tooluniverse-small-molecule-discovery

Use ChEMBL_search_activities with the molecule or target ChEMBL ID and filter with pchembl_value__gte=6 for sub-micromolar potency. BindingDB_get_ligands_by_uniprot is an alternative but frequently times out, so ChEMBL is the reliable fallback.

Pass the SMILES to SwissADME_calculate_adme for physicochemical, pharmacokinetic, and drug-likeness data, or use ADMETAI_predict_* tools for ML-based toxicity, CYP, and BBB predictions. SwissADME takes a string; ADMET-AI requires a list.

SwissADME provides rule-based drug-likeness scores like Lipinski and Veber plus general ADME properties, while ADMET-AI offers ML predictions for hERG, DILI, CYP interactions, and bioavailability. ADMET-AI requires the tooluniverse[ml] extra; SwissADME is the fallback.

Yes, eMolecules_search and Enamine_search_catalog check vendor availability by name or SMILES. Note these tools often return search URLs rather than live data, so results are presented as links to search manually.

The BindingDB REST API frequently times out or is unavailable. Use ChEMBL_get_target_activities or ChEMBL_search_activities instead, which provide equivalent Ki and IC50 data with literature references.

SwissTargetPrediction relies on structure similarity to known drug-target pairs and may time out on complex molecules. As a fallback, run ChEMBL similarity searches and infer targets from known activities of similar compounds.