Step 1
Private scientific staging
Prioritize candidate protein targets from a molecular structure.
Submit one SMILES representation to receive a rank-ordered Holmes v74 target-priority list.
Step 2
Prediction status
Ready for a private request
All 6,155 canonical targets are evaluated before the final top candidate list is selected in real predictor mode.
Completed private result
Top 50 candidate targets
Ranks 1–50 are selected only after all 6,155 canonical targets have been evaluated and ranked.
How to interpret Final Score
Final Score is the Holmes v74 score used to rank candidate targets for the submitted structure. Higher values indicate stronger ligand-library similarity evidence in this model. It is not a probability of binding and does not estimate binding affinity, potency, efficacy, safety, or clinical effect. Absolute scores should not be compared across different submitted compounds unless cross-compound comparability has been validated.
No validated universal cutoff currently defines a “good” Final Score. Interpret Final Score together with rank and the model’s held-out recovery evidence; higher-ranked targets have stronger relative support within that prediction.
| Rank | Protein Name | Abbreviated Protein Name | Organism | ChEMBL ID | UniProt | Final Score |
|---|
Method
How Holmes v74 prioritizes targets
Holmes v74 compares standardized two-dimensional molecular representations with a curated target–ligand evidence library and applies a fixed, target-neutral calibration to rank candidate protein targets. The output is a computational prioritization list, not experimental evidence of binding or biological activity.
Limitations
Interpret scores carefully
Scores rank ligand-library similarity evidence and are not probabilities of binding, potency, safety, or clinical effect. Targets with limited reference data may be underrepresented. Absence from the top candidate list does not demonstrate absence of interaction.
Contact
Contact LigandLens
For feedback, questions, or large screenings contact me at twholmesusn@gmail.com.
Software citation
Cite LigandLens
If you use LigandLens in research, please cite the software:
Holmes, Thomas W. (2026). LigandLens: Computational target prioritization with Holmes v74 (Version 74) [Computer software]. https://ligandlens.com
This is a software citation, not a peer-reviewed paper.