MSigDB v2026.2.Hs (Oct 2026)

Important Notices#

• This page describes updates made to the Molecular Signatures Database Human Collections for release 2026.2 (MSigDB 2026.2.Hs).

• MSigDB v2026.2 is based on gene annotation data from Ensembl Release 116 (September 2026).

• The Ensembl 116 data utilized in MSigDB v2026.2 represents the final snapshot of Ensembl's microarray probe mapping process. Future releases of MSigDB will utilize this snapshot with remapping of gene symbols.

• In order to access the MSigBD mouse collections through the GSEA UI, the GSEA 4.3.0 or newer is required.

Updates to Human Collections (MSigDB v2026.2.Hs)#

C1: positional gene sets#

Updated human gene annotations to Ensembl 116.

C2:CGP#

Gene sets from two publications have been added to C2:CGP, these gene sets consist of:

• Signatures of Fibrolamellar Carcinoma and Intrahepatic Cholangiocarcinoma submitted by David Requena. See: Requena D, Medico JA, Soto-Ugaldi LF, Shirani M, Saltsman JA 3rd, Torbenson MS, Coffino P, Simon SM. Liver cancer multiomics reveals diverse protein kinase A disruptions convergently produce fibrolamellar hepatocellular carcinoma. Nat Commun. 2024 Dec 30;15(1):10887. doi: 10.1038/s41467-024-55238-2. PMID: 39738196; PMCID: PMC11685927.

• A signature describing H3K9ME3 mediated senesence submitted by Clemens Schmitt. This set has also been added to the M2:CGP collection for comparative analysis. See: Schleich K, Kase J, Dörr JR, Trescher S, Bhattacharya A, Yu Y, Wailes EM, Fan DNY, Lohneis P, Milanovic M, Lau A, Lenze D, Hummel M, Chapuy B, Leser U, Reimann M, Lee S, Schmitt CA. H3K9me3-mediated epigenetic regulation of senescence in mice predicts outcome of lymphoma patients. Nat Commun. 2020 Jul 20;11(1):3651. doi: 10.1038/s41467-020-17467-z. PMID: 32686676; PMCID: PMC7371731.

C2:CP:KEGG_MEDICUS#

• KEGG Medicus data has been updated to the KEGG Network data file released as of 2026-07-23. Additionally, the KEGG_MEDICUS subcollation now undergoes comparable Jaccard coefficient based redundancy filtering to the Reactome and GO gene sets in order to optimize the subcollection for GSEA's competitive false-discovery rate testing, see notes for Reactome and GO.

C2:CP:Reactome#

• Reactome gene sets have been updated to reflect the state of the Reactome pathway architecture as of Reactome 97 (+28 gene sets).
• As previously described in the Reactome release notes for MSigDB 7.0, in order to limit redundancy between gene sets within the Reactome subcollection we applied a filtering procedure based on Jaccard coefficients and distance from the top level of the Reactome event hierarchy.

C2:CP:WikiPathways#

WikiPathways gene sets have been updated to the October 10, 2026 release (+2 gene sets).

C3:TFT:GTRD#

Two gene sets were removed from this collection as a result of mapping changes causing them to fall outside of MSigDB inclusion thresholds.

C5:GO (Gene Ontology)#

Gene sets in these subcollections are derived from the controlled vocabulary of the Gene Ontology (GO) project: The Gene Ontology Consortium. Gene Ontology: tool for the unification of biology (Nature Genet 2000). The gene sets are named by GO term and contain genes annotated by that term. This collection has been updated to the most recent GO annotations at the time of release production using annotations in the GO-basic obo file released on 2026-07-26 and NCBI gene2go annotations downloaded on 2026-09-30.

This collection is divided into three subcollections:

• BP: GO Biological process (-633 gene sets). Gene sets derived from the Biological Process Ontology; set names are prefixed with GOBP_.
• CC: GO Cellular component (+14 gene sets). Gene sets derived from the Cellular Component Ontology; set names are prefixed with GOCC_.
• MF: GO Molecular function (+46 gene sets). Gene sets derived from the Molecular Function Ontology; set names are prefixed with GOMF_.

These updates were generated in accordance with the procedure described in the GO release notes for MSigDB 7.0.

The gene sets generated for the GO subcollections have been derived utilizing a new construction process. This process corrects an error identified with the exclusion of genes specifically annotated with the "NOT" annotation qualifier. Previously this small subset of genes were not properly excluded from their associated GO term.

C5:HPO (Human Phenotype Ontology)#

Gene sets in this subcollection have been updated to reflect the 2026-09-01 release of the Human Phenotype Ontology database (+139 gene sets). This subcollection has been redundancy filtered through a procedure comparable to that of the GO and Reactome subcollections.

C9: Computational Perturbation Signature Gene Sets#

Gene sets in this subcollection have been updated to correct the nomenclature used in the gene set name to better reflect the underlying methodology used for set generation. Previously these sets were named GENES_CORRELATED_WITH_[GENE_OF_INTEREST]_DELETION, as these gene sets were generated through comparison of expression profiles to CRISPR deletion screening of the gene of interest across multiple cell lines. However, in MSigDB 2026.2 this has been updated to GENES_CORRELATED_WITH_[GENE_OF_INTEREST]_DEPENDENCY, as DEPENDENCY is a more accurate representation of the nature of the expression correlation analysis.

CHIP file updates#

• MSigDB 2026.2.Hs gene annotations and gene mapping CHIP files have been updated to data from Ensembl 116.
• Gene orthology annotations for mapping mouse and rat genes to their best match mouse orthologs have been updated to Alliance of Genome Resources orthology database release 9.1.0 (2026-09-01)
• Due to early retirement of the Affymetrix Human Transcriptome Array dataset, this CHIP file has been constructed through retrieval of Ensembl 115 archive data and the symbols remapped to the Ensembl 116.