overlap analysis - different results for various threshold

User 8d9e177eee

09-03-2012 10:47:29


I have a question regarding overlap analysis. I use a search mode Similarity with Tanimoto metric (Instant JChem 5.4)
However, I got different results for various number of max hits and similarity threshold values.

I understand, that overlap hits may be different but shouldn't be overlap max values the same? Now, it is different even for the same threshold but for different number of max hits... why's that?

Thanks in advance for any help.

ChemAxon fa971619eb

12-03-2012 09:16:31

Hi, I tried to reproduce this in a recent version of IJC (I used 5.8.2) but could not see what you are reporting. The similarity scores seem consistent across multiple settings.

You mention that you are using 5.4. Are you able to try to update and see if the problem still exists?

(we would recommend backing up the database before updating).


ChemAxon 2bdd02d1e5

12-03-2012 14:55:00

The search of a query structure in target table ends when it founds number of structures specified by Max hits count. E.g. specify Max hits to 2 and run the search...Overlap count will never be grater than 2, some structures where not found becouse the searches was stopped after 2 hits. 

Therefore Overlap max number must not be the same since it tells the similarity of the most similar structure from all Overlap hits(and some has not been found).


User 8d9e177eee

12-03-2012 17:26:52

Thank you so much for your answer - it explains all the problems with sim.search.