Skip to content

Microbiome

The Microbiome section is a portal that hosts every module that operates on microbiome data — community metabolic modelling, pathway differential abundance, feature-association modelling, and (over time) more. Each module is independently executable; you pick the one that answers the question you actually have, without being forced through another one's configuration.

When to use this section

When you have at least one metagenomics dataset in a project (MetaPhlAn shotgun, or an upload whose data_type is set to metagenomics during configure). A paired metabolomics dataset in the same project unlocks GeMMA; the other modules don't require pairing.

The module portal

Clicking Microbiome in the top nav lands you on the portal for the project's first metagenomics dataset. The left panel lists all available modules with one-line descriptions:

Module What it does Requires
GeMMA Community genome-scale metabolic model + punch differential + concordance. Paired metabolomics dataset (HMDB-annotated preferred).
HUMAnN3 MetaCyc pathway abundance from a pre-computed HUMAnN3 run; Wilcoxon per path. Attached pathabundance_cpm.tsv (role humann3_pathabundance).
MaAsLin2 General linear (or linear-mixed) model for microbial feature ↔ metadata. Coming soon — runner not wired yet.
MACADAM Per-taxon metabolic pathway completeness (MetaCyc/KEGG) from taxonomy. Coming soon.
MelonnPan Elastic-net prediction of metabolite profiles from microbiome composition. Coming soon — R-bridge pending.
MIMOSA2 Community-level Metabolic Potential + per-metabolite microbiome attribution. Coming soon — R-bridge pending.

Modules are greyed out when their prerequisites aren't met — hover for the reason.

How modules relate

Each module writes its own database record (GemmaAnalysis, Humann3Analysis, MaaslinAnalysis) and has its own results page. There is no shared "run everything" button; modules don't depend on each other's outputs. You can compare their outputs side-by-side using Compare when both modules use overlapping pathway or taxon identifiers.

Module pages

  • GeMMA — community metabolic model, punch, concordance, Steiner subnetworks.
  • HUMAnN3 — MetaCyc pathway differential (FASTQ → pathabundance → Wilcoxon).
  • MaAsLin2 — microbial feature association modelling (placeholder).
  • MACADAM — per-taxon pathway completeness from taxonomy (placeholder).
  • MelonnPan — statistical microbiome → metabolite prediction (placeholder).
  • MIMOSA2 — metabolite-centric CMP scoring + taxon attribution (placeholder).

Reproducibility

Every module run captures the input file identity, group-comparison parameters, and (for GeMMA) the random seed used for permutation statistics. The pinned VMH metabolite snapshot (data/resources/gemma/apollo/vmh_metabolites.csv) carries a manifest + SHA256 so the HMDB↔VMH↔KEGG mapping is byte-identical across hosts and dates — required for publication-grade reruns.