Mark Thompson

Planaria Software
Seattle, WA
http://www.planaria-software.com
Molecular Docking With Argusa!
"in# the optimal ligan#/protein configurations an# accuratel$ %at least
consistentl$& pre#ict their !in#ing free energ$ without using formal
statistical mechanics approaches.
igan# is fle'i!le.
Protein !in#ing site is fle'i!le %si#e chains an# protein !ack!one&.
Do this in un#er () secon#s on a commo#it$ processor %i#eall$ un#er
* secon# or faster&.
Molecular Docking
Molecular Docking
A complicated search problem
A complicated search problem
We ha+e #e+elope# two #ocking engines an# an empirical scoring function in
Argusa! ,.-.
.
ShapeDock: shape-!ase# metho#, appro'imates e'hausti+e search.
.
GADock: amarckian genetic algorithm similar to AutoDock.
.
AScore: scoring function !ase# on /Score of Wang an# coworkers.
T$pical ShapeDock times for ligan#s with *--*) torsions are 0 1- secon#s on
a 2., 345 Pentium laptop computer.

6ur #ocking co#e is implemente# for !oth interacti+e #ocking an# screening
of ligan# #ata!ases.
Abstract
Abstract
Molecular mo#eling application runs on Win#ows platforms.
ArgusLab 4.0
ArgusLab 4.0

1. Ligand is described as a torsion tree
7o#es are groups of !on#e# atoms that #o not ha+e rotata!le !on#s8
connections !etween no#es are torsions. Topolog$ of tree is crucial to
efficient #ocking. A !alance# tree with a large central no#e is !est.

2. Construct two grids that overlay the binding site
3ri# points marke# as insi#e or outsi#e the free +olume of !in#ing site.
"ine gri# use# to #etermine if atoms of a pose fragment are insi#e or
outsi#e the !in#ing site. 9oarse gri# is use# to esta!lish the search points
insi#e the !in#ing site.

3. At each search point o! interest"
igan#:s root no#e is place# on a search point an# a set of #i+erse an#
energeticall$ fa+ora!le rotations is create#. Translations near the search
point are allowe# to remo+e !umps with the target.
ShapeDock
ShapeDock
Approximates an exhaustive search Approximates an exhaustive search
( (similarities to Fred Dock and !lide" similarities to Fred Dock and !lide"
#. $or each rotation% construct the torsions in breadth&!irst order
;se pre-#efine# torsion +alues !ase# on !on# or#er of central !on#.
;se fine gri# to test newl$ a##e# atoms for !umps with protein an#
intra-ligan# contacts to accept or re<ect pose fragment. Allow small
torsion a#<ustments to minimi5e !umps. %rings are treate# as rigi#&
'. Score pose candidates
Poses can#i#ates are those that sur+i+e the torsion search. The$ are
ranke# to maintain set of 7-lowest energ$ poses %7 t$picall$ )--*)-&.
9lustering poses as the$ are foun# maintains #i+ersit$ in the final set.

(. )pti*i+e the !inal set o! poses
.
9oarse minimi5ation of all poses.
.
=e-cluster an# rank using more aggressi+e cluster cutoff.
.
Minimi5e 2) lowest energ$ poses more aggressi+el$.
.
Stochastic search of 2) lowest poses to fin# near!$ minima.

ShapeDock
ShapeDock

Translation

=igi#-!o#$ rotation

Torsions

Population of in#i+i#uals

"itness of each is #ocking

score

>ach generation:

Select !ree#ing in#i+i#uals

Mutation

9rosso+er

ocal minimi5ation

>litism

9heck for con+ergence

3enetic Degrees of "ree#om Search proce#ure
!ADock
!ADock
Lamarckian !enetic Algorithm
Lamarckian !enetic Algorithm
#ompare
#ompare
!ADock $ ShapeDock
!ADock $ ShapeDock

=o!ust ? 3eneral

Slow, har# to #efine

con+ergence

7ot repro#uci!le
%Stochastic&

9an get caught in a local

minima

Some ligan#/!in#ing site

t$pes still cause pro!lems

"ast@

=epro#uci!le

"ormall$ e'plores all

minima
3ADock ShapeDock
igan#
4$#rogens a##e#
4$!ri#i5ation an# AScore atom t$pes assigne#

Target
9r$stal waters remain
4$!ri#i5ation an# AScore atom t$pes assigne#

Miscellaneous
Atom charges not reAuire#

All steps are #one automaticall$ insi#e Argusa! %no user inter+ention
reAuire#&. 4owe+er, manual mo#ifications to a!o+e ma$ !e #one if #esire#.

,unning the docking calculation:
Select the ligan# an# !in#ing site. Accept #efault parameters %gri# si5e an#
resolution& or mo#if$ them. =un the #ocking.
Sample %reparation and &un
Sample %reparation and &un
2., 345 Pentium%=& , Dell Bnspiron laptop
ShapeDock' ()pical (imings
ShapeDock' ()pical (imings
Target igan# Torsions Time
%sec&
*4PC C/,DE *, 2*
*4C= /F2G1 E *G
,D"= Methotre'ate H 1
*B>P 3lee+ec D **
*9I/ Ien5$lsuccinate ) )
*STP Iiotin ) 1
1PTI Ien5ami#ine - *

AScore
AScore
an empirical scoring *unction an empirical scoring *unction
AScore is !ase# on terms taken from the 4PScore piece of /Score
J*K
J*K L"urther #e+elopment an# +ali#ation of empirical scoring functions for structure-!ase# !in#ing affinit$ pre#ictionM Wang, =, ai, , an# Wang, S. N. 9omp. Ai#e# Mol.
Design *G, **-2G, 2--2
3
!in#
O 3
+#w
P 3
h$#ropho!ic
P 3
4-!on#
P 3
4-!on# %chg&
P 3
#eformation
P 3
-

3
+#w
O 9
CDW
CDW
3
h$#ropho!ic
O 9
h$#ropho!ic
4P
3
4-!on#
O 9
4-!on#
4I
3
4-!on# %chg-chg ? chg-neutral&
O 9
4-!on#%chg&
4I
3
#eformation
O 9
rotor
=T
3
-
O 9
regression

AScore
AScore
Sum is o+er h$#ropho!ic ligan#-protein atom pairs
f%#
i<
& O *.- # 0 #
i<,-
P -.)Q
O 2/1 %#
-
P 2 R #& #
i<,-
P -.) Q 0 # 0O #
i<,-
P 2.- Q
O - # S #
i<,-
P 2.- Q
-. O
( )

protein
j
ij
ligand
i
d f
/D0 O

>
1
1
]
1

,
_

,
_

+
1
1
]
1

,
_

,
_

ligand
i
ligand
i j
ij
ij
ij
ij
protein
j
ij
ij
ij
ij
ligand
i
r
d
r
d
r
d
r
d
,
- ,
E
- ,
,
- ,
E
- ,
2 2
#
i<,-
is sum of +#W ra#ii of atoms i,<
intra-ligan# CDW e'clu#es *-2, *-1 !on#e# pairs.
AScore
AScore
,1 O

RT
i
O - atom i not in+ol+e# in an$ torsion.
O -.) atom i in+ol+e# in * torsion.
O *.- atom i in+ol+e# in 2 torsions.
O -.) atom i in+ol+e# in S 2 torsions.

ligand
i
i
RT
-2 O

protein
j
ligand
i
HB
i<
-2
i<
O f%r
i<
& f%
*,i<
& f%
2,i<
&
r
i<
#istance !etween #onor/acceptor atoms

*,i<
angle !etween #onor root-#onor-acceptor

2,i<
angle !etween #onor-acceptor-acceptor root
>ach term +aries from *.- to -.-
#epen#ing on how close to i#eal
+alue. Ma'imum num!er of 4-
!on#s per #onor/acceptor atom
impose#.
AScore e'ten#s /Score to allow it to !e use# as the #ocking o!<ecti+e
function.
Separate 4-!on# term in+ol+ing charge# #onor an#/or acceptor groups.
Ma'. num!er of 4-!on#s per #onor/acceptor impose# !$ uniforml$ scaling
total foun# to the ma'imum num!er allowe# for an$ gi+en ligan# atom.
igan# has h$#rogens a##e#.
4$#rogens inclu#e# in the CDW term.
9r$stal waters retaine# %!ut h$#rogens not a##e#&. 4-!on#s with cr$stal
waters treate# as ha+ing i#eal 4-!on# geometr$ !ut with a scaling factor fit to
e'periment.
4-!on#s with target metals treate# as i#eal geometr$, !ut with scaling factor fit
to e'periment.
S4 treate# as 4-!on# #onor/acceptor, SS treate# as 4-!on# acceptor.
Bntra-ligan# CDW energ$ inclu#e#.
AScore
AScore
Di**erences +ith ,Score Di**erences +ith ,Score
.Iegin with the pu!lishe# /Score parameters.
J*K
.Iegin with Wang:s #ata set of *-- protein-ligan# structures.
J2K
.=emo+e incorrect structures to get a final training set of E, structures:
1H h$#rophilic, 2- h$#ropho!ic, 2) mi'e#
.Mo#if$ 4-!on# parameters ? other new parameters to impro+e correlation of score of '-ra$
pose an# e'periment !in#ing free.
J*K L"urther #e+elopment an# +ali#ation of empirical scoring functions for structure-!ase# !in#ing affinit$ pre#ictionM Wang, =, ai, , an# Wang, S. N. 9omp. Ai#e# Mol.
Design *G, **-2G, 2--2
J2K L9omparati+e >+aluation of ** Scoring "unctions for Molecular DockingM =en'iao Wang, Tipin u, an# Shaomeng Wang. J. Med. Chem. 2333, 46, 22ED-21-1
Structure
1ype
Correlation
3
!in#
with
3
e'periment

,4SD
2inding
A!!inity
%kcal/mol&
4$#rophilic -.)1 2.1
4$#ropho!ic -.E, 2.-
Mi'e# -.D- 2.*
All Structures 3.53 2.2

%arameteri-ation $ .alidation
%arameteri-ation $ .alidation
(in progress" (in progress"
Dock the training set using the ShapeDock engine.
Structure
1ype
Correlation
3
!in#
with
3
e'periment

,4SD
2inding
A!!inity
%kcal/mol&
Ave.
,4SD%Q&
4$#rophilic -.,1 2., *.,
4$#ropho!ic -.E- 2.2 *.H
Mi'e# -.G* 2., *.D
All
Structures
3.(# 2.3 1.(

%arameteri-ation $ .alidation
%arameteri-ation $ .alidation
J*K LThe >ffect of Small 9hanges in Protein Structure on Pre#icte# Iin#ing Mo#es of Fnown Bnhi!itors of Bnfluen5a Cirus 7eruamini#ase: PM"-Scoring in Dock,M Bngo
Muegge, Med. Chem. Res. 9, *HHH, ,H--)--.
(rial Stud)'
(rial Stud)'
/n*luen-a .irus 0euraminidase
/n*luen-a .irus 0euraminidase
J*K
.
3l$coprotein en5$me clea+es sialic aci# resi#ues from maturing +irus particles.
.
>le+en conser+e# resi#ues make up the !in#ing site.
.
Dominate# !$ 4-!on#ing ? charge-charge group interactions %e.g. car!o'$l :
guani#ino&
DA7A 3A7A
*--,--- ' increase
in !in#ing affinit$
-*-.2 kcal/mol -**.E kcal/mol ( 1' enhancement
0euraminidase Dockings
0euraminidase Dockings
ShapeDock
ShapeDock
6 o! the 13 structures reproduced the e7peri*ental binding *ode.
Correlation of predicted and measured binding affinities
&
1
2 0.30
Ave. &MSD 2 4.55 Angstroms
641
644
640
67
640 67 68 63 69 65 64 6: 61
log IC50
A
S
c
o
r
e

S
c
o
r
e

(
k
c
a
l
/
m
o
l
)
.
ShapeDock an# 3ADock engines %BDock>ngine interface, Dock>ngine"actor$, etc&.
.
AScore scoring function with mo#ifia!le parameter set %BScore interface&.
.
>as$ to make the ligan# an# !in#ing site groups with one mouse click.
.
Dock ligan# as fle'i!le, rigi#, or using onl$ selecte# torsions.
.
Score current pose, optimi5e current pose, an# full #ocking.
.
Scoring function pre-e+aluate# on a scoring gri#%s&.
.
Data!ase #ocking supports SD" file as ligan# #ata!ase %BDataSource&.
.
>fficient reuse of scoring an# #ocking gri#s allows user to interacti+el$
mo#if$ ligan# or choose new ligan# an# Auickl$ #ock new structures.
.
=esults summari5e# in e'ternal file an# in a tree-+iew. ;ser can click on
poses to +iew #etails.
Docking in ArgusLab 4.0
Docking in ArgusLab 4.0
.
3D interactive *olecule builder 8 viewer
.
Co*putational e7peri*ents
.UM: >'ten#e# 4uckel, Semi-empirical %M7D6, AM*, PM1&, VB7D6, an# ab initio %+ia interface
to 3aussian HE/-1&.
.MM: ;ni+ersal "orce "iel# %;""&, 9C"", AMI>=, custom force fiel#s for research. Polari5a!le
molecular mechanics, =appe ? 3o##ar#:s charge eAuili!ration scheme for ;"".
.3eometr$ optimi5ations, electronic e'cite# states, MD simulations, free-energ$ pertur!ation, an#
potential of mean force.
.UM/MM an# UM/MMpol.
.Molecular Docking.
.
.roperties 8 *isc. #ipole moments, atom-charges, transition properties, surface properties,
animate normal mo#es, +iew #ock poses, ri!!ons, sol+ent-accessi!le surfaces, S9=" sol+ent effects,
e'plicit sol+ent, perio#ic !oun#ar$ con#itions, >wal# sums, etc.
.
4anage9organi+e results: tree+iew tool for e#iting structures an# +iewing results. =esults
an# structures can !e sa+e# in Argusa! /M file.
ArgusLab #apabilities
ArgusLab #apabilities

.
Multi-#ocument interface, multi-threa#e#.
.
Written in 9PP %some ol# legac$ 9-co#e is wrappe# in 9PP&
.
;ses 6pen3 for graphics, Win12 APB for win#owing s$stem.
.
3ar!age collection for graphics o!<ects, e+ents, etc.
.
9ustom hash-ta!les ? containers in a##ition to use of ST.
.
9ustom Mo#el-Ciew-9ontroller %MC9& transport la$er.
.
1D e#itor !uilt on a comman# processor mo#el %support un#o/re#o&.

:nstalled ;ser 2ase
.
(2-,--- #ownloa#s/licenses.
.
Popular in uni+ersit$ teaching programs an# with stu#ents. %free &
.
;se# in se+eral in#ustrial settings.
Arguslab Architecture
Arguslab Architecture
Score the %D;bind Database
Score the %D;bind Database
Score the DEG structures from the PDI!in# #ata!ase
J*K

%*, incorrect structures were remo+e# from the original E-- in #ata!ase&
J*K LThe PDI!in# Data!ase: 9ollection of Iin#ing Affinities for Protein-igan# 9omple'es with Fnown Three-Dimensional StructuresM =en'iao Wang, /ueliang "ang, Tipin
u, an# Showmeng Wang. J. Med. Chem. 233#, 47, 2HDD-2HE-
.D2bind
Database
Correlation
3
!in#
with
3
e'periment

,4SD
2inding
A!!inity
%kcal/mol&
DEG Structures -.,D 2.H