What problem does it solve? Genes annotated to many GO terms (like TP53 or BRCA1) appear as top hits in nearly every enrichment, guilt-by-association, or differential expression analysis regardless of the actual biology. This Skill quantifies that bias per gene so you can distinguish genuinely implicated genes from usual suspects. ## Core Features & Use Cases - Multifunctionality scoring: Computes the Gillis & Pavlidis (2011) MF score over propagated GO BP annotations, outputting a ranked TSV with score, rank, and percentile per gene. - Reproducible provenance: Every run writes a sidecar .meta.json recording GAF date, OBO data-version, sha256 hashes, and all filter choices. - Composable background sets: Restricts the computation to a protein-coding universe by default via the sibling gene-set-fetch skill, or accepts a user-supplied TSV. - Use Case: After running a GO enrichment that returns TP53 and MYC as top hits, check their mf_percentile to see whether they rank high simply because they are annotated everywhere. ## Quick Start Run the multifunctionality script for human genes to produce a ranked per-gene TSV of multifunctionality scores.
