sc-pathway-scoring

name	sc-pathway-scoring
description	Single-cell pathway and gene-set activity scoring for preprocessed scRNA-seq data using AUCell or a lightweight normalized-expression module-score path.
version	0.2.0
author	OmicsClaw Team
license	MIT
tags	["singlecell","pathway-scoring","pathway","geneset","aucell","module-score"]
metadata	{"omicsclaw":{"domain":"singlecell","allowed_extra_flags":["--method","--gene-sets","--gene-set-db","--species","--groupby","--top-pathways","--aucell-auc-max-rank","--score-genes-ctrl-size","--score-genes-n-bins","--aucell-py-auc-threshold","--r-enhanced"],"param_hints":{"aucell_r":"[Truncated]","score_genes_py":"[Truncated]","aucell_py":"[Truncated]"},"saves_h5ad":true,"requires_preprocessed":false,"trigger_keywords":["pathway score","pathway scoring","gene set score","module score","pathway activity","signature score"]}}

Single-Cell Pathway Scoring

Why This Exists

• Without it: users often jump from clustering to pathway claims without seeing per-cell score evidence.
• With it: this wrapper scores pathway or signature activity per cell, then optionally summarizes it across clusters or cell types.
• Why OmicsClaw: it keeps the AnnData contract stable, exports grouped tables, and renders a reusable pathway-scoring gallery.

Core Capabilities

1. Official AUCell execution via the Bioconductor R package.
2. Lightweight Python module scoring on normalized expression.
3. Per-cell score export back into processed.h5ad.
4. Grouped pathway summaries using a user-selected or auto-detected label column.
5. Stable tables and gallery figures for downstream interpretation and reuse.

Scope Boundary

Implemented methods:

1. aucell_r
2. score_genes_py

This skill focuses on per-cell pathway or signature scoring. It does not perform ORA or preranked GSEA significance testing. If you want GO/KEGG enrichment significance on a ranked gene list, that should be a separate enrichment skill.

Input Expectations

• Input object: preferably a preprocessed .h5ad
• Gene-set source: one of:
  • --gene-sets <path/to/file.gmt>
  • --gene-set-db hallmark|kegg|go_bp|go_cc|go_mf|reactome
• Optional grouping column: --groupby

Matrix expectations:

• score_genes_py expects X = normalized_expression
• aucell_r prefers normalized expression but can still score a count-like source by ranking genes within each cell

CLI Reference

python omicsclaw.py run sc-pathway-scoring \
  --input data.h5ad --gene-set-db hallmark --groupby leiden --output out/

python omicsclaw.py run sc-pathway-scoring \
  --input data.h5ad --method aucell_r --gene-sets pathways.gmt \
  --aucell-auc-max-rank 250 --top-pathways 25 --output out/

python omicsclaw.py run sc-pathway-scoring \
  --input data.h5ad --method score_genes_py --gene-set-db kegg \
  --groupby cell_type --score-genes-ctrl-size 50 --score-genes-n-bins 25 --output out/

Public Parameters