mdhewei/bioinfor-claw/go-analysis-for-gene-list

GO / KEGG / Reactome Enrichment Plotter

Use this skill to perform over-representation analysis from a user-provided gene list and generate publication-quality enrichment figures and result tables.

Purpose

This skill is designed for routine functional enrichment analysis of a gene list. It retrieves enriched terms from supported Enrichr libraries and produces clean, publication-ready bubble plots together with tabular outputs.

Use this skill when the user wants:

• Gene Ontology enrichment
• KEGG pathway enrichment
• Reactome pathway enrichment
• bubble plots for enriched terms
• result tables suitable for downstream interpretation or figure preparation

This skill is optimized for:

• human and mouse gene lists
• unranked gene lists
• over-representation analysis
• publication-ready static figures

Use when

Use this skill when the user:

• provides a gene list directly or through a file
• asks for GO enrichment analysis
• asks for KEGG or Reactome pathway enrichment
• asks for BP, MF, or CC enrichment
• asks for a bubble plot, dot plot, or enrichment figure
• wants top enriched terms visualized
• wants a result table and figure saved to disk

Typical requests:

• “Run GO enrichment on this gene list.”
• “Make a publication-quality bubble plot for BP terms.”
• “Do KEGG and Reactome enrichment for these genes.”
• “Show the top 15 GO terms with adjusted p-value.”
• “Generate enrichment plots for human genes.”

Do not use when

Do not use this skill when the user:

• provides a ranked gene list and wants GSEA rather than over-representation analysis
• wants gene set enrichment from fold change values or a preranked vector
• wants genome-wide annotation downloads without enrichment analysis
• wants custom pathway databases that are not supported by the script
• wants interactive web-based plots rather than static publication figures
• wants organism support outside the currently supported set
• wants network visualization, pathway maps, or graph-based pathway diagrams

If the user wants ranked-list enrichment or GSEA, use a dedicated GSEA workflow instead of this skill.

Supported organisms

Supported organism inputs are normalized internally.

Accepted user inputs include:

• human
• homo sapiens
• hs
• hsapiens
• mouse
• mus musculus
• mm

These map to:

• human
• mouse

If the user gives an unsupported organism, fail clearly and explain the allowed values.

Supported libraries

The script supports the following categories:

• GO_BP for Gene Ontology Biological Process
• GO_MF for Gene Ontology Molecular Function
• GO_CC for Gene Ontology Cellular Component
• GO for all three GO categories above
• KEGG for KEGG pathway enrichment
• REACTOME for Reactome pathway enrichment
• ALL for GO_BP, GO_MF, GO_CC, KEGG, and REACTOME together

Expected inputs

The skill expects:

• a plain-text gene list file with one gene per line, or equivalent clean gene list input
• an output prefix for saving results

Optional inputs include:

• organism
• library category
• number of top terms to plot
• significance cutoff
• figure width and height
• DPI
• font family
• font size
• colormap

Expected outputs

For each requested category, generate:

1. a full enrichment result table

Example:

• <output_prefix>.GO_BP.tsv
• <output_prefix>.KEGG.tsv

1. a reduced plotting table containing the plotted terms

Example:

• <output_prefix>.GO_BP.plot_input.tsv

1. a publication-quality bubble plot in PNG format

Example:

• <output_prefix>.GO_BP.bubble.png

1. a publication-quality bubble plot in PDF format

Example:

• <output_prefix>.GO_BP.bubble.pdf

Enrichment backends

The script supports two enrichment backends:

1. g:Profiler (primary, default) — Uses official GO consortium annotations via the g:Profiler REST API (biit.cs.ut.ee). Automatically maps gene aliases/outdated symbols. Generally gives more comprehensive results matching what online GO tools produce.
2. Enrichr (fallback) — Uses Enrichr curated libraries via gseapy. Falls back to this if g:Profiler is unreachable.

Control with --backend auto|gprofiler|enrichr (default: auto = try g:Profiler first).

Assumptions

This skill assumes:

• the input is an unranked gene list
• gene symbols are already reasonably clean (g:Profiler will handle alias mapping)
• the user wants over-representation analysis
• enrichment uses g:Profiler (primary) or Enrichr (fallback)
• adjusted p-value is the main significance metric for ranking and plotting

If the user does not specify otherwise, use:

• human as the default organism
• ALL as the default library set only if the user explicitly asks for multiple categories; otherwise use the requested category
• top_n = 15
• sig_cutoff = 0.05
• publication-style figure defaults from the script

Workflow

1. Read the input gene list.
2. Clean and normalize gene symbols by trimming whitespace and converting to uppercase.
3. Normalize the organism input to a supported value.
4. Resolve the requested enrichment library or library panel.
5. Run enrichment analysis using gseapy.enrichr.
6. Validate the returned result table and expected columns.
7. Compute plotting fields such as:

• hit count
• background count
• gene ratio
• -log10(FDR)

1. Rank terms by adjusted p-value and enrichment strength.
2. Select significant terms for plotting. If no term passes the significance cutoff, fall back to the top ranked terms.
3. Generate a publication-quality bubble plot.
4. Save full results, plotting input, PNG figure, and PDF figure.
5. Return the saved output paths and summarize what was generated.

Plot behavior

The default plot is a bubble plot where:

• x-axis = gene ratio
• y-axis = enriched term
• bubble size = hit count
• bubble color = -log10(FDR)

The script is designed for publication-style output:

• wrapped term labels for readability
• clean axes
• no top/right spines
• bubble size legend
• color bar for significance
• PNG and PDF export

Execution

Run the main script:

python scripts/go_enrichment_plotter.py --input <gene_file> --output-prefix <prefix>

Common examples

Run all supported libraries for a human gene list:

python scripts/go_enrichment_plotter.py --input examples/example_genes.txt --output-prefix results/example_all --organism human --library ALL --top-n 15

Run only GO Biological Process:

python scripts/go_enrichment_plotter.py --input examples/example_genes.txt --output-prefix results/example_bp --organism human --library GO_BP --top-n 15

Run only KEGG:

python scripts/go_enrichment_plotter.py --input examples/example_genes.txt --output-prefix results/example_kegg --organism human --library KEGG --top-n 15

Run only Reactome:

python scripts/go_enrichment_plotter.py --input examples/example_genes.txt --output-prefix results/example_reactome --organism human --library REACTOME --top-n 15

Supported arguments

Required:

• --input <gene_file>
• --output-prefix <prefix>

Optional:

• --organism <human|mouse|hs|hsapiens|homo sapiens|mm|mus musculus|mouse>
• --library <GO_BP|GO_MF|GO_CC|GO|KEGG|REACTOME|ALL>
• --backend <auto|gprofiler|enrichr> (default: auto)
• --top-n <int>
• --sig-cutoff <float>
• --fig-width <float>
• --fig-height <float>
• --dpi <int>
• --font-family <string>
• --font-size <float>
• --cmap <matplotlib_colormap>

Parameter decision guide

Choose library and parameters based on the biological question:

Parameter interpretation

Translate natural-language user requests into script arguments where possible.

Examples:

• “Use human genes”

→ --organism human

• “Run BP only”

→ --library GO_BP

• “Run KEGG and Reactome”

→ run the script twice or use separate category handling if needed by the caller

• “Show top 10 terms”

→ --top-n 10

• “Use adjusted p < 0.01”

→ --sig-cutoff 0.01

• “Use Arial font and 12 pt labels”

→ --font-family Arial --font-size 12

• “Make the figure higher resolution”

→ --dpi 300 or higher if explicitly requested

Output conventions

• Save tables as TSV.
• Save figures as both PNG and PDF.
• Preserve original enrichment fields from the enrichment result whenever possible.
• Use deterministic sorting:
• adjusted p-value ascending
• hit count descending
• gene ratio descending
• term name ascending
• If no term passes the significance threshold, still generate a plot from the top ranked terms and note that fallback behavior.

Dependencies

Install with:

pip install -r requirements.txt

Expected packages include:

• pandas
• numpy
• matplotlib
• gseapy (optional, needed for Enrichr fallback)

No extra packages needed for g:Profiler — it uses Python's built-in urllib.

Failure modes

Fail clearly when:

• the input gene list is empty after cleaning
• the organism is unsupported
• the requested library is unsupported
• enrichment returns no results
• the result table is missing expected columns
• no output file can be written
• no enrichment category succeeds when multiple categories were requested

If one category fails but others succeed, continue with the successful categories and report which category failed.

Robot usage notes

When another robot uses this skill:

• prefer explicit user-specified library and organism values
• do not assume ranked-list enrichment; this skill is for unranked lists
• do not silently switch to GSEA
• do not invent unsupported organisms or library names
• return the output file paths in the response
• mention whether the plot used significant terms only or fell back to top ranked terms

Trigger examples

This skill should trigger for prompts like:

• “Run GO enrichment on this gene list.”
• “Make a bubble plot for BP enrichment.”
• “Do KEGG pathway enrichment for these genes.”
• “Run Reactome analysis and generate publication-ready figures.”
• “Perform GO, KEGG, and Reactome enrichment for this list.”

Non-trigger examples

This skill should not trigger for prompts like:

• “Run GSEA on my ranked gene list.”
• “Plot a volcano plot for these DEGs.”
• “Find sgRNAs for these genes.”
• “Curate a literature-based gene list for ferroptosis.”
• “Download all GO annotations for human genes.”