This is the original GitHub repository.
This repository contains the MSigDB gene sets -- specifically the CC-BY subset that includes the cancer hallmark, the GO gene sets, etc., but excludes KEGG and BIOCARTA (and AAAS/STKE) gene sets. The aim is to provide the gene sets in a format that is compressed, easy to load and reference, and to assist code reproducibility.
This dataset is licensed CC(4.0)-BY Broad Institute. See here for details (archived url).
The gene sets are stored in a
single
JSON file (~60MB)
compressed down to (~13MB)
with bzip2.
The file was compiled from
msigdb.v7.1.symbols.gmt
and
msigdb.v7.1.entrez.gmt
obtained from
the downloads
page of MSigDB (2020-06-28).
The file contains both the Entrez Gene Identifiers and the HUGO Gene Symbols.
Please consider hitting the star if this is useful to you.
This is the quickest way to inspect the gene sets in Python:
import pandas as pd
url = "https://github.com/numpde/genesets/raw/53ce4ba8614d6d3ac2ca33243ea3f9f2c1f86ef5/genesets/msigdb/parsed/v7.1/genesets.json.zip"
print(pd.read_json(url)) AAANWWTGC_UNKNOWN ... CTTTGT_LEF1_Q2
card http://www.gsea-msigdb.org/gsea/msigdb/cards/A... ... http://www.gsea-msigdb.org/gsea/msigdb/cards/C...
symbols [MEF2C, ATP1B1, RORA, CITED2, APP, MAP3K4, ATP... ... [MEF2C, AJUBA, SCN3A, RTL8A, SYNCRIP, RORB, AM...
entrez [4208, 481, 6095, 10370, 351, 4216, 493, 2904,... ... [4208, 84962, 6328, 26071, 10492, 6096, 154810...
The above URL refers to a certain commit on GitHub and its content is therefore immutable. Note that this will not cache the downloaded file. You can download the archive programmatically or manually and read from disk, as assumed in the following example.
import pandas as pd
df = pd.read_json("parsed/v7.1/genesets.zip")
i = 'HALLMARK_DNA_REPAIR'
print(df[i].card)
print(df[i].symbols)
print(df[i].entrez)
# http://www.gsea-msigdb.org/gsea/msigdb/cards/HALLMARK_DNA_REPAIR
# ['POLR2H', 'POLR2A', 'POLR2G', 'POLR2E', 'POLR2J', 'POLR2F', ...]
# ['5437', '5430', '5436', '5434', '5439', '5435', '5432', ...]
# All 50 "cancer hallmark" gene sets
print(df.T[[c.startswith("HALLMARK") for c in df]])
# card ... entrez
# HALLMARK_TNFA_SIGNALING_VIA_NFKB http://www.gsea-msigdb.org/gsea/msigdb/cards/H... ... [3726, 2920, 467, 4792, 7128, 5743, 2919, 8870...
# HALLMARK_HYPOXIA http://www.gsea-msigdb.org/gsea/msigdb/cards/H... ... [5230, 5163, 2632, 5211, 226, 2026, 5236, 1039...
# HALLMARK_CHOLESTEROL_HOMEOSTASIS http://www.gsea-msigdb.org/gsea/msigdb/cards/H... ... [2224, 1595, 3422, 2222, 1717, 6713, 3157, 508...The genes (but not necessarily the gene sets) are in the same order as in the original files.
The following is a more Python-native way of loading the file as a dictionary.
import json
from zipfile import ZipFile
with ZipFile("parsed/v7.1/genesets.zip") as zf:
with zf.open("genesets.json") as fd:
genesets = json.load(fd)
print(genesets['HALLMARK_DNA_REPAIR']){
"card": "http://www.gsea-msigdb.org/gsea/msigdb/cards/HALLMARK_DNA_REPAIR",
"symbols": [
"POLR2H",
"POLR2A",
"POLR2G",
...
],
"entrez": [
"5437",
"5430",
"5436",
...
]
}
# install.packages("jsonlite")
tmp <- tempfile()
download.file("https://github.com/numpde/genesets/raw/53ce4ba8614d6d3ac2ca33243ea3f9f2c1f86ef5/genesets/msigdb/parsed/v7.1/genesets.json.zip", tmp)
genesets <- jsonlite::stream_in(unz(tmp, "genesets.json")) # Ignore warnings
genesets$HALLMARK_HYPOXIA$entrezDownload genesets.json.zip.
As of 2020-06-28, the original website (archive) recommends citing
- Subramanian, Tamayo, et al. (2005, PNAS 102, 15545-15550)
and some of
- Liberzon, et al. (2011, Bionformatics)
- Liberzon, et al. (2015, Cell Systems)
- the source for the gene set as listed on the gene set page
as appropriate. Hence, as "the source for the gene set" you can refer to the URL of this repository or the specific link that you used to download the gene sets.
This repository
contains the
SCSig gene sets
(archive),
compiled from
scsig.all.v1.0.1.symbols.gmt,
scsig.all.v1.0.1.entrez.gmt
and
scsig.v1.0.metadata.txt.
The symbols, entrez and metadata files
are
merged into one JSON file.
import pandas as pd
url = "https://github.com/numpde/genesets/raw/c3ee17dfb92b51be82e9bd7ff292a28a801da8aa/genesets/scsig/parsed/v1.0.1/genesets.json.zip"
print(pd.read_json(url)) Fan_Embryonic_CTX_Big_Groups_Cajal_Retzius ... Hu_Fetal_Retina_RPE
card na ... na
symbols [RELN, WFIKKN2, IGFBP5, PRPH, PCP4, LINC01133,... ... [PMEL, TYRP1, PTGDS, SERPINF1, TTR, TYR, ELN, ...
entrez [5649, 124857, 3488, 5630, 5121, 100505633, 79... ... [6490, 7306, 5730, 5176, 7276, 7299, 2006, 430...
organism Homo sapiens ... Homo sapiens
organ_system Central Nervous System ... Visual System
pmid 29867213 ... 31269016
publication_title Spatial transcriptomic survey of human embryon... ... Dissecting the transcriptome landscape of the ...
authors Fan X,Dong J,Zhong S,Wei Y,Wu Q,Yan L,Yong J,S... ... Hu Y,Wang X,Hu B,Mao Y,Chen Y,Yan L,Yong J,Don...
geoid GSE103723 ... GSE107618
exact_source Supplementary information, Table S3: DEGs_of_8... ... S4 Table: DEGs of all cell classes.
external_details_url ...
chip HUMAN_GENE_SYMBOL ... HUMAN_GENE_SYMBOL
category_code SCSig ... SCSig
contributor Anthony Castanza ... Anthony Castanza
contributor_org MSigDB Team ... MSigDB Team
description_brief ... Retinal Pigment Epithelium Cells
raw_publication_members RELN,WFIKKN2,IGFBP5,PRPH,PCP4,LINC01133,NDNF,A... ... PMEL,TYRP1,PTGDS,SERPINF1,TTR,TYR,ELN,TRPM1,TI...
[17 rows x 257 columns]
Refer to the gsea-msigdb website (archive) and the respective publication/s of the gene set/s from the metadata.