Most teams profile late, after lead selection — when failures are expensive
Without model queues, compute allocation, or waiting for a batch job to finish overnight. 122 million compounds already profiled. Novel compounds profiled on the fly with 41 ML models — in under 50 milliseconds.
“Profile this compound library for hERG liability, CYP3A4 inhibition, and aqueous solubility.”
How it works
Submit a SMILES or compound name
Known compounds return pre-computed profiles in under 50ms from the 122M database. Novel compounds are profiled on the fly through 41 ML models — CYP450, toxicity, nuclear receptors, stress response.
Full profile returned inline
84 properties per compound: physicochemical descriptors, drug-likeness rules (Lipinski, Veber, Ghose, Egan, Muegge), QED, SA score, pKa, solubility, and BDE for metabolic soft spots.
Batch or single — same speed
Profile up to 100 compounds in one call. Screen libraries up to 1,000 with filtering. Every result includes FAVES compliance inline.
Proof
NovoExpert-2 SOTA on 5 TDC ADMET endpoints: CYP2D6 Veith 0.778 AUPRC · CYP3A4 Veith 0.916 AUPRC · CYP3A4 Substrate 0.648 AUPRC · Clearance Hepatocyte 0.602 Spearman · DILI 0.922 AUROC.
pKa v9: custom Chemprop D-MPNN on 22,726 IUPAC measurements. SAMPL7 blind test RMSE 0.535. SAMPL8 RMSE 1.255, R² 0.878.
122M compounds in Cosmos DB. 84 fields per molecule. Redis SMILES index: 93M keys, O(1) lookup.
Profile any molecule — in under a second
122 million pre-computed. Novel compounds on the fly. Sign up and profile your first molecule.