KinaseDB: an integrated multi-omics platform for the human kinome with disease-specific target prioritization

Protein kinases govern cellular signalling and disease progression and represent major therapeutic targets, yet comprehensive characterization across the human kinome remains hindered by data fragmentation across specialized resources. Here, we present KinaseDB (https://kinasedb.raylab.iiitd.edu.in), a freely accessible web resource integrating curated multi-omics data for 559 human protein kinases from 11 core resources, supplemented by specialized functional-site annotation datasets. KinaseDB provides structural annotations derived from experimental structures, AlphaFold models, and homology modelling; kinome-wide druggability assessments using fpocket; a provenance-aware kinase-substrate dataset comprising site-resolved and relationship-only interactions; cellular-resolution snRNA-seq expression profiles across 625 tissue-cell-type combinations; disease associations with drug-tractability annotations; and population-scale genetic variation data within a single platform. To enable target prioritization, KinaseDB introduces a kinase prioritization score (KPS) that integrates seven complementary evidence components across 69 disease contexts, comprising 38 571 kinase-disease pairs. The KPS discriminated FDA-approved kinase-inhibitor targets from IDG dark kinases with an AUROC of 0.773 (95% bootstrap CI: 0.615-0.908) and an AUPRC of 0.875 (95% bootstrap CI: 0.747-0.968). Rankings were broadly robust to the tested weight perturbations (Spearman $\rho = 0.84$-0.98 across five alternative weighting schemes). All data, bulk downloads, and a documented REST API are publicly available through KinaseDB.