omicverse-micro-metabol-paired

What problem does it solve?

This Skill turns paired microbiome and metabolomics tables into interpretable cross-omics associations, so you can move from matched sample data to biologically meaningful microbe-metabolite links without manual method switching.

Core Features & Use Cases

• Paired association discovery: Run Spearman correlation with FDR control to rapidly rank candidate microbe-metabolite pairs.
• Shared latent structure: Use CCA to detect a small number of canonical modes that explain joint variation across modalities.
• Co-occurrence modeling: Train MMvec to estimate conditional metabolite profiles, latent embeddings, and top association pairs for deeper interpretation.
• Validation and diagnostics: Compare methods on synthetic paired data and inspect training, co-occurrence, and biplot visualizations to verify results.
• Use case: Analyze matched gut microbiome and metabolomics samples to identify which microbes covary with specific metabolites and whether those links are robust across methods.

Quick Start

Ask the skill to load paired microbiome and metabolomics AnnData objects, run Spearman, CCA, and MMvec, and return the strongest cross-omics pairs with validation plots.