Genomes and metagenomes · What you can do here
Recover genomes from a metagenome
Recover metagenome-assembled genomes from your contigs and read cohort using Bowtie 2, samtools, and MetaCAT, with lineage completeness and contamination checked by BUSCO.
What this involves
Metagenome binning groups assembled contigs into discrete metagenome-assembled genomes (MAGs) based on sequence composition and differential read depth across samples. Reconstructing genomes from microbial communities allows you to study uncultivated organisms, profile strain variations, and link functional capacities to specific taxa. To maximise the power of differential coverage, you should bin a cohort by mapping every sample's reads against a single shared assembly rather than binning individual samples separately.
The procedure requires a pre-computed assembly FASTA and quality-controlled, host-depleted reads for every sample in the cohort. Bowtie 2 aligns each sample's reads against the shared assembly, and samtools sorts and indexes the resulting alignments. The agent then runs MetaCAT to identify single-copy marker gene seeds, calculate pooled contig coverage depths across the BAM files, and cluster the contigs into candidate genomes.
A bin is not automatically a viable genome. Assembly fragmentation can lower completeness metrics without clustering being at fault, while chimeric bins combine DNA from multiple organisms. Each resulting bin is therefore evaluated using BUSCO in genome mode to assess single-copy marker completeness and duplication, the latter serving as a direct signal of contamination.
How it goes
- 1Cross-sample read alignmentBowtie 2 builds an index of the assembly FASTA and aligns paired short reads from each sample, piping the output to samtools to generate coordinate-sorted BAM files.
- 2Marker-gene seed identificationMetaCAT scans the assembly FASTA with its seed command to identify essential single-copy marker genes that anchor individual genome bins.
- 3Cohort coverage profilingMetaCAT processes the sorted BAM alignments across all samples concurrently using its coverage command to build a pooled, compressed coverage profile.
- 4Contig clusteringMetaCAT clusters contigs based on sequence composition, marker seeds, and cross-sample coverage profiles, writing individual FASTA files for each bin alongside a contig mapping file.
- 5Quality evaluationBUSCO evaluates each recovered bin against bacterial lineage datasets to determine marker completeness and duplication rates according to MIMAG standards.
Start it here
Ask in plain language. The agent works on a Linux machine of its own — it installs and runs what the job needs, keeps every file, and reads the results back with the numbers explained. Nothing to set up, and a long run continues after you close the page.
What to watch for
- Contigs shorter than 1,000 bp for a single sample or 500 bp across multiple samples are omitted by default, and clusters smaller than 100 kb are not output.
- The generated intermediate coverage file is gzip-compressed and cannot be read or parsed directly as plain text.
- MetaCAT downstream modules including representative, abundance, variant, and mwas cannot be run locally because they require GTDB-Tk reference data exceeding 100 GB.
- Duplicated single-copy marker genes in BUSCO indicate contamination, which typically occurs when contigs from two distinct organisms are grouped into a single bin.
- High fragmentation in the underlying assembly inflates fragmented BUSCO scores and reduces apparent completeness even when the binning itself is accurate.
What it runs
Programs
- MetaCATInstalled on demandBins metagenome assemblies into genomes using coverage and composition, and links them with host traits.
- Bowtie 2Preinstalled on the research machineAligns sequencing reads to long reference genomes using memory-efficient FM indexing.
- samtoolsPreinstalled on the research machineA command-line toolsuite for reading, writing, sorting, indexing, and viewing SAM, BAM, and CRAM alignment files.
- CheckVRuns as a job on its own machine· BSD-3-Clause-LBNLA command-line pipeline for assessing the completeness, contamination, and quality of viral genomes from metagenomes.
nf-core pipelines
Questions
Do I need to install anything to recover genomes from a metagenome?
No. The research agent does this work on a Linux machine of its own — it runs MetaCAT, Bowtie 2, samtools, CheckV there, keeps the files, and reads the results back to you. Long runs continue if you close the page.
What inputs are required to run MetaCAT?
You need an assembly in FASTA format (compressed or uncompressed) alongside trimmed, host-depleted short reads for each sample, or BAM files already sorted against that exact assembly.
What standards determine whether a bin is a high-quality genome?
Under MIMAG criteria assessed via BUSCO, a high-quality draft genome requires at least 90% completeness and under 5% contamination (represented by duplicated markers), whereas medium-quality requires at least 50% completeness and under 10% contamination.
Does MetaCAT run downstream taxonomy or trait association on this machine?
No, downstream steps such as GTDB-Tk taxonomic assignment, representative genome selection, and microbiome-wide association studies require reference databases over 100 GB that are not hosted on this machine; you can use nf-core/mag or run those steps externally.
Can I use long-read sequencing data?
MetaCAT supports long reads via its coverage subcommand using the -l flag, but this workflow relies on minimap2, which is not preinstalled on the research machine and must be installed during the job.
Written from the method the research agent follows for this work and the documentation of the tools listed above. Statements citing a figure that does not appear there are removed before publication. Corollary is a non-commercial research demonstration; results are not for clinical use.