i-DILI: A Harmonized Multi-Source Database for Drug-Induced Liver Injury (v1.0.0)

i-DILI is an open, harmonized multi-source database for drug-induced liver injury (DILI) labeling and machine-learning benchmarking. Version 1.0.0 contains 17,778 unique compounds with binary DILI labels, a multi-source confidence score (Cf*), canonical SMILES where available, and computed molecular descriptors. Sources and harmonization The pipeline integrates six public resources (LiverTox, DILIrank, DILIst, DrugBank, Open TG-GATEs, Tox21). SIDER is supported in code but contributes zero rows in this release (compound IDs, not drug names). Labels are merged by clinical priority; each compound has a confidence score Cf* for agreement/conflict. Assay bias is reduced via liver-relevant Tox21 filtering and DrugBank approved-drug negatives. Statistics (v1.0.0) 17,778 compounds (22.47% positive); mean Cf* = 0.603; 85.3% valid SMILES; 1,833 multi-source drugs; 848 label conflicts (46.3% of multi-source). Deposit contents i_dili_final.csv, SCHEMA.md, provenance/benchmark reports and figures, frozen splits, CHECKSUMS.sha256, Python rebuild code (build_idili.py, analysis/, reproduce.ps1), and the data descriptor PDF. Raw source downloads are not included (license/size); users obtain them locally per README.md. License CC-BY-4.0 for derived labels and documentation; third-party SMILES/text subject to source terms (LICENSE_SOURCES.md).