Similarity Search

UniChem Similarity Search is an InChI-based search function that will let you find entries for structures similar to your query structure.

• It is a similarity search using 256 bit Morgan fingerprints and radius=2.
• Fingerprints are created from InChI using Tanimoto as a metric.
• The search requires either an InCHI score or a molecule structure, and a % threshold at which to cut off results.

It will return all chemical structures below that threshold from UniCHem. Clicking on a structure will show hits for that compound within UniChem sources.

Exercise

Try a similarity search using the InChI for yohimbine and the default threshold of 80% similarity

Note the number of structures found.

Figure 8: Similarity search results for yohimbine at the default 80% similarity threshold

Click on one of these to view the UniChem records for that structure

Figure 9: UniChem records for a similarity match to yohimbine.

Now repeat the search with a higher threshold, and you can see that fewer structures match this more stringent requirement.

Figure 10: Similarity search results for yohimbine at a 98% similarity threshold