Use exact search to find the encoded molecule, substructure search to find compounds containing a chosen motif, and similarity search to explore related molecular features. These answer different questions; a similarity score is not a prediction of biological activity.
Start with the question you want to answer
An exact query is useful when you already know the compound and want to check the catalogue. A substructure query helps when a ring system or functional group must be present but the rest of the molecule can vary. Similarity search is a broader route to analogues when no single motif defines the whole selection.
| Search mode | Question | What to inspect |
|---|---|---|
| Exact | Is this encoded structure listed? | Stereochemistry, charge and components |
| Substructure | Which compounds contain this motif? | Attachment positions and additional groups |
| Similarity | Which structures share fingerprint features? | The complete structure and the fingerprint method |
Check what your drawing actually specifies
For an exact HTS search, encoded stereochemistry, charge and components are preserved. An unspecified stereocentre is not a specification of a particular enantiomer. A salt, free base and different tautomer may therefore need separate queries. A molecular formula alone cannot distinguish structural isomers.
Begin with a concrete SMILES string or an ordinary structure drawing. If a query is unexpectedly empty, inspect aromatic bonds, formal charges, hydrogen counts and stereochemical annotations before broadening it. Do not silently remove chemically important parts just to obtain a match.
Read similarity scores in context
HTS uses OpenChemLib FFP512 fingerprints and Tanimoto similarity. For binary fingerprints, Tanimoto compares shared set bits with all bits set in either fingerprint. The score depends on the representation, so a numerical threshold from another toolkit may produce a different neighbourhood.
Try a stricter threshold first, then widen it if you need more diversity. Inspect each hit rather than assuming the closest score gives the best synthetic or biological choice. Fingerprint similarity does not establish target binding, selectivity, solubility or cellular activity.
Sources: OpenChemLib: official toolkit source and documentation
A useful HTS search workflow
Calculated descriptors support prioritisation; they are not experimental assay results. Availability and documentation are confirmed with the quotation.
- Select in-stock chemistry for an available-stock shortlist, or include synthesis-required entries to explore the virtual library.
- Choose exact, substructure or similarity mode and enter the drawing or SMILES.
- Combine your query with molecular-weight, cLogP or tPSA filters when appropriate for the project.
- For broad substructure searches, continue loading matches while candidate verification is still in progress.
- Open compound records, keep promising entries in your inquiry list and send the IDs with the quantities you need.
Common questions
Can two different SMILES describe the same molecule?
Yes. SMILES can traverse the same molecular graph in different orders. Compare parsed structures rather than text strings, and keep charge and stereochemistry consistent.
Does a 90% similarity score mean 90% identical activity?
No. It describes overlap under a particular fingerprint and similarity metric, not a proportion of biological activity.
Sources & further reading
YOUR NEXT STEP
Find chemistry to evaluate.
Explore the catalogue or send your IDs, quantities and specifications for a tailored quotation.
Catalogue information is distinct from experimental activity data. Counts refer to entries; categories can overlap. Availability, documentation and synthesis feasibility are confirmed individually. Questions or corrections? Email the team.