Hash code and stereochemistry

User 173254b396

20-01-2009 13:43:19

It seems that the hash codes of stereoisomers are the same. Is there any way to generate hashcode that reflects stereochemsitry?





Cheers, Péter

ChemAxon 9c0afc9aaf

20-01-2009 19:55:37

Hi Peter,





The hash code currently does not include stereo features indeed, and there is no option for it either.





The main reason is that we implemented this for duplicate filtering, and did not see a substantial benefit for that cause: we assumed that "false duplicates" where all structural feature is the same except stereo would be quite rare, so it would not effect performance a lot.





It is important to note that one must always perform a graph search on structures with equal hash, as different structures might also have the same hash code (though with very low probability).





May I ask the way you intend to use the hash code (if other than duplicate filtering) ?





Or if the aim is duplicate filtering it would be nice to know how many "false hits" you get (where hash code equals but the structures are different) - this can be an indicator of the theoretical maximum amount of speedup available via adding more information to the hash code.





You may also send a mail to me directly or to support if the nature of the answer if confidential.





Best regards,





Szilard