Skip to content
MetaGEAR

Presets

Most cohorts do not need one workflow at a time. MetaGEAR Tools ships three presets that run several workflows in dependency order in a single workspace, with --reuse-outputs switched on so each step picks up what the ones before it produced instead of recomputing it.

profiles Reference-based only — no assembly
microbial_profiles
  • MetaPhlAn taxonomic profiles
  • HUMAnN pathway and gene-family abundances
$ metagear profiles --input samples.csv --outdir results/
genomes The assembled bacterial picture
genes → classification → mag → msp
  • Contigs, gene and protein catalogs, with AMRFinderPlus and Pfam annotation
  • Per-contig classification and per-sample bacterial bins
  • A dereplicated MAG catalog with GTDB-Tk taxonomy and MAG×sample abundance
  • Species pangenomes with MSP×sample abundance and a MetaPhlAn cross-walk
$ metagear genomes --input samples.csv --outdir results/
microbiome Everything in genomes, with viruses and plasmids alongside
genomes + virus
  • Everything genomes produces
  • Viral and plasmid contigs, and their gene and protein catalogs
  • Viral-specific annotation — AMGs, PHROG, host prediction and lifestyle calls
$ metagear microbiome --input samples.csv --outdir results/

What a preset is, and is not

A preset is not a separate analysis. It is the same workflows in dependency order, with --reuse-outputs on. Running them by hand in that order, in the same output directory, gives the same result — the preset just saves you from typing four commands and from getting the order wrong.

microbiome is genomes plus virus. The reason virus runs second rather than last is that it performs its own assembly and gene calling; putting it after genes lets it read the existing contigs and gene catalog instead of rebuilding them. Rebuilding would matter: MMseqs2 picks different cluster representatives on a re-run, which silently invalidates the classification and MSP tables.

Before you run one

Reference databases are installed once per machine, not per run:

Terminal window
metagear download_databases

You can see exactly what a preset will do without running any of it:

Terminal window
metagear microbiome --input samples.csv --outdir results/ --preview

When a step fails

The steps after it do not start. Fix what the failing step reported and run the same command again — the steps that already finished resume from their own caches rather than starting over.

Cost

profiles skips assembly, which is the single most expensive stage, so it is the cheapest of the three in CPU-hours. It is not necessarily the quickest to finish: HUMAnN’s translated search against UniRef90 is slow, and on a small cohort it can dominate the wall clock.

genomes and microbiome both assemble. Plan for them at the cohort scale, on a cluster — see running across machines.

Where to go next