Supplementary Data for MOTIVE: Drug-Target Interaction Annotations from jump_compound_annotator

OVERVIEW Supplementary material for Arevalo, Su et al., NeurIPS 2024 - a drug-target interaction graph dataset combining annotations from 8 public databases with Cell Painting image features. The processed MOTIVE dataset is on the Cell Painting Gallery S3 bucket. This Zenodo deposit provides raw annotation pipeline outputs including ID mapping infrastructure and original database downloads. Note: Row counts may differ slightly from S3 annotations due to database updates since original 2024-06 run. DIRECTORY STRUCTURE - annotations/ - Drug-target interaction parquet files- mappings/ - ID mapping files (pointers.csv, UniChem/MyChem mappers)- intermediate/ - Compressed UniChem API query outputs- raw_sources/ - Compressed raw database downloads ANNOTATIONS Drug-target annotations from 8 databases: BioKG, DGIdb, DrugRep, Hetionet, OpenBioLink, OpenTargets, PharmeBiNet, PrimeKG - compound_gene.parquet - Drug-gene annotations- gene_gene.parquet - Gene-gene interactions- compound_compound.parquet - Drug-drug interactions- compound_gene_curated.parquet - Curated drug-gene (standardized rel_types, hub compounds removed) See MOTIVE wiki for schema details: https://github.com/carpenter-singh-lab/2024_Arevalo_NeurIPS_MotiVE/wiki MAPPINGS - pointers.csv - UniChem mappings (InChIKey to DrugBank/ChEMBL/PubChem)- unichem_*_mapper.parquet - InChIKey resolution via UniChem API- mychem_*_mapper.parquet - InChIKey resolution via MyChem API INTERMEDIATE - ids.tar.gz - Batch results from UniChem API queries- errors.tar.gz - Failed lookups- external_ids.tar.gz - Exported external IDs by database RAW SOURCES Compressed downloads from each source database (can be re-downloaded via pipeline):biokg, dgidb, drugrep, hetionet, hgnc, ncbi, openbiolink, opentargets, pharmebinet, primekg CODE Pipeline: https://github.com/broadinstitute/monorepo/tree/main/libs/jump_compound_annotator File renames for S3 consistency:- annotations.parquet -> compound_gene.parquet- gene_interactions.parquet -> gene_gene.parquet- compound_interactions.parquet -> compound_compound.parquet- filtered_annotations.parquet -> compound_gene_curated.parquet CITATION 1. MOTIVE paper: Arevalo J, Su E, Carpenter AE, Singh S (2024) MOTIVE: A Drug-Target Interaction Graph For Inductive Link Prediction. NeurIPS 2024. doi: 10.48550/arXiv.2406.08649 https://github.com/carpenter-singh-lab/2024_Arevalo_NeurIPS_MotiVE 2. Please also cite individual databases as described in the MOTIVE paper. LICENSE Derived from multiple public databases. See MOTIVE paper for data sources and terms of use.