OmicIDX 2026-11

OmicIDX is a regularly refreshed, analysis-ready copy of public genomics metadata: NCBI SRA (studies, samples, experiments, runs, accessions), NCBI GEO (series, samples, platforms), NCBI and EBI BioSample, and BioProject. It also ships the sradb.* and geometadb.* marts. Everything is plain Parquet that any engine can read.

This is the monthly release 2026-11, a verbatim copy of the daily publish v2026-11-02/. Pin it with the @2026-11 revision. For the daily-refreshed copy, use R2 latest/ at https://data.omicidx.cancerdatasci.org/latest/. Documentation lives at https://omicidx.cancerdatasci.org.

PubMed articles (pubmed_articles.parquet) were published through release 2026-10 and are not in later releases. PubMed is moving to its own dataset (https://github.com/seandavi/cdsci-lake/issues/136).

Provenance

Publish date	2026-11-02
Published at	2026-11-02T17:45:14.374442+00:00
Transform SHA	8b6e12367e5659ac1b00fed3bee325ecf99f05b7
Lake snapshot id	4813

The full provenance record is manifest.json.

Tables

28 files, 44.6 GB in total.

View	File	Rows	Size (GB)
bioprojects	bioprojects.parquet	1,134,964	0.25
biosamples	biosamples.parquet	60,718,155	7.99
ebi_biosamples	ebi_biosamples.parquet	54,312,512	7.83
geo_platforms	geo_platforms.parquet	29,170	0.01
geo_samples	geo_samples.parquet	8,823,915	2.52
geo_series	geo_series.parquet	300,067	0.16
geo_series_with_rnaseq_counts	geo_series_with_rnaseq_counts.parquet	27,714	0.00
geometadb.geo_supplemental_files	geometadb/geo_supplemental_files.parquet	7,134,321	0.21
geometadb.gpl	geometadb/gpl.parquet	29,170	0.00
geometadb.gse	geometadb/gse.parquet	300,067	0.14
geometadb.gse_gpl	geometadb/gse_gpl.parquet	345,242	0.00
geometadb.gse_gsm	geometadb/gse_gsm.parquet	10,621,571	0.01
geometadb.gsm	geometadb/gsm.parquet	8,823,915	3.46
sra_accessions	sra_accessions.parquet	163,550,412	5.21
sra_experiments	sra_experiments.parquet	42,902,998	3.86
sra_runs	sra_runs.parquet	45,758,747	1.71
sra_samples	sra_samples.parquet	44,888,669	4.33
sra_studies	sra_studies.parquet	778,502	0.18
sradb.experiment	sradb/experiment.parquet	42,902,998	0.76
sradb.human_runs	sradb/human_runs.parquet	8,589,175	0.12
sradb.mouse_runs	sradb/mouse_runs.parquet	4,434,357	0.06
sradb.rnaseq_runs	sradb/rnaseq_runs.parquet	9,124,312	0.14
sradb.run	sradb/run.parquet	45,904,131	0.74
sradb.run_with_study	sradb/run_with_study.parquet	45,904,131	0.67
sradb.sample	sradb/sample.parquet	44,888,669	1.30
sradb.sra	sradb/sra.parquet	45,904,131	2.62
sradb.study	sradb/study.parquet	778,502	0.15
sradb.wgs_runs	sradb/wgs_runs.parquet	9,681,692	0.16

Querying

DuckDB, one file:

SELECT count(*) FROM 'hf://datasets/seandavis/omicidx@2026-11/sra_studies.parquet';

DuckDB, every table as a view (views.sql points at this release):

curl -sL https://huggingface.co/datasets/seandavis/omicidx/resolve/2026-11/views.sql | duckdb omicidx.duckdb
duckdb omicidx.duckdb -c "SELECT count(*) FROM sradb.study"

Python (polars):

import polars as pl

df = pl.scan_parquet("hf://datasets/seandavis/omicidx@2026-11/sra_studies.parquet").head(10).collect()

R (arrow):

url <- "https://huggingface.co/datasets/seandavis/omicidx/resolve/2026-11/sra_studies.parquet"
tf <- tempfile(fileext = ".parquet")
download.file(url, tf, mode = "wb")
studies <- arrow::read_parquet(tf)

License

The OmicIDX code is MIT-licensed (https://github.com/seandavi/omicidx). The data are redistributed from NCBI (SRA, GEO, BioSample, BioProject) and EBI (BioSamples) and remain under those providers' terms of use.

Citation

@misc{omicidx_2026_11,
  title  = {OmicIDX, release 2026-11},
  author = {Davis, Sean},
  year   = {2026},
  url    = {https://huggingface.co/datasets/seandavis/omicidx/tree/2026-11}
}