activepathways 2.0.6

Integrative pathway enrichment analysis of multivariate omics data (Python port of the ActivePathways R package).

ActivePathways (Python)

ActivePathways is a tool for integrative pathway enrichment analysis of multi-omics data. It identifies gene sets (such as pathways or Gene Ontology terms) that are over-represented in a matrix of genes and their p-values across multiple omics datasets. By fusing multiple datasets through p-value merging, ActivePathways surfaces biological signal that is invisible in any single dataset alone.

This is a Python port of the R ActivePathways package, preserving exact functionality and numerical output.

Citation

If you use ActivePathways please cite:

ActivePathways 2.0 (directional integration): Mykhaylo Slobodyanyuk, Alexander T. Bahcheli, et al. Directional integration and pathway enrichment analysis for multi-omics data. Nature Communications 15, 5690 (2024). doi:10.1038/s41467-024-49986-4

ActivePathways 1.0: Marta Paczkowska, Jonathan Barenboim, et al. Integrative pathway enrichment analysis of multivariate omics data. Nature Communications 11, 735 (2020). doi:10.1038/s41467-019-13983-9

Installation

git clone https://github.com/abahcheli/ActivePathways_python.git
cd ActivePathways_python
pip install -e .

To build a distributable conda package (targets the PyPI tarball):

conda build conda.recipe/
conda install --use-local activepathways

Note: conda.recipe/meta.yaml fetches the package from PyPI. Before the package is published to PyPI, the placeholder SHA256 in the recipe must be replaced with the real tarball hash, or the source section temporarily switched to path: .. for a local build.

Dependencies: numpy>=1.22, pandas>=1.5, scipy>=1.9. Python 3.9+ required.

Quick start

The two required inputs are a p-value matrix (genes × omics datasets, TSV) and a GMT file of gene sets. GMT files for common pathway databases can be downloaded from the Bader Lab gene sets page. Gene symbols must match between the matrix and the GMT file.

Command line

activepathways \
  --scores data/Adenocarcinoma_scores_subset.tsv \
  --gmt    data/hsapiens_REAC_subset.gmt \
  --output results.csv

All options beyond --scores, --gmt, and --output are optional:

activepathways \
  --scores            data/Adenocarcinoma_scores_subset.tsv \
  --gmt               data/hsapiens_REAC_subset.gmt \
  --output            results.csv \
  --merge_method      Brown \
  --cutoff            0.1 \
  --significant       0.05 \
  --correction_method holm \
  --geneset_filter_min 5 \
  --geneset_filter_max 1000

Run activepathways --help for a full listing.

Python

import pandas as pd
from activepathways import active_pathways, export_as_csv

scores = pd.read_csv("data/Adenocarcinoma_scores_subset.tsv", sep="\t").set_index("Gene")
scores = scores.fillna(1.0)  # replace missing p-values with 1 (not significant)

results = active_pathways(scores, "data/hsapiens_REAC_subset.gmt")
export_as_csv(results, "results.csv")

print(results[["term_id", "term_name", "adjusted_p_val", "term_size"]].head())

        term_id            term_name  adjusted_p_val  term_size
0  REAC:2424491      DAP12 signaling    4.491268e-05        358
1   REAC:422475        Axon guidance    2.028966e-02        555
2   REAC:177929    Signaling by EGFR    6.245734e-04        366
3  REAC:2559583  Cellular Senescence    6.636060e-05        196
4   REAC:180292     GAB1 signalosome    1.215316e-02        133

The overlap column lists the genes driving enrichment; evidence lists which input datasets contributed.

For extended examples — directional integration, GMT utilities, Cytoscape output, and merging results — see docs/python_examples.md.

Key parameters

Parameter	Default	Description
scores	required	DataFrame of p-values (genes × datasets). No NAs — replace with 1.0.
gmt	required	Path to a GMT file, or a GMT object from read_gmt().
background	all GMT genes	Custom gene universe for the hypergeometric test.
geneset_filter	(5, 1000)	(min, max) pathway size to retain.
cutoff	0.1	P-value cutoff for including genes in the ranked list.
significant	0.05	Adjusted p-value threshold for reporting pathways.
merge_method	"Fisher"	Method for combining p-values across datasets (see table below).
correction_method	"holm"	Multiple testing correction (holm, BH, bonferroni, etc.).
cytoscape_file_tag	None	File prefix for writing Cytoscape output files.
scores_direction	None	Fold-change direction matrix for directional methods.
constraints_vector	None	Expected directional relationships between datasets (1, -1, 0).

P-value merging methods

Method	Directional	Description
"Fisher"	No	Chi-squared combination of log p-values
"Brown"	No	Fisher's method corrected for between-dataset correlation
"Stouffer"	No	Z-score combination
"Strube"	No	Stouffer corrected for correlation
"Fisher_directional"	Yes	Fisher penalising directional conflicts
"DPM"	Yes	Brown's method with directional penalty (recommended)
"Stouffer_directional"	Yes	Stouffer with directional penalty
"Strube_directional"	Yes	Strube with directional penalty

API reference

Function	Description
active_pathways(scores, gmt, ...)	Main enrichment function
merge_p_values(scores, method, ...)	Combine p-values across datasets
read_gmt(filename)	Load a GMT file → GMT object
write_gmt(gmt, filename)	Write a GMT object to file
make_background(gmt)	Union of all genes across all GMT terms
export_as_csv(results, filename)	Save results table to CSV
merge_results(...)	Merge standard and directional results for Cytoscape
p_adjust(p, method)	Multiple testing correction (mirrors R's p.adjust)
enrichment_analysis(genelist, gmt, background)	Ordered hypergeometric test per term

Differences from the R package

R	Python
ActivePathways()	active_pathways()
export_as_CSV()	export_as_csv()
read.GMT()	read_gmt()
write.GMT()	write_gmt()
data.table output	pandas.DataFrame output
scores[is.na(scores)] <- 1	scores.fillna(1.0)

References

• Slobodyanyuk M*, Bahcheli AT*, et al. Directional integration and pathway enrichment analysis for multi-omics data. Nature Communications (2024). doi:10.1038/s41467-024-49986-4
• Paczkowska M*, Barenboim J*, et al. Integrative pathway enrichment analysis of multivariate omics data. Nature Communications (2020). doi:10.1038/s41467-019-13983-9
• Reimand J*, Isserlin R*, et al. Pathway enrichment analysis and visualization of omics data using g:Profiler, GSEA, Cytoscape and EnrichmentMap. Nature Protocols (2019). doi:10.1038/s41596-018-0103-9