology models can be useful in reaching qualitative
conclusions about the molecular biochemistry of the query sequence.
 Molecular Modelling 4.27

This technique is especially employed in formulating hypotheses

about why certain residues are conserved, which may in turn lead to
wet-lab experiments to test those hypotheses. For example, the spatial
arrangement of conserved residues may suggest whether a particular
residue is conserved to stabilize the folding, to participate in binding
of some small molecule, or to foster association with another protein or
nucleic acid.
Homology modelling can produce high-quality structural models
when the amino acid sequences of the target and template are closely
related. This has inspired the formation of a structural genomics
consortium dedicated to the production of corresponding representative
experimental structures for all classes of protein folds. The method of
homology modelling is based on the observation that the tertiary structure
of a protein is better conserved than the primary structure i.e., the amino
acid sequence. Thus, even proteins that have diverged appreciably in
sequence but still share detectable sequence similarities will also share
common structural properties; particularly the overall fold of the protein
will be similar. Because it is difficult and time-consuming to obtain
experimental structures from methods such as X-ray crystallography
and protein NMR for every protein of interest, homology modelling
can provide useful structural insights for generating hypotheses about
a protein’s function and directing further experimental work. In other
words, two proteins may share a similar fold even if their evolutionary
relationship is so distant that it cannot be discerned reliably.
Steps in model building

The homology modelling technique can be broken down into four

sequential steps:
1. template selection
2. target-template alignment
3. model construction
4. model assessment.
4.28 Introduction to Bioinformatics

The first two steps as mentioned above are often essentially performed
together. This is because the most common methods of identifying
templates rely on the production of sequence alignments. The key to a
good model quality is to have a proper sequence alignment.
The step of model construction can be done in several different
ways. Given a template and a sequence alignment between the target
and the template, the information contained therein is used to generate a
three-dimensional structural model of the target, represented as a set of
Cartesian coordinates for each atom in the protein. Three major classes
of model generation methods have been proposed:
(a) Fragment assembly: This is the original method of homology
modelling. This method relied on the assembly of a complete
model from conserved structural fragments identified in closely
related solved structures. As for an example, a modelling study of
serine proteases in mammals identified a sharp distinction between
“core” structural regions conserved in all experimental structures
in the class, and variable regions typically located in the loops
where the majority of the sequence differences were localized.
Thus, the target proteins could be modeled by first constructing
the conserved core and then substituting variable regions from
other proteins in the set of solved structures. However, the up-to-
date implementations of this method differ mainly in the way they
deal with regions that are not conserved or that lack a template.
The variable regions, i.e, the loop regions are often constructed
with the help of fragment libraries.
(b) Segment matching: This method divides the entire target protein
into short segments of amino acid residues and each segment is
matched with different structural templates. The selection of the
template is done with sequence similarity, comparison of alpha
carbon coordinates and steric clashes between the side chain
atoms.
(c) Satisfaction of spatial restraints: The most commonly used
technique of homology modelling is by the satisfaction of
spatial restraints. This method of homology modelling takes
 Molecular Modelling 4.29

its inspiration from calculations required to construct a three-

dimensional structure from data generated by NMR spectroscopy.
In this method, one or more target-template alignments are used
to construct a set of geometrical criteria that are then converted
to probability density functions (pdfs) for each restraint. A more
recent expansion of the method applies the spatial-restraint model
to electron density maps derived from cryo-electron microscopy
studies, which provide low-resolution information that is not
usually itself sufficient to generate atomic-resolution of structural
models.
The assessment of the homology models is an important task. This is
generally done in two different ways; the statistical potentials and physics
based energy calculations. Both these methods produce energy values of
the modeled proteins.
Statistical potentials are empirical methods based on observed residue-
residue contact frequencies among proteins of known structures in the
Protein Data Bank (PDB)—the structural database for macromolecules.
They assign a probability or energy score to each possible pair wise
interactions between amino acids and combine these pair wise interaction
scores into a single score for the entire homology model.
Physics-based energy calculations generally aim to capture the inter-
atomic interactions that are physically responsible for protein stability
in solution, especially the van der Waals and electrostatic interactions.
These calculations are performed using a molecular mechanics force field.
Such physics based method is based on the energy landscape hypothesis
of protein folding, which predicts that a protein’s native state is also its
energy minimum.
The accuracies of the structures obtained by homology modelling are
highly dependent on the sequence identity between target and template.
Above 50% sequence identity, the homology models tend to be reliable,
with only minor errors in side chain packing and rotameric state. The
overall RMSD between the modeled and the experimental structure are
found to be around 1Å. This error is comparable to the typical resolution
of a structure solved by Nuclear Magnetic Resonance (NMR) techniques.
4.30 Introduction to Bioinformatics

When the sequence identity is in the range of 30–50%, errors can be more
severe and are often located in loops. Below 30% sequence identity,
serious errors occur which sometimes result in the basic fold being mis-
predicted. This low-identity region is often referred to as the “twilight
zone” within which homology modelling is extremely difficult. In such
cases homology modelling methods are possibly less suited than fold
recognition.
The other less popular molecular modelling techniques are Threading
and Ab Initio methods.
(b) Threading

Protein threading, also known as fold recognition, is a method of

modelling protein structures. The method is used to model those proteins
which have the same fold as proteins of known structures, but do not
have homologous proteins with known structure. The threading method
differs from the homology modelling method of structure prediction
as the threading method is used for proteins which do not have their
homologous protein structures deposited in the PDB, whereas homology
modelling is used for those proteins which have homologous protein
structures present in PDB. Threading works by using statistical knowledge
of the relationship between the structures deposited in the PDB and the
sequence of the target protein, the one to be modeled. The prediction
is made by “threading” (i.e. placing, aligning) each amino acid in the
target sequence to a position in the template structure, and progressively
evaluating how well the target fits the template. The best-fit template is
then selected and then the structural model of the sequence is built based
on the alignment with the chosen template. Protein threading is based
on two basic observations: that the total number of different folds in
nature is fairly small (approximately 1300); and that nearly 90% of the
new structures submitted to the PDB in the past three years have similar
structural folds to the ones already in the PDB.
Comparison with homology modelling

Homology modelling and protein threading are both template-based

methods to predict the structures of proteins from their amino acid
 Molecular Modelling 4.31

sequences. There is practically no rigorous boundary between them in

terms of prediction techniques. However, homology modelling is for those
protein targets which have homologous proteins with known structure.
On the other hand, protein threading is for those protein targets with
only fold-level homology found. In a nutshell, homology modelling is
for “easier” targets and protein threading is for comparatively “harder”
targets.
(c) Ab initio method

In ab initio methods, an initial effort is made to analyze the secondary

structures (alpha helix, beta sheet, beta turn, etc.) from primary structure
of a protein. This is done by utilization of physicochemical parameters
and neural network algorithms. From that point, algorithms predict the
tertiary folding in a protein. One drawback to this strategy is that it is
not yet capable of incorporating the locations and orientations of side
chains in amino acid.
The next important molecular modelling technique is the molecular
docking. The molecular docking is one of the most widely used
techniques. This technique is required to generate models of protein and
protein ligand complexes.
(d) Molecular Docking

In the field of molecular modelling, docking is the method which predicts

the preferred orientation of one molecule bound to a second molecule
in a complex. Knowledge of the preferred orientation in turn may
therefore be used to predict the strength of association or binding affinity
between the two molecules using a scoring function. The associations
between biologically relevant molecules such as proteins, nucleic acids,
carbohydrates, and lipids play very vital roles in signal transduction.
Furthermore, the relative orientation of the two interacting partners
(protein with another protein or a ligand) may affect the type of signal
produced. Therefore, docking is useful for predicting both the strength and
type of signal produced. Docking is frequently used to predict the binding
orientation of small molecule drug candidates to their protein targets in
order to predict the affinity and activity of the small molecule. Hence
4.32 Introduction to Bioinformatics

docking plays an important role in the rational drug design endeavors.

Given the biological and pharmaceutical significance of molecular
docking, considerable efforts have been directed towards improving the
methods to predict the structure of the complex obtained by docking.
As the subject is still very young, researches are going on to come
up with many other plausible ways of molecular techniques to build
the structures of proteins and protein ligand complexes. In the present
scenario, the present aim of the chapter is to give a fair idea about the
basics of the molecular modelling principles as utilized in the field of
protein-protein and protein ligand complexes. There are very few articles
available that deal with the basics of such techniques. In the present
chapter emphasis is given on the elucidation of the existing knowledge
about the molecular modelling techniques. The chapter is intended for
both the computer scientists and molecular biologists. The basics of
molecular biology are discussed. The protein structure is presented and
these are linked to the theory of computations. The chapter may be used
as a first hand guide for the researchers interested in the field of Molecular
Modelling and Drug Design.

Prediction of protein-protein and protein-ligand interactions

The protein-protein and protein-ligand interaction prediction
methodologies can be broadly classified into two categories, viz., the
experimental determination techniques and the computational. In most of
the cases the two types of methods are combined together to complement
each other.

4.3 EXPERIMENTAL METHODOLOGIES

The prediction of protein-protein and protein-ligand interactions requires

the determination of the quaternary structures of the proteins. This needs
the background knowledge of the subunit composition of the system under
investigation. The subunit composition of the proteins may be determined
by applying chemical methods like introducing chemical cross-links
between the polypeptide chains. This may also be done by comparing
the molecular masses of the native protein and its constituent chains.
The molecular masses of the subunits are obtained using denaturing gel
electrophoresis.
 Molecular Modelling 4.33

The most accurate and important method of prediction of structures

of protein-protein and protein-ligand complexes is X-ray crystallography.
However, there are several other experimental techniques for the purpose,
viz., NMR spectroscopy, fluorescence resonance energy transfer, yeast
two-hybrid, affinity purification/mass spectrometry, protein chips to
name a few.

X-ray crystallography

X-ray crystallography is a method to determine the arrangement of

atoms within a crystal. In this method, a beam of X-rays are allowed
to strike a crystal and then it scatters into many different directions.
The electron density map of the molecule can be generated from the
angles and intensities of these scattered X-ray beams to build a three-
dimensional picture of the distribution of electrons within the crystal.
This leads to the determinations of the mean positions of the atoms in
the crystal, as well as, their chemical bonds with which the atoms are
held together and the disorder and various other structural properties.
A fully grown crystal is mounted on the apparatus called goniometer.
After that the crystal is rotated gradually and X-rays are passed through
it, which produce a diffraction pattern of spots regularly spaced in two
dimensions. These are known as reflections. A three-dimensional electron
density model of the whole crystal is then created from the images of the
crystal which are taken by rotating it at different orientations with the
help of Fourier transforms, as well as with previous chemical data for
the crystallized sample. Small crystals or deformities in crystal packing
lead to erroneous results. X-ray crystallography is related to several other
methods for determining atomic structures. Similar diffraction patterns
can be produced by scattering electrons or neutrons, which are likewise
interpreted as a Fourier transform.
The structure of Hexamethylenetetramine was solved in 1923, and
this happened to be the structure of the first organic molecule to be solved
by X-ray crystallography. After that, structures of a number of important
bioorganic molecules like, porphyrin, corrin and chlorophyll were solved.
4.34 Introduction to Bioinformatics

X-ray crystallography of biological molecules took off with Dorothy

Crowfoot Hodgkin, who solved the structures of cholesterol (1937),
vitamin B12 (1945) and penicillin (1954). In 1969, she succeeded in
solving the structure of insulin.
X-crystallography has a widespread use in the elucidation of protein
structure function relationship, mutational analysis and drug design. The
challenge lies in the prediction of structures of membrane proteins, like ion
channels and receptors as it is difficult to find appropriate system for them
to crystallize as these proteins are integral parts of the cell membranes
and it is hard to get the protein part out from the membrane component.
Nuclear Magnetic Resonance (NMR) Techniques

Protein nuclear magnetic resonance spectroscopy (usually abbreviated as

protein NMR) is a field of structural biology in which NMR spectroscopy
is used to obtain information about the structure and dynamics of protein-
protein and protein-ligand complexes. Structure determination by NMR
spectroscopy usually consists of following phases, each using a separate
set of highly specialized techniques. The sample is prepared; resonances
are assigned; restraints are generated and a structure is calculated and
validated.
FÖRSTER Resonance Energy Transfer / Fluorescence Resonance Energy
Transfer (FRET)

It is a mechanism describing energy transfer between two molecules both

of which should be sensitive to light. This method is used to study protein
dynamics, protein-protein and protein-ligand interactions. For FRET
analysis of protein interactions, the cyan fluorescent protein (CFP)-yellow
fluorescent protein (YFP) pair, which is the colour variants of the green
fluorescent protein (GFP), is currently the most useful protein pair that
is being employed in biology. The interactions between the proteins are
determined by the amount of energy that is being transferred between
the proteins, thereby creating a large emission peak of YFP obtained by
the overlaps of the individual fluorescent emission peaks of CFP and YPF
as the two proteins are near to each other.
 Molecular Modelling 4.35

Yeast Two-Hybrid System

One of the important methods to analyze the physical interactions

between proteins or proteins with ligand is the yeast two-hybrid
screening. The basic principle behind the process comes from the
fact that the close proximity and modularity of the activating and the
binding domains of most of the eukaryotic transcription factors lead to
the interactions between themselves albeit indirectly. This system often
utilizes a genetically engineered strain of yeast which does not possess
the biosynthetic machinery required for the biosynthesis of amino acids
or nucleic acids. Yeast cells do not survive on the media lacking these
nutrients. In order to detect the interactions between the proteins, the
transcription factor is divided into two domains called the binding (BD)
and the activator domain (AD). Genetically engineered plasmids are made
to produce a protein product with the DNA binding domain (BD) attached
onto a protein. Another such plasmid codes for a protein product having
the activation domain (AD) tagged to another protein. The protein fused
to the BD may be referred to as the bait protein and is typically a known
protein that is used to identify the new binding partners. The protein
fused to the AD may be referred to as the prey protein and can either be a
single known protein or a collection of known or unknown proteins. The
transcription of the reporter gene(s) occurs if and only if the AD and the
BD of the transcription factors are connected bringing the AD close to the
transcription start site of the reporter gene, which justifies the presence
of physical interactions between the bait and the prey proteins. Thus, a
fruitful interaction between the proteins fused together determines the
phenotypic change of the cell.
Affinity Purification

It is a technique to study protein-protein and protein-ligand interactions.

It involves creating a fusion protein with a designed piece, the tag, on
the end. The protein of interest with the tag first binds to beads coated
with Immunoglobulin G. Then the tag is broken apart by an enzyme.
Finally a different part of the tag binds reversibly to beads of a different
type. After the protein of interest has been washed through two affinity
columns, it can be examined for binding partners.
4.36 Introduction to Bioinformatics

Protein Chips / Protein Microarray

This method is sometimes referred to as a protein binding microarray.

It provides a multiplex approach to identify protein-protein and protein-
ligand interactions. This method is mainly applied to identify transcription
factor protein-activation, or to identify the targets of biologically active
small molecules. On a piece of glass, different protein molecules or DNA
binding sequences of proteins are affixed orderly at different locations
forming a microscopic array. A commonly used microarray is obtained
by affixing antibodies which bind antigen molecules from cell lysate
solutions. These antibodies can easily be spotted with appropriate dyes.
The aforementioned experimental tools are routinely used in
laboratories to detect protein-protein and protein-ligand interactions.
However, these methods are not full proof. On top of that, these
experimental tools are labor-intensive and time-consuming. The methods
are expensive and often they include false positives. These factors
necessitate the use of other methods in order to verify the results. All
these led to the development of computational methods which are capable
of prediction of protein-protein and protein-ligand interactions with
sufficient accuracies.
Computational Predictions of Protein-Protein and Protein-Ligand
Interactions

Computational methodologies to predict protein-protein and protein-

ligand interactions can broadly be classified as numerical value-based
and probabilistic. Both of them involve training over a dataset containing
protein structural and sequence information.
Numerical methods use a function of the form F = f (pi, pj ∈ ni, x)
Where, pi = input data for the amino acid residue i under consideration.
pj = the corresponding properties of the spatially neighboring residues
and j ∈ ni
x = the collection of coefficients to be determined by training
The value of F determines the characteristics of residue under
consideration. It can either be I for interface or be N for non-interface.
 Molecular Modelling 4.37

If the value of F is above a certain threshold i it is considered to be in I

state otherwise in N state.
The value-based methods are classified as
• Linear regression: This method is used to predict the values of the
unknown variables from a set of known variables. It also tests the
relationship among the variables. In case of predictions of protein-
protein and protein-ligand interactions solvent accessibilities of
the amino acid residues of the proteins are taken as inputs. The
different amino acid residues have different solvent accessibility
values depending on whether they are exposed or buried. Based
on a collection of such values of known proteins linear regression
methods may be used to predict the nature of amino acids in
unknown proteins.
• Scoring function: This is a general knowledge based approach.
Scoring functions are based on empirical energy functions made
up of contributions from various experimentally validated data.
In this approach, information are generated from know protein-
protein and protein-ligand complexes. The approach takes into
account various types of physico-chemical parameters of the
protein complexes, like solvation potential, solvent accessibilities,
interaction free energies and entropies for example. These data are
used to generate the scoring functions for the individual atoms of
the amino acids constituting the protein complexes. The functions
can then be used in case of unknown proteins to predict its mode
of binding.
• Support Vector Machines (SVMs): This is a supervised
learning method for classification, function approximation, signal
processing and regression analysis. In this method, the input data
are divided into two different sets, viz., the interface (I) and non-
interface (N) sets. The SVM will create a separating hyper-plane
which maximizes the margin between the two different types of
the data. The basic principle of SVM is first to train it with a set
of known data to create a classifier. The classifier is then used to
predict whether an amino acid residue is in I or N state by giving
4.38 Introduction to Bioinformatics

it a probability score. SVM training is done by using a training file

consisting of feature vectors generated using various information
about the protein-protein and protein-ligand complexes. The
typical information used for the purpose are hydrophobicities,
accessible surface areas, electrical charges, sequence similarity
and sequence conservation scores of amino acids etc. These
information are combined into feature vectors and are used as
input to train the SVM.
• Neural network: This is an interconnected group artificial
neurons (In biological perspective, neurons are connectors that
transmit information via chemical signaling between cells), which
are basically an adaptive system. There are hidden layers whose
output is fed into a final output node. Protein information like
the solvent accessibilities, free energy of interactions etc. are fed
into the hidden layers to train them. Next the trained model can
be used to produce the output from a set of unknown data.
• Random Forests: This is a combination of tree predictors. Each of
the tree is dependent on the respective values of a random vector
that is sampled independently following the same distribution
for all the trees in the whole forest generated by the method. The
more the number of trees the less is the error. Due to the law of
Large Numbers there are no over-fitting problems.
The other class of computational prediction methodology is
the use of probabilistic methods. The probabilistic methods
are employed to find the conditional probability p(s| x1,.,xk ),
where s is either I or N for the range of input data x1,.,xk for a
residue under consideration. Interface is predicted if the value of
p(s| x1,.,xk) becomes greater than a threshold value.
• Naïve Bayesian: This is a supervised machine learning method.
It takes input data, which are assumed to be independent. Naïve
Bayes is a well-known machine learning tool. This is a classifier
that assumes no dependencies between the variables to predict
the class of the object under study.
 Molecular Modelling 4.39

• Bayesian network: In this case, the input data are dependent on

each other. So a joint probability is calculated. For two dependent
input data, x1 and x2, their joint probability P(S|x1,x2) is calculated
as P(x1, x2|S).
P(S) = fraction of state S in the training dataset. In case of protein-

protein interactions S represents whether an amino acid residue
of the protein under study is in the interface or not.
• Hidden Markov Model (HMM): Hidden Markov Models
(HMMs) belong to the class of directed graphical models.
HMMs define a factored probability distribution of p(x,y). The
mathematical formulation of the model is expressed as
p(x,y) = ∏ p(xi|yi)p(yi|yi–1)
This model is often referred to as a generative model. The term
p(xi|yi) is considered to be the probability that the observed
result xi is generated from the input feature yi. The second term,
p(yi | yi–1) is the first-order Markov assumption term. This term
represents that the probability of a label variable yi is not related
to the other label variables yi-1 which are used in the study.
In case of prediction of protein-protein and protein-ligand
interactions this method takes into account a multiple sequence
alignment (MSA) of known proteins. The amino acid residues
which are found to be conserved in the MSA are used to construct
a sequence profile which is used to predict the nature of the amino
acids from the protein of interest.

4.4 SOLUTIONS AND RECOMMENDATIONS

One of the major advantages of the use of molecular modelling techniques

in solving structural problems in biology is the speed of calculations. A
user can run several modelling jobs in parallel and is therefore able to
perform with much higher speed. There are several experimental methods
(like Yeast-two-hybrid, X-Ray Crystallography and Nuclear Magnetic
Resonance spectroscopy) available for the analysis of protein interactions.
4.40 Introduction to Bioinformatics

Side by side a number of computational algorithms have been developed

to predict amino acid residues in protein-protein interfaces with accuracies
typically in the range of 70%. There is no denying that the experimental
approaches are more accurate and would give more comprehensive
and reliable results; but they are time consuming and labour intensive
besides being expensive. As alternatives to the experimental approaches,
computational methods have been developed. They are comparatively
less accurate but often give an overall idea of the whole process. There
are various computational approaches with somewhat varying degrees of
accuracies. The most important aspect is that computational approaches
are cost effective, and requires less time. A word of caution is that none
of the methods are cent percent accurate. To properly predict a PPI a
number of information is needed. The best way to perform an experiment
is first to use the computational algorithms to find a PPI and then test that
via experimental means.
This is a list of docking servers and software tools. This is given for
an easy reference of the computational tools available to study protein
interactions.
http://protein3d.ncifcrf.gov/~keskino
http://dockground.bioinformatics.ku.edu/
http://www.ces.clemson.edu/compbio/protcom/
http://mips.gsf.de/proj/ppi/
http://mips.gsf.de/proj/yeast/CYGD/interaction/
http://dip.doe-mbi.ucla.edu/dip/Main.cgi
http://www.thebiogrid.org/
http://www.ncbi.nlm.nih.gov/RefSeq/HIVInteractions/
http://www.hprd.org/
http://wilab.inha.ac.kr/hpid/
http://www.ihop-net.org/UniPub/iHOP
http://insilico.csie.ntu.edu.tw:9999/point/
 Molecular Modelling 4.41

http://point.bioinformatics.tw/
http://www.compbio.dundee.ac.uk/www-pips
http://www.jcvi.org/mpidb/about.php
http://www.molecularconnections.com/home/en/home/products/
NetPro
http://www.proteinlounge.com/inter_home.asp
http://itolab.cb.k.u-tokyo.ac.jp/Y2H/
http://mips.gsf.de/genre/proj/mpact/index.html

SOME IMPORTANT QUESTIONS AND ANSWERS

1. What is Molecular Modelling?
2. What is it good for?
3. What are the goals of using force fields?
4. How are the parameters in the force fields calculated?
5. What are topological properties of molecules?

6. What are structural parameters of molecules?

7. What are other properties of molecules?

8. What are structural properties of molecules?

9. What are electron densities?

10. What is electrostatic potential?

11. What is electrostatic potential map?

12. What is the correct name for the following peptide.

4.42 Introduction to Bioinformatics

(a) L-alanyl-L-phenylalanyl-glycine
(b) glycyl-L-phenylalanyl-L-alanine
(c) L-phenylalanyl-L-alanyl-glycine
(d) L-alanyl-glycyl-L-phenylalanine
13. Identify the single letter code from the following single letter to represent
the structure below

(a) DVYGSA (b) ASGYVD

(c) EVFGSA (d) ASGFVE
14. Identify the true statement about a peptide bond (RCONHR’)?
(a) It is non planar.
(b) It is capable of forming a hydrogen bond.
(c) The cis configuration is favoured over the trans configuration.
(d) Single bond rotation is permitted between nitrogen and the carbonyl
group.
15. Identify the incorrect statement about protein secondary structure?
(a) The alpha helix, beta pleated sheet and beta turns are examples of
protein secondary structure.
(b) The ability of peptide bonds to form intramolecular hydrogen bonds
is important to secondary structure.
(c) The steric influence of amino acid residues is important to secondary
structure.
(d) The hydrophilic/hydrophobic character of amino acid residues is
important to secondary structure.
16. Which of the following terms would refer to the arrangement of different
protein subunits in a multiprotein complex?
 Molecular Modelling 4.43

(a) primary structure

(b) secondary structure
(c) tertiary structure
(d) quaternary structure
17. Which of the following terms would refer to the order in which amino
acids are linked together in a protein?
(a) primary structure
(b) secondary structure
(c) tertiary structure
(d) quaternary structure
18. Which of the following terms would refer to the overall three-
dimensional shape of a protein?
(a) primary structure
(b) secondary structure
(c) tertiary structure
(d) quaternary structure
19. Which of the following terms would refer to ordered region in a protein?
(a) primary structure
(b) secondary structure
(c) tertiary structure
(d) quaternary structure
20. What is the strongest form of intermolecular bonding that could be
formed involving the residue of the amino acid serine.
(a) ionic bond
(b) hydrogen bond
(c) van der Waals interactions
(d) none of the above
4.44 Introduction to Bioinformatics

ANSWERS
1. Molecular Modelling is the science that deals with the description
of the atomic and molecular interactions. These interactions govern
microscopic and macroscopic behaviors of physical systems
2. The utility of molecular modelling is to make connections between
the microscopic and the macroscopic world provided by the theory of
statistical mechanics
3. The basic aim of the force fields is to define the empirical potential
energy functions V(r) to model the molecular interactions. These
potential energy functions need to be differentiable for the computations
of the forces acting on each atom: F= –DV(r)
4. At first the theoretical analytical functional forms of the interactions are
derived. For that purpose, the entire system is divided into a number
of atom types which differ by their atomic number and chemical
environment. Thus, the carbons in C=O or C-C are not of the same type
as they are present in different chemical environments. The bonding
parameters, like the bond enthalpy and bond free energy values, are
then determined so as to reproduce the interactions between the various
atom types by fitting procedures. The experimental enthalpy values
of the covalent bonds are used in CHARMM force field. On the other
hand, GROMOS and AMBER force fields are based on experimental
free energies.
5. The topological properties are the descriptions of the covalent
connectivity of the molecules to be modeled.

6. It is the initial or the starting conformation of a molecule, obtained from

an X-ray structure, NMR data or a theoretical model.

7. These are the different types of forces acting on the molecule based on
the force-field parameters.

8. These are the thermodynamic ensembles corresponding to the

experimental conditions.

9. Electron densities require no prior knowledge about bonding which

is quite different from the traditional methods. Therefore, they can be
used to elucidate bonding. Electron densities also provide information
about the transition states in absence of experimental evidences.
 Molecular Modelling 4.45

10. The electrostatic potential is the interaction energy of a point positive

charge (known as an electrophile) with the electrons of a molecule.
Negative electrostatic potentials would indicate areas that are prone to
electrophilic attack.

11. The electrostatic potential map of a molecule provides the information

about the distribution of charges in the molecule. It also gives
information about the delocalization of electric charges.

12. (a) 13. (b)

14. (b) 15. (d)
16. (d) 17. (a)
18. (c) 19. (b)
20. (b)
 CHAPTER

5 Phylogenetic Analysis

5.1 INTRODUCTION

Phylogenetic analysis deals with the study of evolutionary history of

organisms. It also analyzes the relationships between various species.
The Phylogenetic relationships are drawn by using common heritable
elements like DNA sequences, r-RNA sequences, amino acid sequences
of proteins. The results of these analyses are presented in the form of a
tree known as the Phylogenetic tree. On the other hand, taxonomy which
is generally used synonymously with Phylogenetic study is the branch of
science that deals with the nomenclature, classification and identification
of organisms. The purpose of Phylogenetic study is manifold as follows:
1. To study the evolutionary history: The main aim of a Phylogenetic
study is to reconstruct the evolutionary history of the organism.
The species diversities are also studied using this technique.
2. Another important application of the study is to determine the
closest relative of the organism of interest.
3. With the help of Phylogenetic analyses, the functions of new
genes may be identified or at least predicted.
4. Similarly the origin of the evolved gene of interest may be
determined.
5. Phylogenetic studies based on biological sequence information
could provide valuable information with accurate descriptions of
5.2 Introduction to Bioinformatics

patterns of relatedness between the species before the advent of

modern day molecular sequencing methods.
6. Phylogenetic studies are also useful in forensic sciences.
Phylogenetic analyses are used to assess the DNA samples
submitted in court cases to solve the problems of paternity and
other crimes.
7. Phylogenetic analyses are routinely used to analyze the mode
of pathogen outbreak. The Phylogenetic analyses are capable
of elucidating the characters of new pathogenic species,
their associations with their hosts and further their modes of
transmissions.
8. Another important aspect of Phylogenetic studies is that the study
provides ideas to conservation biologists to identify the nature of
a species. The study also provides tentative ideas regarding the
probabilities of a species to get extinct.

5.2 IMPORTANCE OF MOLECULAR SEQUENCE DATA

The Phylogenetic studies are almost entirely based on molecular sequence

information. The mostly used molecular sequence data are the nucleotide
sequences of DNA, r-RNA and the amino acid sequences of proteins.
The molecular sequence data are used for Phylogenetic studies for the
following reasons:
1. DNA is mainly the genetic material. It contains the genetic
information and it is inherited from the ancestors.
2. Another important molecular sequence data that are mostly used
in Phylogenetic analyses are the r-RNA. r-RNA is present in all
cells. Thus, r-RNA encoding genes are sequenced and the r-RNA
sequence is analyzed to identify taxonomic group of the organism.
From the use of r-RNA sequence data the related species are also
identified and consequently the rates of species divergence can
be estimated.
3. The amino acid sequence of a protein can also be used as the
starting material for Phylogenetic studies. The amino acid
 Phylogenetic Analysis 5.3

sequences of proteins are the fundamental building blocks

responsible for the structure and function of the protein.
4. The most important aspect of the molecular sequences is that the
molecular sequences are highly specific and rich in information.
A word of caution regarding the use of molecular sequence
information for Phylogenetic analyses is that DNA sequences can only
be used if there are high sequence identities between them. Otherwise,
the DNA sequences are to be converted to amino acid sequences for the
analysis purpose.
However, a situation might arise where the molecular sequence data
are not available. For example, when analyzing the ancient fossil samples,
the molecular sequence data might not be available. In such situations,
morphological features may be used for the analysis purposes. It is
redundant to mention that such procedures are less accurate as the same
morphological feature might arise from a large number of independent
evolutionary events. It may further be noted that biochemical properties
can also be used in Phylogenetic studies.
However, an empirical guiding principle for Phylogenetic analyses
is as follows:
(a) In order to construct a Phylogenetic relationship between closely
related species a rapidly evolving molecule, like mitochondrial
DNA, is used.
(b) On the other hand, for finding the Phylogenetic relationship
between diverse species, a highly conserved molecule like
ribosomal RNA (r-RNA) is to be used.
Phylogenetic studies can be analyzed in three major ways. They are
• Phenetics: In this type of Phylogenetic analyses, the different
species are grouped on the basis of their phenotypic resemblances.
In such studies, all characteristics are taken into considerations.
• Cladistics: In this type of Phylogenetic analyses, the different
species are grouped only with those species that share the similar
characteristics.
5.4 Introduction to Bioinformatics

• Evolutionary systematics: In this type of Phylogenetic study, both

the phonetic and cladistics principles are used.
Among the aforementioned methods cladistics is considered to be
the best method for Phylogenetic analyses as it incorporates the latest
evolutionary theory.

5.3 SOME IMPORTANT TERMINOLOGIES AND DEFINITIONS

The Phylogenetic studies often involve the following terms:

A. Phylogenetic tree: It is the representation of evolutionary
relationships between organisms. It is basically a tree like diagram
where the branches of the tree would represent the speciation
events. A Phylogenetic tree can be of different types.
(a) The most commonly used one is the Dendrogram. A
Dendrogram is a broad term which encompasses all the
different types of Phylogenetic trees.
(b) A Phylogram is a branching diagram representing
Phylogenetic relationship. A Phylogram provides an estimate
of a Phylogenetic relationship where the lengths of the tree
are proportional to the extent of inferred evolutionary change.
(c) A Cladogram is a Phylogenetic diagram which is used to
show the relations which is used to reveal the relationship
between the ancestors and the descendants. In a Cladogram
the lines would represent the different organisms coming
from a common ancestor.
(d) A Chronogram is another type of Phylogenetic tree that is
constructed on the basis of evolutionary time scale.
(e) A Spindle Diagram is a special type of Phylogenetic
relationship which would represent the evolutionary history
and the distribution of various taxonomic classes.
(f) Rooted Phylogenetic tree: A rooted Phylogenetic tree is a
directed tree with a unique node which would represent the
most common ancestor of all the species used in the study.
 Phylogenetic Analysis 5.5

(g) Unrooted Phylogenetic tree: An unrooted Phylogenetic tree

would depict the relatedness between the different species
under study without the knowledge of a common ancestor.
(h) Homologs: Studies of protein and gene evolution are known
to involve the comparison of homologous species. The
sequences that have common origins but may or may not
have common activity are considered to be belonging to
homologous species. However, homologous species are
considered to have a common ancestor.
(i) Orthologs: Orthologous species are homologs produced by
speciation. These species are known to contain genes derived
from a common ancestor that diverged due to divergence
of the organisms they are associated with. However, the
orthologous genes would tend to have similar functions.
(j) Paralogs: Paralogous species are homologs produced by
gene duplications. These species are known to contain genes
produced by gene duplications. However, the paralogous
genes would tend to have different functions.
(k) Xenologs: Xenologs are homologous species produced
by horizontal gene transfer between two organisms. The
G+C content are found to be vastly different between the
organisms. However, the xenologous genes tend to have
similar functions.
(l) Last Common Ancestor (LCA): The species from which the
new species are evolved.
(m) Mitochondrial DNA (mt-DNA): The DNA that is available in
the mitochondria. Mt-DNA has a higher mutation rate than
the nuclear DNA and therefore evolves more rapidly than the
nuclear DNA.
(n) Ribosomal RNA (r-RNA): The ribosomal RNA has a specific
secondary structure. It is a slowly evolving molecule.
5.6 Introduction to Bioinformatics

(o) Distance matrix method: This is a basic method which

involves the selection of two of the most closely related
species from a bunch of species used for Phylogenetic study.
These two species are clustered together and the operation is
repeated until only one cluster is left.
(p) Unweighted Pair Group Method using Arithmetic Mean
(UPGMA): In this method, the distance between the different
species is calculated as arithmetic mean as all the species are
given equal statistical weights.
(q) Weighted Pair Group Method using Arithmetic Mean
(WPGMA): In this method, the distance between the different
species is calculated on the basis of the population size of
the cluster to which the species belongs. The cluster with a
higher population is given a larger statistical weightage.
(r) Neighbour Joining (NJ): This is clustering method. This
method is based on the principle of UPGMA technique. In this
method, the closely related species as obtained from multiple
sequence alignment are connected by internal nodes and they
are referred to as Neighbours. The process is repeated until
all the species under consideration are joined together to form
a Phylogenetic tree.
(s) Maximum parsimony method: This method is based on the
principle of finding the mutations that are needed to generate
a sequence from the other. This is generally obtained from
the analyses of the multiple sequence alignment results. The
grouping of the species in the Phylogenetic tree is done by
finding the species having sequences with minimum number
of mutations.
(t) Maximum likelihood method: This method is also based on
the analyses of the results of multiple sequence alignment.
However, the method allows user intervention in selecting
the closely related species which might not be reflected in
the multiple sequence alignment.
 Phylogenetic Analysis 5.7

5.4 SOME STATISTICAL ASPECTS IN PHYLOGENETICS

(i) Akaike information criterion (AIC): It is an estimator obtained

from information theory. It measures the distance between the
true model and the estimated model. The value of the AIC is
derived from a maximum likelihood estimate that is penalized for
the number of estimable parameters in the model. The selection
of the model by AIC is based on a comparison between nested
and non-nested models. The AIC framework also helps in the
assessment of the uncertain parameters in model selection.
(ii) Bayes factors: It is a Bayesian analog to the likelihood ratio test.
In this case, the fitness of the model is derived from integration
over all possible parameter values that are required for fitness
evaluation. The most important aspect of Bayes factors is that
the model selection process does not require an a priori model
selection criterion.
(iii) Bayesian information criterion (BIC): This parameter is similar
to the AIC as it compares transformed likelihoods. However, it
differs by penalizing the sample size and the number of estimable
parameters. The BIC is an approximation of the natural logarithm
of the Bayes factor. This makes this parameter as computationally
more reliable than Bayes factors for Phylogenetic purposes.
However, BIC generally selects less complex models than those
selected by Bayes factors.
(iv) Branch length estimations: It represents the number of estimated
changes in each lineage used in the Phylogenetic analyses. It is
usually calculated as the number of substitutions per site. The
extent of the measurements of the branch lengths depends on the
actual number of substitutions in the sequence. This is generally
reflected as the adequacy of the model.
(v) Heterotachy: It is the property of change in evolutionary rate at
a given sequence position through time. This is very vital as any
uncorrected data may lead to mismatching in Phylogenetic study.
This is a source of generation of systematic error.
5.8 Introduction to Bioinformatics

(vi) Jukes-Cantor model: It is the simplest substitution model that

is used for DNA. There are different assumptions. The model
assumes equal base frequencies and equal mutation rates in the
DNA sequences. The only parameter of this model is the overall
substitution rate. This variable turns out to be a constant when
the mean-rate of base substitutions is normalized to 1.
(vii) Likelihood ratio test (LRT): It is a method to compare the
maximum likelihood estimates of two nested models obtained
using the same one data set. The significance of the models is
assessed by an arbitrarily attained significance value (usually p =
0.05). It is to be noted that the test requires that one of the nested
models being compared is the true model obtained from the data.
(viii) Model averaging: It is an information theoretic technique and it
makes formal inferences based on a set of adequate models. This
parameter is generally used when there are more than one models
obtained for the same dataset with reasonably same accuracies.
Model averaging techniques judge the relative importance of
the individual models by calculating the relative weights. This
technique utilizes the incorporations of uncertainties in model
selection and in the estimation of model parameters.
(ix) ModelTest: It is a widely used computer programing technique
that quickly fits one of the candidate models from a set of built
models to a nucleotide data set using hierarchical likelihood ratio
tests. This technique builds a neighbor-joining Jukes-Cantor tree.
Then the maximized model likelihoods are estimated from the
tree using the program PAUP. ModelTest also comes up with an
AIC score for each model.
(x) Nested models: These are the statistical models which include
all special cases of the most general model in the candidate set.
(xi) Phylogeny estimation: There are various approaches for
inferring evolutionary relationships among living organisms (i.e.
a phylogeny). All the methods for molecular data depend upon
 Phylogenetic Analysis 5.9

an implicit or explicit mathematical model which describes the

evolution of aligned nucleotide or amino acid sequence characters.
Thus, a Phylogenetic tree is referred to as an estimation of the
true phylogeny of a group of organisms.
(xii) Systematic error: In Phylogenetic study, there are errors which
arise due to the use of an inappropriate model of character
evolution. The error is accentuated by the addition of sufficient
data points. This process makes it ‘systemic’. In Phylogenetics,
less number of parameters might lead to under-fitting problems.
On the other hand, the presence of too many parameters in a
dataset would lead to over-fitting problem. All these might lead
to create systematic error.

5.5 DIFFERENT ASPECTS OF PHYLOGENETIC STUDIES

The important features of Phylogenetic analyses are the visualizations

of the species relationships. The visualizations are done by different
types of Phylogenetic trees. The Phylogenetic trees have some important
features as follows:
1. The topology of the tree: The topology of a Phylogenetic tree
refers to the relatedness among the various species used for the
construction of the tree. Phylogenetic trees having the same
patterns of species distributions would reflect the similar kinds
of biological relatedness among the species.
2. Branches of the tree and branch lengths: The branch lengths
of the trees would represent the transmission of the genetic data
among the species. The branch lengths would reflect the genetic
changes or the mutations that are responsible for the speciation.
A longer branching would indicate more variations among the
species. The differences in genetic data among the species are
measured by calculating the number of changes in the nucleotide
or amino acid sequences.
3. Nodes in Phylogenetic trees: The nodes in a Phylogenetic tree are
the end points in a Phylogenetic tree. The nodes would typically
5.10 Introduction to Bioinformatics

represent a species in the Phylogenetic tree. The nodes can be of

three different types:
(a) Tip: The sequences used to construct the Phylogenetic trees are
present at the single terminal branches. They are called tips or
external nodes.
(b) Internal nodes: The nodes where the different branches meet
in the Phylogenetic trees are called internal nodes. The internal
nodes would represent the ancestral sequence.
(c) Root: The root is the most recent common ancestor of all the
sequences used for the construction of the Phylogenetic tree.
It is the oldest part of the Phylogenetic tree. The root node in a
Phylogenetic tree would determine the direction of flow of the
genetic information. However, it is always difficult to identify the
common ancestor of all the species and consequently, the addition
of a root node in the Phylogenetic tree is not an easy task. The
following figure would depict the rooted and un-rooted trees.

Figure: A, B, C, D and E are the tip nodes.

It is therefore very important to put a root node in a Phylogenetic tree

as this is the most critical aspect of a Phylogenetic analysis. The presence
of the root node would be the most critical aspect of determining the
direction of flow of genetic information.

5.6 POSITING OF ROOT NODE IN A PHYLOGENETIC TREE

As mentioned earlier that the presence of a root node is the most

critical aspect of a Phylogenetic analysis as this root node would guide
 Phylogenetic Analysis 5.11

the direction of flow of genetic data among the species. However, the
positioning of the root node in a Phylogenetic tree is based on the
following two assumptions:
1. Addition of the root node as and out-group: In this assumption
a specific molecular information (sequence of DNA, r-RNA or
amino acid sequence of a protein) is chosen in such a way that the
sequence belongs to the same biological family of the sequences
of interest to be used for the Phylogenetic analysis but it is more
distantly related to them. The aforementioned specific molecular
information is then considered as the out-group. The positioning of
the root then simply depends on the sequence similarity measures
between the sequences of interest and the out-group sequence.
2. Addition of the root to the midpoint of the Phylogenetic tree: In
this assumption, it is considered that all the sequences are being
evolved at the same time. The root node in such cases should be
placed at the midpoint of the two longest branches.

5.7 STEPS FOR CONSTRUCTION OF PHYLOGENETIC TREES

There is no short-cut ways to construct a Phylogenetic tree. However, while

building a Phylogenetic tree the following points are to be considered:
1. To identify a sequence of interest: The first step in Phylogenetics
study is the presence of a sequence of interest for which the
Phylogenetic relationship is to be studied. In principle anything
starting from a single gene to the whole genome may be used for
Phylogenetic analysis. However, the choice of sequence of interest
is guided by the availability of the sequence. For example to study
the microbial evolution pattern, the small subunit of r-RNA is
generally used. However, the main drawback of using the r-RNA
sequence is that the rates of evolution of the r-RNA sequences vary
between different species. Another problem associated with the
use of the r-RNA sequence is that the rate of evolution of r-RNA
sequence is very slow, which makes it a difficult choice to study
the Phylogenetic relationships of rapidly evolving genes.
5.12 Introduction to Bioinformatics

2. To gather molecular sequence information: The other related

sequences are to be gathered from existing databases.
3. Alignment of the sequences: The quality of a Phylogenetic tree
depends on a proper alignment of the biological sequence under
consideration. A faulty sequence alignment would produce an
erroneous tree. The details of sequence alignment methods can
be obtained from chapter 3.
4. Determining the tree building method (substitution model):
The most commonly used methods for tree building are based
on distance matrices. The method starts with creating n numbers
of single clusters containing one species from the list of species
under consideration based on the sequence alignments. This
would create a n × n matrix. Then differences between the clusters
are determined using a distance function. The clusters with the
minimum distances between them are grouped together. This
process is repeated until one cluster which is the most distant
one from the rest of the clusters is left. There are different
varieties of the distance matrix methods available. One of them
is the Unweighted Pair Group Method Using Arithmetic Mean
(UPGMA) method. In this method the distances between the
clusters are determined as a simple average as each candidate is
weighted equally. The UPGMA method assumes that the evolution
occurs at the same rate on all branches of the tree and the distances
between the nodes of the tree are additive. In the Weighted Pair
Group Method Using Arithmetic Mean (WPGMA) method the
clusters are weighted according to their sizes. The method of
Neighbour Joining (NJ) is somewhat similar to UPGMA but it
does not use the principle of addition. This method chooses the
species on the basis of multiple sequence alignment results. The
species with the maximum numbers of matching residues are
considered to be belonging to the same cluster. Another method
called Maximum Parsimony Method uses the information
about the minimum numbers of mutations required to convert
one sequence to another. This is generally achieved from the
 Phylogenetic Analysis 5.13

multiple sequence alignment results. The Maximum Likelihood

Method uses a user defined expectation model for determining
the relatedness between the sequences.
5. Building of Phylogenetic trees: The Phylogenetic trees are
constructed by using the information from the substitution model.
The popular software tools for such analyses are PAUP and
PHYLIP. There are other packages also like MEGA and MacClade
which are also used these days.
6. Evaluations of the trees: There are no specific ways to verify
the reliability of a Phylogenetic tree. If the different methods
of tree construction would come up with similar results then it
would indicate a good tree quality. Another way to analyze the
reliability of a Phylogenetic tree is bootstrapping. In this method,
the data are randomly sampled from any position within a multiple
sequence alignment. A good Phylogenetic tree would have a high
percentage of bootstrapping score.
7. Determining the errors associated with the tree building
method: All the methods have their own short-comings. The
results of Phylogenetic analyses are to be interpreted considering
the inherent approximations associated with the methods
employed.
8. To come up with new biological ideas: The ultimate goal of
the Phylogenetic analysis is to identify the positioning of the
sequence of interest in the realm of biological environment. A
proper Phylogenetic analysis is capable of providing insights into
the identity of the sequence of interest.

5.8 THE RELIABILITY OF A PHYLOGENETIC ANALYSIS

The most important aspect of Phylogenetics is the reliability issue.

This is very critical for all types of computational analyses. In case of
phylogenetics, there is no existing information regarding the ancestor for
the sequence of interest. It is also a fact that some sequences are more
informative than others. However, the most commonly used method of
5.14 Introduction to Bioinformatics

providing the confidence about a Phylogenetic analysis is bootstrapping.

In the following figure the bootstrap values are given in a Phylogenetic
tree.

Figure: A Phylogenetic tree with bootstrap values of 60, 80 and 100

From the above figure it is clear that the Phylogenetic analyses have
been performed 100 times with A, B, C and D. Out of 100 such replicates,
80% of times B appeared at the same position as presented in the tree
while the same for A and C are 60% and 100% respectively. Thus, it is
conclusive that C and D are distant neighbours and C is related to B.
All said and done, all the Phylogenetic analyses methods have their
own shortcomings. Among them the major ones are:
(a) Limitations in sequence alignment methodologies
(b) Not being able to differentiate between the differences in rates of
evolutions in the different parts of the same sequence
(c) Not being able to differentiate between the differences in rates of
evolutions of the same sequence in the different species.
Different labs are constantly trying to sort out these problems. A few
recent papers justifying such endeavors are presented in following section:
1. RAxML version 8: a tool for Phylogenetic analysis and
post-analysis of large phylogenies, Alexandros Stamatakis,
Bioinformatics, Volume 30, Issue 9, 1 May 2014, Pages 1312–
1313, https://doi.org/10.1093/bioinformatics/btu033
2. Verdant: automated annotation, alignment and Phylogenetic
analysis of whole chloroplast genomes, Michael R. McKain,
Ryan H. Hartsock, Molly M. Wohl, Elizabeth A. Kellogg,
Bioinformatics, Volume 33, Issue 1, 1 January 2017, Pages
 Phylogenetic Analysis 5.15

130–132, https://doi.org/10.1093/bioinformatics/btw583
3. ClockstaR: choosing the number of relaxed-clock models in
molecular Phylogenetic analysis, Sebastián Duchêne, Martyna
Molak, Simon Y. W. Ho, Bioinformatics, Volume 30, Issue
7, 1 April 2014, Pages 1017–1019, https://doi.org/10.1093/
bioinformatics/btt665
4. MulRF: a software package for Phylogenetic analysis using
multi-copy gene trees,
Ruchi Chaudhary, David Fernández-Baca, John Gordon Burleigh,
Bioinformatics, Volume 31, Issue 3, 1 February 2015, Pages
432–433, https://doi.org/10.1093/bioinformatics/btu648
5. Unrealistic Phylogenetic trees may improve Phylogenetic
footprinting, Martin Nettling, Hendrik Treutler, Jesus Cerquides,
Ivo Grosse, Bioinformatics, Volume 33, Issue 11, 1 June 2017,
Pages 1639–1646, https://doi.org/10.1093/bioinformatics/btx033
6. markophylo: Markov chain analysis on Phylogenetic trees,
Utkarsh J. Dang, G. Brian Golding, Bioinformatics, Volume 32,
Issue 1, 1 January 2016, Pages 130–132, https://doi.org/10.1093/
bioinformatics/btv541

5.9 DETERMINATION OF EVOLUTIONARY RATE AND TIME

It is a well established fact that the genetic change in a lineage is measured

as the product of evolutionary rate and time of divergence.
Genetic change: The genetic change is expressed as the number of
mutations or substitutions per site.
Evolutionary rate: The evolutionary rate is the amount of genetic
change per unit time. The time scale is measured in year.
Time of divergence: The time of divergence is the total time required
for the entire speciation.
The mathematical form of calculation of the genetic change is
Genetic change = Evolutionary rate X Time of divergence
5.16 Introduction to Bioinformatics

However, the appropriate calculation of the genetic change would

require the knowledge of the root node.

Fish Frog Lizard Mouse Human

Figure: Evolution of vertebrates with their ancestors

represented by the circles

From the above figure it can be concluded that human is closely related
to mouse as both of them have a common ancestor expressed by the black
circle. Similarly frogs share a recent ancestor with fish represented by
another circle.

FINAL WORDS
Phylogenetic analyses would involve the use of biological sequences.
The most commonly used sequences are amino acid sequences and DNA
sequences. However, the choice of the biological sequence depends
on the problem. For the evolutionary study of primates mitochondrial
DNA is preferred. To study the relatedness between divergent species,
Ribosomal RNA sequence is generally used. However, the success behind
a good Phylogenetic analysis depends on the accuracy of the sequence
alignment methods. The problem of sequence alignment can be solved by
developing a robust alignment method. It is also to be noted that there is
no guarantee that a Phylogenetic tree would represent a true evolutionary
pathway. However, in order to solve the problem different methods of
tree building may be employed. If the same result is obtained by all the
methods, the Phylogenetic analyses may be considered to represent a
true evolutionary picture. In general, bootstrapping is commonly used
to randomly sample the available data. In an ideal case, the Phylogenetic
trees obtained from bootstrapping analyses should match the original tree.
However, in reality a bootstrapping result of 70% for any given branch in
Phylogenetic tree is considered to provide nearly 95% confidence level
about the correctness of the branch.
 Phylogenetic Analysis 5.17

Some computational resources for Phylogenetic analyses

Ensemble (http://www.ensembl.org/index.html): This is a repository

of genomic information
Ensembl compara (http://www.ensembl.org/info/genome/compara/
index.html): This contains Phylogenetic information which can be freely
downloaded.
ClustalW2Phylogeny(http://www.ebi.ac.uk/Tools/phylogeny/simple_
phylogeny/): This tool can be used to construct Phylogenetic trees.
EMBOSS (https://www.ebi.ac.uk/Tools/sfc/emboss_seqret/): Another
tool for Phylogenetic analysis.
Phyllip (http://evolution.genetics.washington.edu/phylip.html): A
free software package for Phylogenetic analyses.
Dendroscope (http://ab.inf.uni-tuebingen.de/software/dendroscope/):
It is a freely available tool for Phylogenetic study.
HyPhy (http://www.hyphy.org/): It is an online tool for testing the
hypothesis of Phylogenetics.
MEGA (http://www.megasoftware.net/): It is a comprehensive tool
for Phylogenetic analyses.
PAUP (http://www.paup.sc.fsu.edu/): It is a comprehensive tool for
Phylogenetic analyses.
Splitstree (http://www.splitstree.org/): It is a tool for building un-
rooted tree.
Treefinder (http://www.treefinder.de/): It is a comprehensive tool for
Phylogenetic analyses.
Ugene (http://ugene.net/): It is a bioinformatics software
T-REX (www.trex.uqam.ca.): A webserver for Phylogenetic study

SOME IMPORTANT QUESTIONS AND ANSWERS

1. Overall phenotypic similarity may not always reflect evolutionary
relationships—
5.18 Introduction to Bioinformatics

(a) due to convergent evolution

(b) because of variation in rates of evolutionary change of different
kinds of characters
(c) due to homoplasy
(d) all of the above.
2. Cladistics
(a) is based on overall similarity of phenotypes
(b) requires distinguishing similarity due to inheritance from a common
ancestor from other reasons for similarity
(c) is not affected by homoplasy
(d) none of the above.
3. The principle of parsimony in Phylogenetics
(a) helps evolutionary biologists distinguish among competing
Phylogenetic hypotheses
(b) does not require that the polarity of traits be determined
(c) is a way to avoid having to use outgroups in a Phylogenetic analysis
(d) cannot be applied to molecular traits.
4. The concept of Phylogenetic species
(a) depends on whether individuals from different populations can
successfully breed
(b) is indistinguishable from the biological species concept
(c) does not apply to allopatric populations
(d) is based on evolutionary independence among populations.
5. Parsimony would suggest that parental care in birds, crocodiles, and
some dinosaurs
(a) evolved independently, multiple times by convergent evolution
(b) evolved once in an ancestor common to all three groups
(c) is a homoplastic trait
(d) is not a homologous trait.
 Phylogenetic Analysis 5.19

6. The reappearance of lost traits, especially if they are complex

(a) can be identified with Phylogenetic analyses
(b) never happens
(c) is not an example of a reversal
(d) does not affect interpretation of evolutionary relationships.
7. The term molecular clock pertaining to evolutionary biology and
Phylogenetics
(a) refers to a group of proteins that induce endogenous circadian
rhythms in animals
(b) is an undisputed assumption that all biological molecules evolve
at a constant rate
(c) may help provide a way to estimate the absolute timing of historical
events in evolution
(d) applies only to organisms that reproduce sexually.
8. A taxonomic group containing a common ancestor, which leaves out a
descendant group is known as
(a) paraphyletic
(b) monophyletic
(c) polyphyletic
(d) a good cladistic group.
9. The forelimbs of birds and rhinoceros
(a) are homologous and symplesiomorphic
(b) are not homologous but are symplesiomorphic
(c) are homologous and synapomorphic
(d) are not homologous but are synapomorphic.
10. In order to determine polarity in different states of a character
(a) there must be a fossil record of the groups in question
(b) genetic sequence data must be available
5.20 Introduction to Bioinformatics

(c) an appropriate name for the taxonomic group must be selected

(d) an outgroup must be identified.
11. A paraphyletic group
(a) includes an ancestor and all of its descendants
(b) an ancestor and some of its descendants
(c) descendants of more than one common ancestor
(d) all of the above.
12. Sieve tubes and sieve elements
(a) are homoplastic because they have different function
(b) are homologous because they have similar function
(c) are homoplastic because their common ancestor was single-celled
(d) are structures involved in transport within animals.
13. The phylogeny of dinosaurs leading to birds
(a) demonstrates that the first function of feathers was flight
(b) demonstrates that feathers and wings evolved simultaneously
(c) suggests that complex characters evolve rapidly, in one step
(d) reveals many transitional forms between modern birds and their
ancestors.
14. The Phylogenetic analysis of HIV suggests
(a) a single origin of HIV from primates
(b) multiple origins of HIV from several different primate species
(c) multiple origins of HIV from a single primate species
(d) that SIV originated from HIV.
 Phylogenetic Analysis 5.21

ANSWERS

1. (d) 2. (b) 3. (a) 4. (d) 5. (b) 6. (a)

7. (c) 8. (a) 9. (a) 10. (d) 11. (b) 12. (c)
13. (d) 14. (b)
 Few More Important
Questions and Answers

1. The term Bioinformatics was invented by

(a) Paulien Hogeweg, 1979.

(b) Dr Margaret Oakley Dayhoff, 1976.

(c) Robert Ledley, 1978. D. David W Mount, 1977.

2. Find the odd one out

(a) GenBank.

(b) DDBJ.

(c) EMBL.

(d) TREMBL.

3. Find the odd one out

(a) SWISS­PROT.

(b) EMBL.

(c) DDBJ.

(d) NCBI.

4. Find the secondary database

(a) DDBJ.

(b) PROSITE.

(c) NRDB.

(d) OWL.
QA.2 Introduction to Bioinformatics

5. Find a composite database.

(a) PROSITE.

(b) DDBJ.

(c) NRDB.

(d) EMBL.

6. Find a primary protein structure database.

(a) PDB.

(b) PubChem.

(c) ChemBank.

(d) SCOP.

7. Find a secondary protein structure database

(a) PubChem.

(b) PDB.

(c) ChemBank.

(d) SCOP.

8. FASTA format starts with which symbol?

(a) /.

(b) *.

(c) >.

(d) #.

9. A complementary DNA database is

(a) Swiss­Prot.

(b) GenBank.

(c) UniSTS.

(d) NRDB.
 Few More Important Questions and Answers QA.3

10. Find a bibliographic database.

(a) PubMed.

(b) Entrez.

(c) PIR.

(d) EBI.

11. A life science search engine is

(a) PubMed.

(b) Entrez.

(c) Mozilla.

(d) EBI.
12. PubMed is maintained by
(a) NCBI.
(b) EMBL.
(c) DDBJ.
(d) SWISS­PROT.
15. Entrez is maintained by
(a) SWISS­PROT.
(b) EMBL.
(c) DDBJ.
(d) NCBI.
16. Find a similarity search tool.
(a) BLAST.
(b) CLUSTALW.
(c) CLUSTALX.

(d) RASMOL.
QA.4 Introduction to Bioinformatics

17. BLAST contains ------- programs.

(a) Four.
(b) Five.
(c) Six.
(d) Seven.
18. BLAST accepts ----- sequence format.
(a) GenBank.
(b) EMBL.
(c) FASTA.

(d) PIR.
19. Which tool compares protein sequence against protein databases?
(a) blastp.
(b) blastn.
(c) blastx.

(d) tblastx.
20. Which tool compares nucleotide sequence against DNA databases?
(a) blastn.
(b) blastp.
(c) tblastx.

(d) tblastn.
21. Which tool compares translated nucleotide query sequence against
protein databases?
(a) blastp.
(b) tblastn.
(c) blastx

(d) tblastx.
 Few More Important Questions and Answers QA.5

22. Which tool compares protein sequence against translated nucleotide

databases?
(a) blastp.
(b) tblastx.
(c) blastn.

(d) tblastn.
23. Which tool compares translated nucleotide query sequence against
translated nucleotide databases?
(a) blastp.
(b) blastn.
(c) tblastx.

(d) tblastn.
24. Which E­value provides the more significant the hit?
(a) lower. (b) higher.

(c) average. (d)superior.

25. Swiss­Prot is housed and maintained by.
(a) NCBI. (b) NBRF.

(c) SIB. (d) DDBJ.

26. The full form of ExPASy is
(a) Expert Protein Analysis Server.
(b) Exponential Protein Analysis Server.
(c) Expert Protein Analysis System.
(d) Exponential Protein Analysis System.
27. Which web resource is maintained by NCBI for the retrieval of gene
based information?
(a) LocusLink. (b) PDB.
(c) MSD. (d) PRF.
QA.6 Introduction to Bioinformatics

28. EST is
(a) Expressed Sequence Tag.
(b) Expressed Site Tag.
(c) Expressed Structure Tag.
(d) Expressed Symbol Tag.
29. SNP is
(a) Small Nucleic Polymorphism.
(b) Single Nucleic Polymorphism.
(c) Single Nucleotide Polymorphism.
(d) Small Nucleotide Polymorphism.
30. What is E­value?
(a) The chance that a random sequence could achieve a better score
than the query.
(b) The chance that a homologous sequence could achieve a similar
score to the query.
(c) The chance that a random sequence could achieve a worse score
than the query.
(d) The chance that a homologous sequence could achieve a better
score than the query.
31. PROSITE is
(a) A database of protein structures.
(b) A database of protein sequences.
(c) A database of protein motifs.
(d) Options a and b.
37. Fingerprint is
(a) A protein family discriminator built from a set of regular
expressions.
(b) A protein family discriminator built from a set of conserved motifs.
(c) A cluster of protein sequences gathered from a BLAST search.
(d) A cluster of protein sequences gathered from a FASTA search.
 Few More Important Questions and Answers QA.7

38. What are the conserved regions in multiple sequence alignments?

(a) Reflect areas of structural importance.
(b) Reflect areas of functional importance.
(c) Reflect areas of both functional and structural importance.
(d) Reflect areas likely to be of functional and/or structural importance.
39. How is a motif identified?
(a) COPIA. (b) Patternhunter.
(c) PROSPECT. D. BLAST.
43. A PATTERN is
(a) block. (b) profile.
(c) regular expression. (d) fuzzy regular expression.
44. InterPro is.
(a) integrated protein family.
(b) integrated protein sequence.
(c) integrated protein structure.
(d) integrated protein interaction.
45. More than two sequences are aligned by
(a) Multiple sequence alignment
(b) Pairwise sequence alignment
(c) Global alignment
(d) Local alignment
46. The family of related genes in an organism is called
(a) orthologs. (b) zoologs.

(c) paralogs. (d) xenologs.

47. PAM and BLOSUM are
(a) Distance Matrix (b) Sequence search method
(c) Both (d) None.
QA.8 Introduction to Bioinformatics

48. Which characteristics of a protein family are defined by the existence

of a multiple sequence alignment of a group of homologous sequences?
(a) Domains. (b) DNA.
(c) Proteins. (d) RNA.
49. ClustalW accepts inputs in the ---- format.
(a) genbank.
(b) embl.
(c) pdb.

(d) fasta.
50. Comparison of two sequences is called
(a) Global alignment
(b) Local alignment
(c) Pairwise sequence alignment
(d) Multiple sequence alignment
51. Comparison of more than two sequences is called
(a) Global alignment
(b) Local alignment
(c) Pairwise sequence alignment
(d) Multiple sequence alignment
52. Highly similar sequences are aligned by
(a) Global alignment
(b) Local alignment
(c) Pairwise sequence alignment
(d) Multiple sequence alignment
53. An optimum alignment _____ number of matches and ______ the
number of gaps.
 Few More Important Questions and Answers QA.9

(a) minimizes, maximizes.

(b) maximizes, minimizes.
(c) degrades, upgrades.
(d) upgrades, degrades.
54. Multiple sequence alignment method is known as.
(a) global. (b) local.
(c) progressive. (d) non­
progressive.
55. A Phylogenetic tree showing the branching representing evolutionary
distance is
(a) Phylogram.
(b) Cladogram.
(c) A guide tree.
(d) Cardiogram.

56. Pfam database is obtained from .

(a) SMART.

(b) PRINTS.

(c) PROSITE. D. PRODOM.

57. High quality in Pfam is given by

(a) Pfam­A.

(b) Pfam­B.

(c) Pfam­C. D. Pfam­D.

58. Low quality of Pfam is given by

(a) Pfam­D.

(b) Pfam­B.

(c) Pfam­A. D. Pfam­C.

QA.10 Introduction to Bioinformatics

59. Full form of CDD is

(a) Conserved Domain Database.

(b) Conserved Dictionary Database.

(c) Conserved Domain Dictionary.

(d) Conserved Dictionary Database.

60. Which BLAST program is used to search conserved domains?

(a) BLASTN.

(b) BLASTP.

(c) SNP­BLAST.

(d) PSI­BLAST.

61. Which is RPS BLAST?

(a) PHI­BLAST.

(b) PSI­BLAST.

(c) BLASTN.

(d) TBLASTX.
62. The PRINTS database is linked to
(a) SwissProt/TrEMBL.
(b) SwissProt/EMBL.
(c) PIR/TrEMBL.

(d) PIR/EMBL.
63. Dot­matrix is
(a) as the coordinates of a two­dimensional graph.
(b) are represented in the form of trees.
(c) as the coordinates of a 3D graph.
(d) not represented as graph.
 Few More Important Questions and Answers QA.11

64. PAM is
(a) Parallel Align Mutation.
(b) Point Altered Mutation.
(c) Point Accepted Mutation.
(d) Point Arranged Mutation.
65. Local alignment is done by
(a) Needleman and Wunsch. (b) PAM.
(c) Smith­Waterman. (d) All the above.
66. Global alignment is done by
(a) Needleman and Wunsch. (b) Smith­
Waterman.

(c) BLAST. (d) PAM .

67. PAM was developed by
(a) Needleman and Wunsch, 1976.
(b) Smith­Waterman, 1978.
(c) Dayhoff et al., 1978.

(d) Henikoff 1992.

68. BLOSUM is
(a) Blocks of Amino Acid Substitution Mutation.
(b) Basic Amino Acid Substitution Mutation.
(c) Blocks of Amino Acid Substitution Matrix.

(d) Basic Amino Acid Substitution Matrix.

69. PAM matrices are obtained from .
(a) Needleman and Wunsch.
(b) Smith­Waterman.
(c) Dayhoff model.
(d) Markov model.
QA.12 Introduction to Bioinformatics

70. A mismatch refers to

(a) gap.
(b) deletion.
(c) insertion.

(d) substitution mutation.

71. Gene duplication leads to
(a) orthologs.
(b) paralogs.
(c) xenologs.
(d) zoologs.
79. Speciation leads to
(a) zoologs.
(b) paralogs.
(c) xenologs.
(d) orthologs.
80. The paired dots in the sequence alignments would determine
(a) conserved substitutions.
(b) semi­conserved substitutions.
(c) gaps.
(d) identity.
81. A single dot in the sequence alignments would determine
(a) identity.
(b) semi­conserved substitutions.
(c) conserved substitutions.
(d) gaps.
82. BLAST2 is used for comparison of .
(a) two. (b) three.
(c) four. (d) five.
 Few More Important Questions and Answers QA.13

83. Two main ways to construct guide tree in progressive alignment is

(a) UPGMA and Neighbor joining method.
(b) Maximum Parsimony.
(c) Maximum Likelihood.
(d) all the above.
84. Genetic recombination is shown by
(a) Progressive.
(b) Dynamic Programming.
(c) Genetic Algorithm.
(d) Hidden Markov Model.
85. Related protein sequences are used for
(a) PAM1.
(b) PAM45.
(c) PAM60.
(d) PAM250.
86. Distantly related proteins are used in
(a) PAM1.
(b) PAM250.
(c) PAM60.
(d) PAM45.
87. A Twilight Zone is
(a) Where alignments appear plausible and are statistically significant.
(b) Where alignments may appear plausible to the eye, but are no longer
statistically significant.
(c) Where alignments neither appear plausible nor statistically
significant.
(d) Where alignments share 30% identity.
QA.14 Introduction to Bioinformatics

88. Protein domains are presented by

(a) class. (b) architecture.
(c) taxonomy. (d) homologs.
89. CATH and SCOP are meant for
(a) the structural family to which a protein belongs.
(b) the generic family to which a protein belongs.
(c) homologous proteins.
(d) analogous proteins.
90. The coordinates of the amino acid residues in proteins are present in
(a) CATH. (b) SCOP.
(c) PDBsum. (d) PDB.
91. Homology modelling is
(a) Due to low sequence similarity between proteins of unknown and
known structure, the structure is predicted from first principles.
(b) Due to high sequence similarity between proteins of unknown and
known structure, the same function is assumed for both.
(c) Due to high sequence similarity between proteins of unknown and
known structure, the structure of the latter is used as a template to
model the former.
(d) A protein of unknown structure is compared against a library of
fold templates to find the best match.
92. Threading is
(a) Due to low sequence similarity between proteins of unknown and
known structure, the structure is predicted from first principles.
(b) Due to high sequence similarity between proteins of unknown and
known structure, the same function is assumed for both.
(c) Due to high sequence similarity between proteins of unknown and
known structure, the structure of the latter is used as a template to
model the former.
(d) A protein of unknown structure is compared against a library of
fold templates to find the best match.
 Few More Important Questions and Answers QA.15

93. PDBID is expressed as

(a) SMILES. (b) ROSDAL.
(c) WLN. (d) ALPHANUMERIC.
94. PDB does not accept the structures obtained by
(a) X­
ray crystallography. (b) NMR.
(c) Mass spectrometry. (d) Comparative modelling.
95. Fold recognition is
(a) Comparative modelling. (b) Threading.
(c) Abinitio. (d) Homology modelling.
96. SCOP is
(a) Similar Classification of Proteins.
(b) Structural Classification of Proteins.
(c) Similar Characterization of Proteins.
(d) Similar Classification of Proteins.
97. What is present in SCOP?
(a) primary structure of protein.
(b) secondary structure of protein.
(c) protein fold.
(d) mutated protein sequences.
98. Template based protein modelling is done in
(a) comparative modelling. (b) surface modelling.
(c) threading. (d) abinitio prediction.
99. Gaps are allowed in
(a) Maximum parsimony.
(b) Maximum likelihood.
(c) Neighbor Joining.
(d) Unweighted pair group method with arithmetic mean.
QA.16 Introduction to Bioinformatics

100. Distance based method is

(a) Unweighted pair group method with arithmetic mean.
(b) Jukes­Cantor.
(c) Neighbuor Joining
(d) Jukes-Cantor

ANSWERS

1. (a) 2. (d) 3. (a) 4. (b) 5. (c)

6. (a) 7. (d) 8. (c) 9. (c) 10. (a)
11. (b) 12. (a) 15. (d) 16. (a) 17. (b)
18. (c) 19. (a) 20. (a) 21. (c) 22. (d)
23. (c) 24. (a) 25. (c) 26. (c) 27. (a)
28. (a) 29. (c) 30. (b) 31. (c) 37. (b)
38. (b) 39. (a) 43. (c) 44. (a) 45. (a)
46. (c) 47. (a) 48. (a) 49. (d) 50. (c)
51. (d) 52. (a) 53. (b) 54. (c) 55. (a)
56. (d) 57. (a) 58. (b) 59. (a) 60. (d)
61. (b) 62. (a) 63. (a) 64. (c) 65. (c)
66. (a) 67. (c) 68. (c) 69. (c) 70. (d)
71. (b) 79. (d) 80. (a) 81. (b) 82. (a)
83. (a) 84. (c) 85. (a) 86. (b) 87. (c)
88. (a) 89. (a) 90. (d) 91. (c) 92. (d)
93. (d) 94. (d) 95. (b) 96. (b) 97. (c)
98. (a) 99. (a) 100. (a)
 S
Further Reading

Allen, M. P., & Tildesley, D. J. (1989) Computer Simulation of

Liquids, 1989, Oxford University Press, ISBN 0-19-855645-4.
Alberts, J., Lewis, R., & Roberts, W. (2008) Molecular Biology of
the Cell.
Anfinsen, C. (1972). The formation and stabilization of protein
structure. Biochem. J. 128 (4): 737–749.
Attwood, T.K, Parrysmith D-J (2001). Introduction to Bioinformatics.
1stedition. Published by Benjamin Cummings. ISBN-10: 0582327881,
ISBN-13: 978-0582327887.
Baker, D., Sali, A. (2001). Protein structure prediction and structural
genomics. Science 294 (5540): 93–96.
Baron, R., Setny, P., & Andrew M.J. (2010). Water in Cavity-Ligand
Recognition”. Journal of the American Chemical Society 132 (34):
12091–12097.
Baxevanis A. D., Francis Ouellette, B.F (2004). Bioinformatics: A
Practical Guide to the Analysis of Genes and Proteins. Wiley-Blackwell;
3rd Revised edition. ISBN-10: 0471478784, ISBN-13: 978-0471478782.
Branden, C., & Tooze, J. (1999) Introduction to Protein Structure,
Second Edition. Garland Publishing Inc., 19 Union Square West, NY 10003.
Bradford et al., (2006). Insights into protein–protein interfaces using
a Bayesian network prediction method. J. Mol. Biol. 362 (2).
Bradford, J.R. & Westhead, D.R. (2005). Improved prediction of
protein–protein binding sites using a support vector machines approach.
Bioinformatics. 21(8).
F.2 Introduction to Bioinformatics

Bradford, J.R. et al., (2006). Insights into Protein–Protein Interfaces

using a Bayesian Network Prediction Method. J.Mol.Biol. 362(2).
Bogan, A.A. &Thorn, K.S. (1998) Anatomy of hot spots in protein
interfaces. J.Mol.Biol. 280(1).
Bogan, A.A. &Thorn, K.S. (2001). ASEdb: a database of alanine
mutations and their effects on the free energy of binding in protein
interactions. Bioinformatics 17(3).
Bowie, J.U., Lüthy, R., & Eisenberg, D. (1991). A method to identify
protein sequences that fold into a known three-dimensional structure.
Science 253 (5016): 164–170.
Brown, S.M. (2000) Bioinformatics: A Biologist’s Guide to
Biocomputing and the Internet. Eaton Pub Co. ISBN-10: 188129918X,
ISBN-13: 978-1881299189.
Buehler, L.K, Rashidi, H.H (2005). Bioinformatics Basics:
Applications in Biological Science and Medicine. CRC Press. ISBN
9780849312830.
Burgoyne, N.J. & Jackson, R.M. (2006) Predicting protein interaction
sites: binding hot-spots in protein-protein and protein-ligand interfaces.
Bioinformatics 22(11).
Cavanagh, J., Fairbrother, W.J., Palmer, A.G. III, Skelton, N.J., Rance,
M. (2006) Protein NMR Spectroscopy: Principles and Practice, Second
Edition, Academic Press.
Chiang, Y.S., Gelfand, T.I., Kister, A.E., & Gelfand, I.M. (2007).
New classification of supersecondary structures of sandwich-like proteins
uncovers strict patterns of strand assemblage. Proteins. 68 (4): 915–921.
Chothia, C., & Lesk, A.M. (1986). The relation between the
divergence of sequence and structure in proteins. EMBO J 5 (4): 823–6.
Chung, S.Y., & Subbiah, S. (1996) A structural explanation for the
twilight zone of protein sequence homology. Structure 4: 1123–27.
 Further Reading F.3

Clavarie J-M, Notredame C. (2006) Bioinformatics For Dummies.

Second Edition. Wiley. ISBN: 978-0-470-08985-9.
Creighton T.E. (1992) Proteins: Structures and Molecular Properties,
Second Edition. W. H. Freeman, NY.
Deniz, A. et al., (2000) Single-molecule protein folding: Diffusion
fluorescence resonance energy transfer studies of the denaturation of
chymotrypsin inhibitor 2. Proc. Natl Acad. Sci. USA, 97(10).
Dill, Ken A. et al. (2007). The protein folding problem: when will it
be solved? Current Opinion in Structural Biology, 17:342–346.
Drenth, J. (2007) Principles of Protein X-ray Crystallography, Second
Edition. Springer.
Ezkurdia et al. (2009) Progress and challenges in predicting protein-
protein interaction sites. Briefings in Bioinforamtics 10(3).
Fariselli, P. et. al., (2002). Prediction of protein–protein interaction
sites in heterocomplexes with neural networks. Eur. J. Biochem 269 (5).
Fiser, A, & Sali, A. (2003). “ModLoop: automated modelling of loops
in protein structures”. Bioinformatics 19 (18): 2500–1.
Frenkel, D., & Smit, B.(1996) Understanding Molecular Simulation:
From Algorithms to Applications ISBN 0-12-267370-0.
Friedrich et al., (2006). Modelling interaction sites in protein domains
with interaction profile hidden Markov models. Bioinformatics. 22(23).
Fresht, A. (1999) Structure and Mechanism in Protein Science: A
Guide to Enzyme Catalysis and Protein Folding, First Edition. W. H.
Freeman, NY.
Gibas, C., Jambeck P. (2001) Developing Bioinformatics Computer
Skills An Introduction to Software Tools for Biological Applications.
O’Reilly Media.
Govindarajan, S., Recabarren, R., & Goldstein, R.A. (1999).
Estimating the total number of protein folds. Proteins. 35 (4): 408–414.
F.4 Introduction to Bioinformatics

Jones, D.T., Taylor, W.R., & Thornton, J.M. (1992). A new approach
to protein fold recognition. Nature 358 (6381): 86–89.
Jones, S. & Thornton, J.M. (1997). Analysis of protein-protein
interaction sites using surface patches. J.Mol.Biol. 272(1).
Jones, S. & Thornton, J.M. (1997). Thornton, Prediction of protein-
protein interaction sites using patch analysis. J.Mol.Biol. 272(1).
Kaczanowski, S., & Zielenkiewicz, P. (2010). Why similar protein
sequences encode similar three-dimensional structures?. Theoretical
Chemistry Accounts 125: 643–650.
Kitchen, D.B., Decornez, H., Furr, J.R., & Bajorath, J. (2004).
Docking and scoring in virtual screening for drug discovery: methods
and applications. Nature reviews. Drug discovery3 (11): 935–949.
Koike, A., & Takagi, T. (2004). Prediction of protein–protein
interaction sites using support vector machines. Protein Eng. Des. Sel.
17(2).
Kyte, J (2006). Structure in Protein Chemistry, Second Edition.
Garland Publishing Inc., 19 Union Square West, NY 10003
Leach, A. R. (2001). Molecular Modelling: Principles and
Applications, ISBN 0-582-38210-6.
Lengauer, T., & Rarey, M. (1996). Computational methods for
biomolecular docking. Curr. Opin. Struct. Biol. 6 (3): 402–406.
Lesk, A. (2008). Introduction to Bioinformatics. OUP Oxford; 3
edition, ISBN-10: 0199208042.
ISBN-13: 978-0199208043.
Levitt, M. (1992). Accurate modelling of protein conformation by
automatic segment matching. J Mol Biol 226 (2): 507–33.
Mount, D.M. (2004). Bioinformatics: Sequence and Genome Analysis
2nd ed. Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY.
Nelson, D. L., & Cox, M. L. (2008) Principles of Biochemistry 5th
Edition. M. W. H. Freeman; June 15, 2008. ISBN 1429224169.
 Further Reading F.5

Ofran, Y. & Rost, B (2003). Analysing six types of protein–protein

interfaces. J.Mol.Biol. 325(2).
Orego, C, Jones, D. , Thornton, J. (2002) Bioinformatics: Genes,
Proteins and Computers. Taylor & Francis; 1 edition. ISBN-10:
1859960545, ISBN-13: 978-1859960547.
Ramachandran K.I Deepa, G. & Krishnan Namboori. P.K. (2008)
Computational Chemistry and Molecular Modelling Principles and
Applications. ISBN 978-3-540-77302-3 Springer-Verlag GmbH.
Rapaport,D. C. (2004) The Art of Molecular Dynamics Simulation,
2004, ISBN 0-521-82568-7.
Sali, A, Blundell, T.L. (1993). Comparative protein modelling by
satisfaction of spatial restraints. J Mol Biol 234 (3): 779–815.
Voet, D., & Voet, J.G. (2010) Biochemistry, Second Edition. John
Wiley & Sons, INC, NY.
Xiong, J. (2006). Essential Bioinformatics. Cambridge University
Press. ISBN-10: 0521600820, ISBN-13: 978-0521600828.
Zhang Y (2008). Progress and challenges in protein structure
prediction. Curr Opin Struct Biol 18 (3): 342–348.
Zemla, A. (2003). LGA – A Method for Finding 3-D Similarities in
Protein Structures. Nucleic Acids Research 31 (13): 3370–3374.
Zvelebil, M, Baum, Jeremy O. (2007). Understanding Bioinformatics.
Garland science publisher. ISBN: 9780815340249.
 Index

A Errat, 1.17
E-value, 3.3
Ab Initio Modelling, 4.1
AIC, 5.7 F
Amino Acid, 1.5 Family, 1.5
B FASTA, 2.5
Force Field, 4.14
Bayesian, 4.38
FRET, 4.34
BIC, 5.7
FSSP, 2.18
Bioinformatics, 1.1
BLAST, 1.7 G
BLOSUM, 1.7 GenBank, 2.8
Bootstrapping, 1.11 Global Alignment, 1.7
C GO, 2.18

CATH, 1.5 H
CLASS, 2.16 Hidden Markov Model (HMM), 1.19
Conjugate Gradient (CG), 4.20 HMM, 1.9
Crystallography, 4.33 Homology Modelling, 4.1
D Human Genome Project, 1.2

Dali, 1.5 I
Database, 1.6 Interactions, 4.4
DAVID, 2.18 InterPro, 1.5
DDBJ, 2.3
DNA, 1.2 L
Docking, 4.31 Lincs, 4.24
Domain, 1.3 Local Alignment, 1.7
Dynamic Programming, 1.7 LRT, 5.8
E M
EBI, 2.3 Machine Learning, 1.20
EMBL, 2.3 Maximum Likelihood, 1.10
I.2 Introduction to Bioinformatics

Micro RNA, 1.3 S

Modelling, 1.15
SCOP, 1.5
Molecular Dynamics, 4.14
Molecular Mechanics, 4.1 Secondary Database, 1.4
MSA, 1.11 Secondary Structure, 1.21
Mutation, 1.6 Sequence Alignment, 1.7
Shake, 4.24
N
Smith-Waterman Algorithm, 3.6
NCBI, 2.8 SRS, 2.4
Neighbour Joining, 1.11 SSAP, 3.8
Neural Network, 4.38 Steepest Descent (SD), 4.20
NMR, 4.33
Substitution Matrix, 3.2
Node, 5.9
Super Secondary Structure, 4.3
P Support Vector Machine (SVM), 4.37
PAM, 1.7 SwissProt, 2.11
PDB, 2.10 Systems Biology, 1.20
Periodic Boundary Condition (PBC), 4.17
T
Pfam, 1.5
Phyllogeny, 1.10 Tertiary Structure, 4.3
PIR, 2.9 Threading, 4.1
Primary Database, 2.3 Tip, 5.10
Primary Structure, 1.21 Tree, 5.10
PRINTS, 2.22 Two-Hybrid, 4.35
Procheck, 1.17
Prosite, 1.5 U
Proteomics, 120 UniProtKB, 1.5
Prove, 1.17
UPGMA, 1.9
PSI-BLAST, 1.8
UPGMA, 5.6
Pubmed, 2.18
R V

Rampage, 1.17 Verify3D, 1.17

Random Forest, 4.38
W
RefSeq, 1.5
RNA, 1.3 What_check, 1.17
Root, 5.10 WPGMA, 5.6