Off-Target Predictor

Rank the human protein targets each molecule most resembles known actives of. For every target it reports a calibrated probability, the highest ECFP4 Tanimoto similarity to that target's recorded ChEMBL actives, how many actives stand behind the target, and whether the molecule lies within the chemistry the reference covers. Results are hypotheses ranked by similarity to recorded actives, not predicted activity at any dose, and a target without a recorded active is not evidence of inactivity. Reference data: ChEMBL (EMBL-EBI), CC BY-SA 3.0.
0 molecules ready
Estimated minimum cost: $—

Examples

Molecules

Paste SMILES or InChI one per line, or mol blocks / SDF records. At most 10,000 molecules. A salt or mixture is scored on its largest fragment, the way the reference compounds were prepared; the results list the molecule as you entered it.

or drop a file here

Ranking options

Results are hypotheses ranked by similarity to recorded actives, not predicted activity.

Limit to protein families (optional): All families

Only targets in the selected ChEMBL protein families are ranked. A broad family such as Enzyme also includes its subfamilies (Kinase, Protease, ...). Leave all unchecked to consider every family.

Reference data: ChEMBL (EMBL-EBI), CC BY-SA 3.0 Unported, reduced to fingerprints and target links and modified by CogniChem. The ChEMBL release used is recorded in each result. Target accessions are from UniProt, CC BY 4.0. Zdrazil B, et al. The ChEMBL Database in 2023. Nucleic Acids Res. 2024;52:D1180–D1192. doi:10.1093/nar/gkad1004