Skip to main content

Pandas API for Gene Set Enrichment Analysis in Python (GSEApy, cudaGSEA, GSEA)

Project description

Build Status MIT License DOI

Pandas API for Gene Set Enrichment Analysis in Python (GSEApy, cudaGSEA, GSEA)

  • aims to provide a unified API for various GSEA implementations; uses pandas DataFrames and a hierarchy of Pythonic classes.
  • file exports (exporting input for GSEA) use low-level numpy functions and are much faster than in pandas
  • aims to allow researchers to easily compare different implementations of GSEA, and to integrate those in projects which require high-performance GSEA (e.g. massive screening for drug-repositioning)
  • provides useful utilities for work with GMT files, or gene sets and pathways in general in Python


To install the API use:

pip3 install gsea_api

See below for the instructions on installation of specific GSEA implementations.

Example usage

from pandas import read_table
from gsea_api.expression_set import ExpressionSet
from gsea_api.gsea import GSEADesktop
from gsea_api.molecular_signatures_db import GeneSets

reactome_pathways = GeneSets.from_gmt('ReactomePathways.gmt')

gsea = GSEADesktop()

design = ['Disease', 'Disease', 'Disease', 'Control', 'Control', 'Control']
matrix = read_table('expression_data.tsv', index_col='Gene')

result =
    # note: contrast() is not necessary in this simple case
    ExpressionSet(matrix, design).contrast('Disease', 'Control'),

Where expression_data.tsv is in the following format:

Gene    Patient_1   Patient_2   Patient_3   Patient_4   Patient_5   Patient_6
TACC2   0.2 0.1 0.4 0.6 0.7 2.1
TP53    2.3 0.2 2.1 2.0 0.3 0.6

MSigDB integration

Molecular Signatures Database (MSigDB) can be downloaded from the Broad Institute GSEA website. It provides expert-curated gene set collections, as well as curated subset of pathway databases (Reactome, KEGG, Biocarta, Gene Ontology) trimmed to remove redundant, overlapping and and otherwise little-value terms (if needed).

You can download all the pathways collections at once (search for ZIPped MSigDB on the download page). After downloading and un-zipping (e.g. to a local directory named msigdb), you can access the gene sets from MSigDB with:

from gsea_api.molecular_signatures_db import MolecularSignaturesDatabase

msigdb = MolecularSignaturesDatabase('msigdb', version=7.1)

msigdb.gene_sets returns a list of dictionaries describing auto-detected pathways:

    {'name': 'c1.all', 'id_type': 'symbols'},
    {'name': 'c1.all', 'id_type': 'entrez'},
    {'name': 'c2.cp.reactome', 'id_type': 'symbols'},
    {'name': 'c2.cp.reactome', 'id_type': 'entrez'}
    # etc..

Information about the location on disk and version are avilable in msigdb.path and msigdb.version.

msigdb.load loads the specific collection into a GeneSets object:

> kegg_pathways = msigdb.load('c2.cp.kegg', 'symbols')
> print(kegg_pathways)
<GeneSets 'c2.cp.kegg' with 186 gene sets>

This object can be passed to any of the supporteed GSEA implementations; please see below for a detailed description of the GeneSets object.

GeneSets objects

GeneSets represents a collection of sets of genes, where each set is represented as GeneSet object.

You can check the number of sets contained within a collection with:

> len(kegg_pathways)

The gene sets are accessible with gene_sets (tuple) and gene_sets_by_name (dict) properties:

> kegg_pathways.gene_sets[:2]
(<GeneSet 'KEGG_TIGHT_JUNCTION' with 132 genes>, <GeneSet 'KEGG_RNA_DEGRADATION' with 59 genes>)
> kegg_pathways.gene_sets_by_name
    'KEGG_TIGHT_JUNCTION': <GeneSet 'KEGG_TIGHT_JUNCTION' with 132 genes>,
    # etc.

Subseting collections

Sometimes only a subset of genes is measured in an experiment. You can remove gene sets which do not contain any of the measured genes from the collection:

> measured_genes = {'APOE', 'CYB5R1', 'FCER1G', 'PVR', 'HK2'}
> measured_subset = kegg_pathways.subset(measured_genes)
> print(measured_subset)
<GeneSets with 12 gene sets>

The skipped gene sets are accessible in measured_subset.empty_gene_sets for inspection.

Trimmming collections

> kegg_pathways.trim(min_genes=10, max_genes=20)
<GeneSets with 21 gene sets>

Prettify names

def prettify_kegg_name(name):
    return name.replace('KEGG_', '').replace('_', ' ')

kegg_pathways_pretty = kegg_pathways.format_names(prettify_kegg_name)
# (<GeneSet 'TIGHT JUNCTION' with 132 genes>, <GeneSet 'RNA DEGRADATION' with 59 genes>)

Other properties

Other properties and methods offered by GeneSets include: - all_genes: return a set of all genes which are covered by the gene sets in the collection - name: the name of the collection - to_frame() return a pandas DataFrame describing membership of the genes (gene sets = rows, genes = columns), which can be used for UpSet visualisation (e.g. with ComplexUpset) - to_gmt(path: str) exports the gene set to a GMT (Gene Matrix Transposed) file

Installing GSEA implementations

Following GSEA implementations are supported:

GSEA from Broad Institute

Login/register on the official GSEA website and download the gsea_3.0.jar file (or a newer version).

Provide the location of the downloaded file to GSEADesktop() using gsea_jar_path argument, e.g.:

gsea = GSEADesktop(gsea_jar_path='downloads/gsea_3.0.jar')


To use please install it with:

pip3 install gseapy

Use it with:

from gsea_api.gsea import GSEApy

gsea = GSEApy()


Please clone this fork of cudaGSEA to thirdparty directory and compile the binary version (using the instructions from this repository):

git clone

or use the original version, which does not implement FDR calculations.

Use it with:

from gsea_api.gsea import cudaGSEA

# CPU implementation can be used with use_cpu=True
gsea = cudaGSEA(fdr='full', use_cpu=False)



Please also cite the authors of the wrapped tools that you use.


The initial version of this code was written for a Master thesis project at Imperial College London.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for gsea-api, version 0.3.4
Filename, size File type Python version Upload date Hashes
Filename, size gsea_api-0.3.4.tar.gz (16.6 kB) File type Source Python version None Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page