Gene Expression — Data Braids

Three datasets covering drug perturbation effects, baseline tissue expression, and cancer cell line profiling. The combination is the foundation for computational drug repurposing.

LINCS L1000 Level 5 + metadata

473K gene expression signatures across 12K genes. Drug perturbation, gene knockdown, and overexpression profiles across 77 cell lines. The core dataset for computational drug repurposing — each signature records how a cell’s gene expression changes in response to a chemical or genetic perturbation.

What’s Possible

• Combine with ChEMBL 37 for drug mechanism inference — match chemical bioactivity to gene expression changes
• Map perturbation signatures to GTEx tissue expression for tissue-specific drug effect prediction
• The gen5 critical path for tideGlass drug repurposing: LINCS + ChEMBL + GTEx = computational pharmacology without wet lab access

The Braid

sweetGrass braid.create produced a W3C PROV-O JSON-LD attestation:

{
  "@context": "https://www.w3.org/ns/prov#",
  "@id": "urn:braid:lincs-l1000-westgate-20260729",
  "prov:wasGeneratedBy": {
    "@type": "prov:Activity",
    "prov:used": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE92742",
    "prov:wasAssociatedWith": "did:eco:westgate"
  },
  "prov:wasAttributedTo": "did:eco:westgate",
  "prov:generatedAtTime": "2026-07-29T...",
  "eco:license": "CC-BY-4.0",
  "eco:blake3_root": "...",
  "eco:file_count": 6,
  "eco:size_bytes": 21474836480
}

GTEx V8 expression

Gene expression across 54 human tissues from 948 donors. TPM and read count matrices for tissue-specific expression analysis. The baseline map of where genes are expressed in the human body.

What’s Possible

• Map tissue-specific protein expression with UniProt Swiss-Prot
• Identify tissue selectivity of drug candidates with ChEMBL + LINCS
• Feed healthSpring clinical models with tissue context

The Braid

sweetGrass braid.create produced a W3C PROV-O JSON-LD attestation:

{
  "@context": "https://www.w3.org/ns/prov#",
  "@id": "urn:braid:gtex-v8-westgate-20260729",
  "prov:wasGeneratedBy": {
    "@type": "prov:Activity",
    "prov:used": "https://gtexportal.org/home/downloads/adult-gtex/bulk_tissue_expression",
    "prov:wasAssociatedWith": "did:eco:westgate"
  },
  "prov:wasAttributedTo": "did:eco:westgate",
  "prov:generatedAtTime": "2026-07-29T...",
  "eco:license": "dbGaP (public summary data)",
  "eco:blake3_root": "...",
  "eco:file_count": 4,
  "eco:size_bytes": 2576980378
}

GEO SOFT cancer series (11 series)

11 GEO cancer expression series including CCLE (GSE36139), GDSC (GSE68379), and other cancer cell line profiling datasets. Pre-processed SOFT format.

What’s Possible

• Drug sensitivity modeling: combine CCLE/GDSC expression with ChEMBL bioactivity
• Cancer-specific perturbation profiles for tideGlass
• Cross-reference with TCGA Xena clinical data for translational analysis

The Braid

sweetGrass braid.create produced a W3C PROV-O JSON-LD attestation:

{
  "@context": "https://www.w3.org/ns/prov#",
  "@id": "urn:braid:geo-soft-cancer-westgate-20260801",
  "prov:wasGeneratedBy": {
    "@type": "prov:Activity",
    "prov:used": "https://www.ncbi.nlm.nih.gov/geo/",
    "prov:wasAssociatedWith": "did:eco:westgate"
  },
  "prov:wasAttributedTo": "did:eco:westgate",
  "prov:generatedAtTime": "2026-08-01T...",
  "eco:license": "Public Domain",
  "eco:blake3_root": "...",
  "eco:file_count": 11,
  "eco:size_bytes": 3221225472
}

See Also

• Drug Discovery — ChEMBL, ZINC20, PubChem (the pharmacology layer)
• Cancer Genomics — TCGA Xena (clinical data)
• Structural Biology — PDB, UniProt (protein targets)
• What’s Possible — the drug repurposing combination