gsea-enrichment_skill

Overview

This skill performs gene set enrichment analysis (GSEA) and pathway enrichment within OmicVerse with explicit handling of geneset formats. It provides stepwise guidance to load pathway databases correctly and avoid common API errors. The goal is reliable enrichment results across bulk, single-cell, and spatial RNA-seq workflows.

How this skill works

The skill inspects the input gene list and runs ov.bulk.geneset_enrichment() after converting pathway files into the required dictionary format. It requires you to first download pathway databases and load them with ov.utils.geneset_prepare(), then pass the returned dictionary to the enrichment call. Results can be plotted with ov.bulk.geneset_plot() or compared across databases using ov.bulk.geneset_plot_multi().

When to use it

• Running GO, KEGG, Reactome, or custom pathway enrichment on DEG lists from bulk, single-cell, or spatial RNA-seq.
• When you need to combine multiple pathway databases and compare results visually.
• If enrichment calls are failing with FileNotFoundError or type errors referencing pathways_dict.
• When ensuring organism-specific gene symbol handling (Human vs Mouse) is required.

Best practices

• Always run ov.utils.download_pathway_database() once to populate local geneset files.
• Load geneset files using ov.utils.geneset_prepare() and pass the returned dictionary to geneset_enrichment(), never a file path string.
• Set organism explicitly ('Human' or 'Mouse') and confirm gene symbol capitalization matches the organism.
• Provide a background gene list where possible and use a gene list with >10 genes for robust hits.
• Combine and color-code multiple enrichment outputs with ov.bulk.geneset_plot_multi() for side-by-side comparison.

Example use cases

• Run GO Biological Process enrichment on DEGs from a bulk RNA-seq experiment using geneset_prepare + geneset_enrichment.
• Perform KEGG pathway enrichment on upregulated genes from single-cell cluster markers, ensuring Human vs Mouse mapping is correct.
• Download the pathway database once, prepare multiple geneset dictionaries (GO_BP, GO_MF, KEGG) and plot the top 5 terms across each database.
• Resolve errors when geneset_enrichment complains that pathways_dict is not a dict by showing the correct load-and-pass pattern.