tooluniverse-drug-repurposing

Start with OpenTargets to get genetically associated targets for the disease, then query DGIdb and DrugBank for approved drugs hitting those targets. Score candidates on target association, safety profile, literature evidence, and drug properties using the 0-100 viability framework.

The workflow integrates OpenTargets for disease-target associations, DrugBank and DGIdb for drug-target interactions, ChEMBL for bioactivity data, FDA and FAERS for safety profiles, and PubMed plus ClinicalTrials.gov for evidence. All are accessed through the ToolUniverse SDK.

Compare the drug's IC50 or Ki values for the new target against its original target using ChEMBL bioactivity data. If the new target requires more than 10x the concentration achievable at approved doses, the candidate is unlikely to be clinically feasible.

Disease lookups fail when using non-standard terminology or synonyms not in the EFO ontology. Try alternate disease names, broader disease categories, or query directly with a known EFO identifier instead of the disease name.

Yes, adverse events can indicate therapeutic effects in different contexts, such as weight loss side effects suggesting obesity treatment potential. Query FAERS reaction counts by drug, then filter for approved drugs with acceptable serious-event rates.

Network proximity and target overlap alone do not establish efficacy; mechanism rationale and dose feasibility are required. Computational candidates need experimental validation, and existing clinical trial evidence should always be checked first before prioritizing predictions.