Meteor is a plateform for quantitative metagenomics profiling of complex ecosystems. Meteor relies on genes catalogue to perform species-level taxonomic profiling (Bacteria, Archaea and Eukaryotes), functional analysis and strain-level population structure inference.
Besides python packages dependencies, Meteor requires:
- python 3.11+
- bowtie2 2.3.5+
- freebayes 1.3.6+
Meteor is available with conda which includes all its dependencies:
conda create --name meteor -c conda-forge -c bioconda meteor
Or with pip with a recent Python 3.10+:
pip install meteor
You can test the installation of meteor with:
meteor test
For automated pipeline execution, a Nextflow wrapper nf-meteor.nf is available that streamlines the entire Meteor workflow:
nf-meteor.nf --in <fastq_dir> --catalogue_name <catalogue_name> --out <output_dir> --cpus <nb_cpus> -w <temp_work_dir>
Parameters:
--in Directory containing paired fastq.gz files (default: ).
--out Output directory (default: ).
--cpus Number of cpus to use (default: 4).
--catalogue_name Name of the prebuilt catalogue to use (default: none). Allowed values are: fc_1_3_gut, gg_13_6_caecal, clf_1_0_gut, hs_10_4_gut, hs_8_4_oral, hs_2_9_skin, mm_5_0_gut, oc_5_7_gut, rn_5_9_gut, ssc_9_3_gut
--catalogue Path to a catalogue (overrides --catalogue_name if both are provided).
--fast Enable fast mode for meteor (no functional analysis) (default: false).
--check_catalogue Check md5sum of the catalogue is compatible with the input reads (default: false).
--resistance Variants of the antibiotic resistance genes: meteor mapping --kr, meteor resistance and meteor resistance_summary (complete catalogue only) (default: false).
--resistance_trees With --resistance, one tree per resistance gene (meteor resistance_summary --trees) (default: false).A basic usage of meteor will require to:
- Download or build a reference catalogue
- Structure the raw fastq files
- Map reads against the reference catalogue
- Compute taxonomical and/or functional abundances
- Strain profiling
- Resistance variants (optional: variants of the antibiotic resistance genes)
Meteor requires to download locally a microbial gene catalogue specif, either in 'full' or 'light' version. The 'full' version contains all genes of the catalogue, whereas the 'light' version contains only the marker genes that will be used to infer species abundance profiles. Of note, no functional profiling can be performed when using the 'light' version of a catalogue.
Ten catalogues are currently available:
| Microbial gene catalogue | <name> | Genes count (M) | Metagenomic Species Pan-genomes (MSPs) | Size (full) (GB) | Size (light) (GB) | Description |
|---|---|---|---|---|---|---|
| Canis lupus familiaris | clf_1_0_gut | 0.95 | 234 | 1.4 | 0.1 | link |
| Felis catus | fc_1_3_gut | 1.3 | 344 | 2.0 | 0.2 | link |
| Gallus gallus domesticus | gg_13_6_caecal | 13.6 | 2420 | 19.6 | 1.2 | link |
| Homo sapiens gut | hs_10_4_gut | 10.4 | 1990 | 12.6 | 0.7 | link |
| Homo sapiens oral | hs_8_4_oral | 8.4 | 853 | 13.7 | 0.5 | link |
| Homo sapiens skin | hs_2_9_skin | 2.9 | 392 | 3.9 | 0.2 | link |
| Mus musculus | mm_5_0_gut | 5.0 | 1252 | 10.3 | 0.6 | link |
| Oryctolagus cuniculus | oc_5_7_gut | 5.7 | 1053 | 8.0 | 0.4 | link |
| Rattus norvegicus | rn_5_9_gut | 5.9 | 1627 | 7.0 | 0.6 | link |
| Sus domesticus | ssc_9_3_gut | 9.3 | 1523 | 11.3 | 0.7 | link |
These references can be downloaded with the following command:
meteor download -i <name> -c -o <refdir>
The 'light' catalogues are available with the tag (--fast) :
meteor download -i <name> -c --fast -o <refdir>
Meteor requires a first of fastq indexing:
meteor fastq -i <fastqdir> [-p paired reads] -o <outputdir>
When multiple sequencing are available for a library, the option -m allows to group these samples. Example:
Illumina_lib1-SAMPLE_01.fastq
Illumina_lib1-SAMPLE_02.fastq
Illumina_lib2-SAMPLE_01.fastq
Illumina_lib2-SAMPLE_02.fastq
In this case, the following command will group these samples the same library:
meteor fastq -i ./ -m SAMPLE_\\d+ -o outputdir
The fastq files are mapped against a catalogue to generate a gene count table with the following command:
meteor mapping -i <fastqdir/sampledir> -r <refdir> -o <mappingdir>
We recommend to first filter out reads with low-quality, length < 60nt or belonging to the host.
Genes from the catalogue are clustered into Metagenomic Species Pangeomes (MSP) with MSPminer, and are functionnaly annotated against KEGG r107, DBcan (carbohydate active enzymes) and MUSTARD (antibiotic resistant determinants).
MSP and functional profiles are computed from the gene count table with the following command:
meteor profile -i <mappingdir/sampledir> -o <profiledir> -r <refdir> -n coverage
The "-n" parameter ensures read count normalization for gene length. If omitted, no normalization will be performed on the gene table.
This profiling step will generate:
- a Species abundance table;
- an ARD abundance table (full catalogue only);
- a DBCAN abundance table (full catalogue only);
- a Gut Metabolic Modules (GMM) abundance table (from the KO annotation) (full catalogue only).
- a Gut Brain Modules (GBM) abundance table (from the KO, EGGNOG and TIGRFAM annotations) (full catalogue only).
To merge output from different samples into a single table, use the following command:
meteor merge -i <profiledir> -r <refdir> -o <mergingdir>
All tables are merged by default, except the gene tables (-g). -t merges only the tables given, e.g. the species table alone to look for cross-sample contamination with CroCoDeEL:
meteor merge -i <profiledir> -r <refdir> -o <mergingdir> -t msp
Tables: msp (species), genes (gene counts, same as -g), modules_completeness, modules, and the annotation tables mustard, resfinder, resfinderfg, dbcan, kegg (each _as_genes_sum and _as_msp_sum). Functional tables need a complete catalogue.
-l <samples.txt> merges only the samples named in a text file, one sample name per line (first column if the file is tab-separated; empty lines and lines starting with # are ignored), in the order of the file. Samples of the list that are not found are reported in the log.
meteor merge also writes a report of the merged samples:
<prefix>_qc.tsv: one row per sample with its reads (total_read_count), the reads mapped and counted on the catalogue genes (mapped_reads,mapped_read_ratein %), the number of MSPs detected (msp_richness) and the % of the gene signal carried by the MSPs (msp_signal_rate);<prefix>_phylum.tsv: relative abundance (%) of each phylum in each sample, from the species table before the-a/-nfilters (when the species table is merged);<prefix>_*_mqc.json: the same data as MultiQC custom content (general statistics, mapped reads, MSP richness against the number of reads, phylum composition).
MultiQC is not a dependency of meteor: install it separately (pip install multiqc or conda install -c bioconda multiqc), then build the HTML report from the merging directory:
multiqc <mergingdir> -o <reportdir>
MultiQC finds the *_mqc.json files and writes <reportdir>/multiqc_report.html with a "Meteor" section (mapped reads, MSP richness, phylum composition) and the meteor columns of the "General Statistics" table. -n <name>.html names the report and -f overwrites an existing one. To read only the meteor files (faster when the merging directory holds large tables such as the gene table), give them directly: multiqc <mergingdir>/*_mqc.json -o <reportdir>. Other directories given to MultiQC (fastp, bowtie2 logs, ...) are added to the same report:
multiqc <mergingdir> <fastpdir> -o <reportdir>
With -m, the species abundance table is also saved in the style of a merged MetaPhlAn table (<prefix>_msp_mpa.tsv), for the tools that read MetaPhlAn profiles. Each row is a clade from superkingdom (k__) to MSP (t__), named by its lineage (k__Bacteria|p__Pseudomonadota|...|s__Escherichia_coli|t__msp_0005). Values are relative abundances in percent of the sample: the MSPs written in <prefix>_msp.tsv (after the -a, -n and -s filters) sum to 100, and a clade is the sum of its MSPs, so each rank sums to 100. Ranks missing from the catalogue taxonomy are named unclassified. --tax_lev keeps a single rank (k, p, c, o, f, g, s or t). The first line names the meteor version, the catalogue and the GTDB release of its taxonomy.
Meteor is capable of profiling strains in large metagenomic datasets. It identifies specific mutations from strains and applies them to the gene catalog MSPs.
To use Meteor for strain profiling, use the following command:
meteor strain -i <mappingdir/sampledir> -o <straindir> -r <refdir>
Meteor computes mutation rates and trees between strains from samples using a GTR+GAMMA model with the following command:
meteor tree -i <straindir> -o <treedir>
-l <samples.txt> builds the trees with only the samples of the list (same format as for meteor merge). An MSP carried by none of the samples of the list gets no tree; when no sample of the list is in the strain directory, meteor tree warns and builds no tree. When the output directory already holds the tree of an MSP computed from other samples, it is recomputed.
This profiling step will generate:
- a mutation rate matrix,
<msp>.tsv: the tip-to-tip distances of the UPGMA tree; - the distances of the model (GTR, or TN93 with
-r) before UPGMA,<msp>_raw_distances.tsv: a square matrix of the samples, in the order of the alignment; - a fasta file of each strain for each sample;
- a table giving detailed comparison per strain. The file is a tab-separated values (TSV) file with one row per sample pair. Each row contains the following columns:
| Column | Description |
|---|---|
sample1 |
First sample in the comparison |
sample2 |
Second sample in the comparison |
total_length |
Total length of the alignment (in bases) |
overlap_noN_info_count |
Number of positions where both samples have minimal information (A,C,G,T or IUPAC codes excluding N, gaps, and ?) |
overlap_noIUPAC_info_count |
Number of positions where both samples have maximal information (strictly A,C,G,T only) |
overlap_noN_info_pc |
Percentage of total_length with minimal information overlap |
overlap_noIUPAC_info_pc |
Percentage of total_length with maximal information overlap |
noN_info_pc_sample1 |
In sample1, percentage of positions with minimal information |
noN_info_pc_sample2 |
In sample2, percentage of positions with minimal information |
noIUPAC_info_pc_sample1 |
In sample1, percentage of positions with maximal information |
noIUPAC_info_pc_sample2 |
In sample2, percentage of positions with maximal information |
distance |
Genetic distance between samples (0.0 = identical) |
distance_category |
Categorical classification: same_strain, same_subspecies, or divergent |
Samples are automatically classified based on their genetic distance:
| Category | Distance Threshold | Approximate Similarity | Biological Interpretation |
|---|---|---|---|
same_strain |
≤ 0.0001 | ≥ 99.99% | Same strain/clone |
same_subspecies |
≤ 0.015 | ≥ 97% | Same subspecies |
divergent |
> 0.015 | < 97% | Different lineage |
Note:
- Maximal Information: Unambiguous nucleotides (A, C, G, T only)
- Minimal Information: Includes IUPAC ambiguity codes (R, Y, S, W, K, M, B, D, H, V) but excludes N, gaps (-), and unknown (?)
Samples sitting on abnormally long branches of a strain tree (rogue taxa) inflate their distances to every other sample. meteor tree --treeshrink runs the per-gene test of TreeShrink (Mai & Mirarab, BMC Genomics 2018) on each MSP tree and writes, next to the tree:
<msp>_qcflags.tsv: for each sample of the best removal sets, its diameter ratio (largest reduction of the tree diameter obtained by removing it), the threshold of the MSP (log-lognormal, alpha = 0.05) and whether it is flagged;<msp>_shrink.tsv: the distance matrix without the flagged samples (only when samples are flagged).
At the end of meteor tree, the comparison tables of all MSPs are gathered into three tables:
strain_sharing_events.tsv: one row per strain-sharing event, i.e. a sample pair sharing the strain of an MSP (msp_name,sample1,sample2,relation,distance,overlap_noN_info_pc,threshold);strain_sharing_samples.tsv: one row per sample pair, with its strain-sharing rate (sharing_rate: fraction of the MSPs compared between the two samples whose strain they share) and the numbers it is computed on (n_mspcompared,n_msp_shared);strain_sharing_msp.tsv: one row per MSP, with itstransmissibility(fraction of the compared pairs of samples of different individuals that share its strain) and the numbers it is computed on (n_sample_pairs,n_shared_pairs), and the pairs of samples of the same individual (n_same_individual_pairs,n_same_individual_shared).
Every rate is given with the number of pairs or MSPs it is computed on, however small: filter on these numbers rather than on the rate alone.
Two samples are compared on an MSP when the sites informative (not N) in both cover at least --min_overlap % of its alignment (default 10): a pair overlapping on a few sites only is neither compared nor shared. They share its strain when their distance is at most 0.0001 (99.99 % identity, the same_strain category of the comparison tables), the same threshold for every MSP.
--ind <individuals.tsv>, a tab-separated file giving the individual of each sample (sample, individual), marks the pairs of samples of the same individual (time points of a subject): they are listed (relation = same_individual) but left out of the transmissibility, since sharing a strain with oneself is not a transmission. Samples of different individuals (different_individuals), mother and infant included, count. Without --ind, every sample is a distinct individual.
A tree directory computed in GTR by meteor 2.0.22 or earlier has wrong distances in its comparison tables (issue #133): meteor tree recomputes them when it runs again in that directory.
The tree and the distance matrix are not modified.
This mode is dedicated to the variants (single nucleotide variants) of the antibiotic resistance genes, not to their presence or abundance:
- the abundance of the resistance genes (the resistome) is given by the standard workflow, in the
mustard,resfinderandresfinderfgtables ofmeteor merge, with no extra option; meteor resistancetells, for each resistance gene with reads, how much the gene of the sample differs from the catalogue gene (major variants), how polymorphic it is in the sample (minor variants), whether these variants change the protein, and how this compares with the marker genes of the same species;- these variants are not resistance mutations: the catalogue gene is a cluster representative, not a susceptible reference, so a variant is a difference; the known resistance positions of the target genes are read separately, when the catalogue has them (see below).
meteor resistance counts the single nucleotide variants of the antibiotic resistance genes of a sample against the catalogue genes, with the marker genes of the same species as controls. The resistance genes are the catalogue genes annotated by ResFinder, ResFinderFG or MUSTARD (acquired genes), and the target genes, resistant by point mutations or inactivation: the genes of the KEGG orthologs of gyrA (K02469), gyrB (K02470), parC (K02621) and parE (K02622) (quinolones), rpoB (K03043) (rifampicin), rpsL (K02950) (streptomycin), folP (K00796) (sulfonamides), folA (K00287) (trimethoprim), murA (K00790), glpT (K02445) and uhpT (K07784) (fosfomycin), nfsA (K10678) and nfsB (K10679) (nitrofurans) and basR (K07771), basS (K07643), phoP (K07660), phoQ (K07637) and adeS (K18143, the CrrB sensor of Klebsiella) (colistin), and the genes of the resistance_targets annotation of the catalogue, target genes without a KEGG ortholog found by alignment on a reference protein (gene_id, family, e.g. mgrB of Klebsiella pneumoniae). The sample must be mapped with --kr, which writes the filtered alignments on the resistance genes to <sample>_resistance.cram, next to <sample>.cram, the alignments on the marker genes used as controls (--kr implies --kf; <sample>.cram is the same as with --kf alone). A resistance gene missing from <sample>_resistance.cram (a sample mapped before some of the targets were added to --kr) is read from <sample>.cram when it is a marker gene, and has no reads otherwise (genes_read_from_markers in the census):
meteor mapping -i <fastqdir/sampledir> -o <mappingdir> -r <refdir> --kr
meteor resistance -i <mappingdir/sampledir> -o <resistancedir> -r <refdir>
The bases of every alignment kept by the mapping step (best identity, a multi-mapped read on each of its genes) are counted at each position of the gene (base quality >= -q, default 20). A site is covered when at least -p reads (default 10) cover it. On covered sites:
- a major variant is a site whose most frequent base is not the catalogue base, with at least
-areads (default 3): the gene of the sample differs from the catalogue gene; on an evenly split site the catalogue base wins; - a minor variant is a polymorphic site, whose second most frequent base has at least
-areads and a frequency of at least-f(default 0.1): several alleles in the sample.
A variant is synonymous or not by comparing the two alleles in the codon of the catalogue gene (genes are CDS in frame); synonymous and non-synonymous sites are counted on the fully covered codons (Nei & Gojobori).
<resistancedir>/<sample>/<sample>_resistance.tsv has one row per resistance gene with reads:
| Column | Description |
|---|---|
gene_id, msp_name |
Catalogue gene and the MSPs it belongs to (empty for a gene of no MSP) |
resfinder, resfinderfg, mustard, mustard_prediction, target |
Annotations of the gene (target: gyrA, gyrB, parC, parE, rpoB, rpsL, folP, folA, murA, glpT, uhpT, nfsA, nfsB, basR, basS, phoP or phoQ) |
gene_length, alignments, multimapped_fraction |
Alignments on the gene and the fraction with a MAPQ <= 1 (mostly reads with equally good hits on other genes; it also holds rare unique alignments of low MAPQ) |
mean_depth, covered_bp, covered_fraction, interpretable |
Coverage; a gene is interpretable when covered_fraction >= -c (default 0.8) |
major_snv, major_syn, major_nonsyn, minor_snv, minor_syn, minor_nonsyn |
Variants |
syn_sites, nonsyn_sites |
Synonymous and non-synonymous sites of the covered codons |
major_per_kb, minor_per_kb |
Variants per kb covered |
major_dnds, minor_pnps |
(non-synonymous variants / non-synonymous sites) / (synonymous variants / synonymous sites), against the catalogue (major) and within the sample (minor); a simple count ratio, not a model-based dN/dS; NA without synonymous variant |
control_genes, control_covered_bp, control_major_per_kb, control_minor_per_kb |
The same rates on the interpretable marker genes (--core_size) of the MSPs of the gene, without the gene itself; the controls of a target gene hold the other target genes of its MSP. A gene of no MSP (often a mobile gene) has no controls |
reads_resistant, reads_wt, reads_other, reads_linked |
Reads of the gene at its known resistance positions (NA without them), read by read: carrying a resistance residue at one position at least / reading a codon without resistance residue / reading no codon of a known position / carrying a resistance residue at two positions at least (linked substitutions, e.g. gyrA S83L + D87N) |
The catalogue gene is the representative of a gene cluster, not a susceptible reference: a variant against the catalogue is not a resistance mutation. Close homologues share their reads (multimapped_fraction), and reads diverging from a gene by more than the mapping identity threshold are not aligned on it.
<resistancedir>/<sample>/<sample>_resistance_consensus.fasta.xz holds the consensus sequence of each interpretable resistance gene in the sample (>gene_id): the major base of each covered site, N elsewhere, on the coordinates of the catalogue gene. A sample carrying several alleles of a gene (minor variants) gets the major one at each site.
When the catalogue has a resistance_positions annotation, <resistancedir>/<sample>/<sample>_resistance_positions.tsv gives the residue carried by the reads at the known resistance positions of the target genes with reads. The annotation is built outside of meteor, one row per catalogue gene x known resistance position (e.g. the union of PointFinder and AMRFinderPlus, the known resistance substitutions of the reference species transferred onto the gene):
| Column | Description |
|---|---|
gene_id |
Catalogue gene of the row |
family |
Family of the target gene (gyrA, gyrB, parC, parE, rpoB, rpsL, folP, ...) |
codon_pos |
Position in the numbering of the reference (e.g. gyrA 83 of E. coli) |
ref_residue |
Residue of the reference at this position |
resistance_residues |
Known resistance residues, comma-separated |
gene_aa_pos |
Position of the codon in the catalogue gene (amino acid, 1-based) |
gene_nt_pos |
First base of the codon in the catalogue gene, 1-based |
gene_residue |
Residue of the catalogue gene at this position |
window_identity |
Identity of the alignment of the gene on the reference around the position |
source, residue_sources, residue_classes |
Optional: the origin of the position (PointFinder, AMRFinderPlus), the origin of each resistance residue, and the antibiotic class of each resistance residue (L:QUINOLONE;W:QUINOLONE/AMINOCOUMARIN) |
The optional columns, when they are in the annotation, are carried as is to <sample>_resistance_positions.tsv and to the merged tables of meteor resistance_summary.
The residues are counted read by read: a read counts when the 3 bases of the codon are aligned without indel, with a base quality >= -q. One row per position read by at least one read:
| Column | Description |
|---|---|
gene_id, msp_name, family, codon_pos, ref_residue, resistance_residues, gene_aa_pos, gene_nt_pos, gene_residue, window_identity, residue_classes |
The position, from the annotation (residue_classes when it is in the annotation) |
depth, readable |
Reads carrying the codon; the position is read when depth >= -p |
major_residue, major_frequency, major_is_resistance |
Residue of most reads, its frequency, and whether it is a resistance residue |
resistance_reads, resistance_frequency |
Reads carrying a resistance residue, and their fraction |
known_resistance |
The position is read and at least -a reads and -f of the reads carry a resistance residue (a resistant minor population counts) |
residue_counts |
Reads per residue (S:16,L:4) |
The positions come from the reference species (E. coli, K. pneumoniae): in another species, the wild-type residue can be a "resistance" residue. meteor resistance_summary compares each sample with the residue of the cohort.
The abundance of the resistance genes is given by meteor merge (mustard, resfinder, resfinderfg tables): meteor resistance only adds their variants. The tables of several samples are merged, and with --trees one tree is built per resistance gene, by:
meteor resistance_summary -i <resistancedir> -o <mergedir> [--trees]
Every sample must have been analysed against the same catalogue with the same parameters. -l <samples.txt> merges only the samples of the list, in its order (same format as for meteor merge). It writes in <mergedir>:
<prefix>_resistance.tsv: the tables of every sample, one row per sample and gene with reads (samplecolumn first);<prefix>_resistance_major_per_kb.tsv,<prefix>_resistance_minor_per_kb.tsv: gene x sample matrices of the variants per kb covered,NAwhere the gene is not interpretable;<prefix>_resistance_interpretable.tsv: gene x sample matrix, 1 where the gene is interpretable;<prefix>_resistance_genes.tsv: one row per gene, with its annotations and MSPs, the number of samples where it has reads (samples_with_reads) and where it is interpretable (samples_interpretable;interpretable_fractiondivides it by the number of merged samples), the medians of its rates and of the rates of its controls over the samples where it is interpretable (median_major_per_kb,median_minor_per_kb,median_control_major_per_kb,median_control_minor_per_kb), the number of these samples where it has controls (samples_with_controls), and the fraction of them where its rate is strictly higher than that of its controls (major_above_controls_fraction,minor_above_controls_fraction);- with the known resistance positions,
<prefix>_resistance_positions.tsv: the position tables of every sample (samplecolumn first), with the residue of the cohort at each position (cohort_residue: the individual residue of most of the individuals reading the position, the individual residue of a sample being the residue of its reads over the genes of the same family in the same MSP, weighted by depth, as several catalogue genes can be alleles of the gene of one species; a gene of no MSP alone;cohort_samples: the individuals where the position is read;cohort_share: the fraction of them carrying the cohort residue, to filter the near-even splits between two lineages) anddiffers_from_cohort; - with the known resistance positions,
<prefix>_resistance_positions_samples.tsv: per sample and target family, the genes read at one known position at least (genes_readable), and among them the genes with a resistance residue at one position at least (genes_resistance,known_resistance: minor populations included), as the major residue (genes_major_resistance), and as a major residue different from the cohort residue (genes_acquired: candidate acquired mutations, a resistance residue that is the wild type of the species being the cohort residue), with the fractionsfraction_resistance,fraction_major_resistanceandfraction_acquired(divided bygenes_readable); and the reads of the family genes at the known positions, summed over the genes (reads_resistant,reads_wt,reads_other,reads_linkedas in<sample>_resistance.tsv) withresistant_reads_fraction(the resistant reads over the reads reading a position at least); when the annotation hasresidue_classes, the same counts and fractions per antibiotic class (genes_<class>_readable: the genes read at one position of the class at least,genes_<class>_major_resistance,genes_<class>_acquired,fraction_<class>_major_resistance,fraction_<class>_acquired), so that a family carrying positions of several classes (gyrB: quinolones and aminocoumarins) is no longer mixed; <prefix>_resistance_census.json: the catalogue, the parameters and the numbers of samples, genes and trees.
The genes are the catalogue genes (gene_id), as in the tables of meteor merge.
With --trees, every resistance gene interpretable in at least -m samples (default 4) gets a tree of these samples, built from their consensus sequences as meteor tree builds the tree of an MSP (same site filter -g, minimum number of sites -s, GTR or TN93 distance -r, UPGMA). The sequences share the coordinates of the catalogue gene, so no alignment is needed; one tree is built per catalogue gene, not per family. The tree tells which samples carry the same allele of the gene; compared with the tree of the MSP carrying it (meteor tree), a discordance points to a transfer of the gene between strains. Close homologues share their reads (multimapped_fraction), which can bring alleles closer. In <mergedir>/<prefix>_resistance_trees/:
gene_<gene_id>.fasta: the consensus sequences of the samples;gene_<gene_id>.tree,gene_<gene_id>.tsv(tip-to-tip distances),gene_<gene_id>_raw_distances.tsv(distances of the model before UPGMA),gene_<gene_id>_comparison.tsv(pairs of samples), as formeteor tree.
<prefix>_resistance_trees.tsv indexes them: one row per gene with gene_id, family (the first annotation among ResFinder, ResFinderFG, MUSTARD and the target gene), samples, msp_name, the annotations, and tree (NA when the gene has fewer sites than -s). A tree is recomputed only when the sequences of its gene change.
The options of every command are also listed by meteor <command> -h.
| Option | Description | Default |
|---|---|---|
-i USER_CHOICE |
Select the catalogue to download. Choices: fc_1_3_gut, gg_13_6_caecal, clf_1_0_gut, hs_10_4_gut, hs_8_4_oral, hs_2_9_skin, mm_5_0_gut, oc_5_7_gut, rn_5_9_gut, ssc_9_3_gut. |
required |
--fast |
Select the short catalogue version (only for taxonomic profiling). | off |
-o REF_DIR |
Directory where the downloaded catalogue is saved. | required |
-c |
Check the md5sum of the catalogue after download. | off |
| Option | Description | Default |
|---|---|---|
-i INPUT_FASTQ_DIR |
Directory containing all input fastq files with .fastq or .fq. extensions (gzip, bzip2 and xz compression accepted). Paired-end files must be named : file_R1.[fastq/fq] & file_R2.[fastq/fq] or file_1.[fastq/fq] & file_2.[fastq/fq] | required |
-p |
Fastq files are paired. | off |
-m MASK_SAMPLE_NAME |
Regular expression (between quotes) for extracting sample name. | none |
-o FASTQ_DIR |
Directory where the fastq repository is created. | required |
| Option | Description | Default |
|---|---|---|
-i FASTQ_DIR |
Directory corresponding to the sample to process. (contains sequencing metadata files ending with _census_stage_0.json) | required |
-r REF_DIR |
Directory corresponding to the gene catalog against which reads are mapped. (contains a file ending with *_reference.json) | required |
-o MAPPING_DIR |
Directory where mapping and raw gene counts of the sample are saved. | required |
-p MAPPING_TYPE |
Strategy to map reads against the catalogue. Choices: end-to-end, local. |
end-to-end |
--trim TRIM |
Trim reads exceeding TRIM bases before mapping. If 0, no trim. | 80 |
--align ALIGNMENT_NUMBER |
Maximum number alignments to report for each read. | 100 |
-c COUNTING_TYPE |
Strategy to calculate raw gene counts. Choices: total, smart_shared, strict_shared, unique. smart_shared splits each multi-mapped read between its genes in proportion to their unique reads (evenly when none has unique reads); strict_shared does the same but does not count the multi-mapped reads reaching only genes without unique reads. |
smart_shared |
--core_size CORE_SIZE |
Number of signature genes per species (MSP) used to estimate their respective abundance. | 100 |
--id IDENTITY_THRESHOLD |
Keep only the read alignments with a nucleotide identity >= IDENTITY_THRESHOLD. If 0.0, no filtering. | auto: 0.95 (complete catalogue), 0.98 (taxonomic catalogue) |
--ka |
Keep raw bowtie2 output in cram format. Required for calculating gene counts with another strategy. | off |
--kf |
Keep filtered alignments on marker genes. Required for strain analysis. | off |
--kr |
Keep filtered alignments on antibiotic resistance genes (ResFinder, ResFinderFG, MUSTARD annotations, and gyrA, gyrB, parC, parE, rpoB, rpsL, folP, folA, murA, glpT, uhpT, nfsA, nfsB, basR, basS, phoP, phoQ and adeS from the KEGG annotation, and the genes of the resistance_targets annotation) in <sample>_resistance.cram; implies --kf. Required for meteor resistance. |
off |
--tmp TMP_PATH |
Directory where temporary files (e.g. cram) are stored | none |
-t THREADS |
Number of alignment threads to launch. | 1 |
| Option | Description | Default |
|---|---|---|
-i MAPPED_SAMPLE_DIR |
Directory with raw gene counts of the sample to process. (contains a metadata file ending with _census_stage_1.json) | required |
-r REF_DIR |
Directory corresponding to the catalog used to generate raw gene counts. (contains a file ending with *_reference.json) | required |
-o PROFILED_SAMPLE_DIR |
Directory where species and functional abundance tables of the sample are saved. | required |
-l RAREFACTION_LEVEL |
Rarefaction level. If <= 0, no rarefation is performed. | 0 |
--seed SEED |
Seed of the random number generator used for rarefaction. | 1234 |
-n NORMALIZATION |
Normalization applied to raw gene counts. Choices: coverage, fpkm, raw. |
coverage |
-c COVERAGE_FACTOR |
Multiplication factor for coverage normalization. | 100.0 |
--core_size CORE_SIZE |
Number of signature genes per species (MSP) used to estimate their respective abundance. | 100 |
--msp_filter MSP_FILTER |
Minimal proportion of core genes detected in a sample to consider a species (MSP) as present. | auto: 0.1 (complete catalogue), 0.2 (taxonomic catalogue) |
--completeness COMPLETENESS |
Cutoff above which a module is considered as present in a species. Value between 0.0 and 1.0. | 0.9 |
| Option | Description | Default |
|---|---|---|
-i PROFILE_DIR |
Directory containing subdirectories (one per sample) with abundance tables to be merged. (each subdirectory contains a metadata file ending with _census_stage_2.json) | required |
-r REF_DIR |
Directory corresponding to the gene catalog used to generate the abundance tables. (contains a file ending with *_reference.json) | required |
-a MIN_MSP_ABUNDANCE |
Minimum msp abundance. | 0.0 |
-n MIN_MSP_OCCURRENCE |
Report only species (MSPs) occuring in at least n samples. | 1 |
-s |
Remove samples with no detected species (MSPs). | off |
-m |
Save the merged species abundance table in the style of MetaPhlAn: relative abundances (%) of the clades from superkingdom (k__) to MSP (t__). | off |
-b |
Save the merged species abundance table in biom format. | off |
--tax_lev MPA_TAXONOMIC_LEVEL |
Taxonomic level of the MetaPhlAn-style table (-m): a all levels, k kingdoms, p phyla, c classes, o orders, f families, g genera, s species, t MSPs only. |
a |
-o MERGING_DIR |
Directory where the merged abundance tables are saved. | required |
-p PREFIX |
Prefix added to output filenames. | output |
-g |
Merge gene abundance tables. | off |
-t TABLE [TABLE ...] |
Merge only these tables: msp (species), genes (gene counts, same as -g), modules_completeness, modules, and the annotation tables mustard, resfinder, resfinderfg, dbcan, kegg (each _as_genes_sum and _as_msp_sum). Functional tables need a complete catalogue. |
all but genes |
-l SAMPLE_FILE |
Merge only the samples of this file, in its order: one sample name per line (first column if tab-separated; empty lines and lines starting with # ignored). | none |
| Option | Description | Default |
|---|---|---|
-i MAPPED_SAMPLE_DIR |
Path to the mapped sample directory. | required |
-r REF_DIR |
Path to reference directory (Path containing *_reference.json) | required |
-d MAX_DEPTH |
Maximum read depth taken in account at each position. | 100 |
-p MIN_DEPTH |
Minimum read depth of a site: sites covered by fewer reads are masked in the consensus. Values should be comprised between 1 and the maximum depth (-d). |
3 |
-s MIN_SNP_DEPTH |
Minimum read depth of a site to call a variant. Values should be comprised between 1 and the maximum depth (-d), and be >= the minimum depth (-p). |
3 |
-f MIN_FREQUENCY |
Minimum allele frequency to call a variant. | 0.1 |
-l PLOIDY |
Ploidy given to the variant caller. | 1 |
-m MIN_MSP_COVERAGE |
Minimum percentage (1 to 100) of the signature genes of an MSP passing -c to build its strain. |
80 |
-c MIN_GENE_COVERAGE |
Minimum fraction of a signature gene covered by reads, from 0 to 1. | 0.5 |
--core_size CORE_SIZE |
Number of signature genes per species (MSP) used to estimate their respective abundance. | 100 |
-o STRAIN_DIR |
Path to output directory. | required |
--kc |
Keep the consensus of the marker genes, needed to run meteor strain again on the sample. | off |
--tmp TMP_PATH |
Path to the directory where temporary files are stored | none |
-t THREADS |
Number of threads to perform variant calling. | 1 |
| Option | Description | Default |
|---|---|---|
-i MAPPED_SAMPLE_DIR |
Path to the mapped sample directory (meteor mapping --kr). | required |
-r REF_DIR |
Path to reference directory (Path containing *_reference.json) | required |
-o RESISTANCE_DIR |
Path to output directory. | required |
-p MIN_DEPTH |
Minimum read depth of a covered site. | 10 |
-a MIN_ALT_COUNT |
Minimum number of reads carrying a variant allele. | 3 |
-f MIN_FREQUENCY |
Minimum frequency of a minor allele. | 0.1 |
-c MIN_GENE_COVERAGE |
Minimum fraction of a gene covered at the minimum depth for the gene to be interpretable, from 0 to 1. | 0.8 |
-q BASE_QUALITY |
Minimum base quality of a counted base. | 20 |
--core_size CORE_SIZE |
Number of signature genes per species (MSP) used as controls, as at the mapping step. | 100 |
--tmp TMP_PATH |
Path to the directory where temporary files are stored | none |
-t THREADS |
Number of threads. | 1 |
| Option | Description | Default |
|---|---|---|
-i RESISTANCE_DIR |
Directory containing the sample directories of meteor resistance. | required |
-o MERGING_DIR |
Directory where the merged tables (and the trees) are saved. | required |
-p PREFIX |
Prefix added to output filenames. | output |
-l SAMPLE_FILE |
Merge only the samples of this file, in its order: one sample name per line (first column if tab-separated; empty lines and lines starting with # ignored). | none |
--trees |
Build one tree per resistance gene from the consensus sequences of the samples where it is interpretable. | off |
-m MIN_SAMPLES |
Minimum number of samples where a gene is interpretable to build its tree. | 4 |
-g MAX_GAP |
Removes sites constitued of >= cutoff gap character. | 1.0 |
-s MIN_INFO_SITES |
Minimum number of informative sites in the alignment. | 10 |
-r |
Compute tn93 model (faster). | GTR |
--tmp TMP_PATH |
Path to the directory where temporary files are stored | none |
-t THREADS |
Number of threads. | 1 |
| Option | Description | Default |
|---|---|---|
-i STRAIN_DIR |
Path to the strain directory. | required |
-g MAX_GAP |
Removes sites constitued of >= cutoff gap character. | 1.0 |
-s MIN_INFO_SITES |
Minimum number of informative sites in the alignment. | 10 |
-r |
Compute distances with the TN93 model (faster) instead of GTR. | GTR |
--treeshrink |
Flag the samples on abnormally long branches of each strain tree (per-gene test of TreeShrink): writes _qcflags.tsv and _shrink.tsv. | off |
-l SAMPLE_FILE |
Build the trees with only the samples of this file: one sample name per line (first column if tab-separated; empty lines and lines starting with # ignored). | none |
--ind INDIVIDUALS_FILE |
Tab-separated file of the individual of each sample (sample, individual). Pairs of samples of the same individual (time points) are left out of the transmissibility of the MSPs (default: every sample is a distinct individual). | none |
--min_overlap MIN_OVERLAP |
Minimum % of the MSP alignment informative (not N) in both samples for the pair to be counted in strain sharing. | 10.0 |
-f FORMAT |
Output image format of the trees. Choices: png, svg, pdf, txt. |
no image |
-w WIDTH |
Output image width. | 500 |
-H HEIGHT |
Output image height. | 500 |
-o TREE_DIR |
Path to output directory. | required |
--tmp TMP_PATH |
Path to the directory where temporary files are stored | none |
-t THREADS |
Number of threads when infering each tree. | 1 |
Please cite the following publication if you use Meteor2:
Accurate profiling of microbial communities for shotgun metagenomic sequencing with Meteor2.
Amine Ghozlane, Florence Thirion, Florian Plaza Oñate, Franck Gauthier, Emmanuelle Le Chatelier, Anita Annamalé, Mathieu Almeida, Stanislav D. Ehrlich, Nicolas Pons.