Stereochemistry (tetrahedral, double-bond) options?

[chembl]

Stereochemistry (tetrahedral, double-bond) options?

[eloyfelix]

Hi Andrew,

FPSim2 is using this RDKit function to generate Morgan fingerprints:

http://rdkit.org/docs/source/rdkit.Chem.rdMolDescriptors.html#rdkit.Chem.rdMolDescriptors.GetMorganFingerprintAsBitVect

which means that you could use the useChirality flag which is set as False by default when creating the fingerprints file
{'radius': 2, 'nBits': 2048, 'useChirality': True}

hope it helps

[someonetookmyname]

Thank you for the response.

Is it ok to use Morgan fingerprints for substructure searches? https://chembl.github.io/FPSim2/source/user_guide/subs.html recommends to use RDKPatternFingerprints, but I cannot find a similar "useChirality" option for that fingerprint type?

Thank you.

[eloyfelix]

Be aware that the substructure function in FPSim2 does not run a full substructure search. It is mentioned in the documentation and you coud bet a getter idea of what you'll get from it in Greg's blogpost: http://rdkit.blogspot.com/2013/11/fingerprint-based-substructure.html TL;DR section.

There is no such parameter in RDKit for the RDKPatternFingerprints. It is recommended by FPSim2 because is the best option in RDKit :) I didn't experimented with Morgan but I'm afraid they can have much worse accuracy than the specificaly designed fingerprints.

[someonetookmyname]

Thank you for this extra information!