[Rdkit-discuss] Gobbi hydrogen bond acceptor definition

[Konrad K. <kon...@ic...>]

Hi All,

The following is a relatively minor point, but IMHO, the Gobbi hydrogen bond acceptor definition probably should be modified so that only aromatic nitrogen atoms with two and not three non-hydrogen attachments are considered as acceptors.  For example, the aromatic nitrogen atom pyridine is an acceptor whereas the nitrogen in indole or N-methylindole is not.  Also fluorine atoms (see https://doi.org/10.1016/0040-4020(96)00749-1 <http://dx.doi.org/10.1016/0040-4020(96)00749-1>) do not appear to be good acceptors.  In addition, thiophenes are very questionable acceptors.

Hence in: 

in ~/envs/my-rdkit-env/lib/python2.7/site-packages/rdkit/Chem/Pharm2D/Gobbi_Pharm2D.py

I have made the following modification:

# Replaced the following line to the next line:
# DefineFeature Acceptor [$([O,S;H1;v2]-[!$(*=[O,N,P,S])]),$([O,S;H0;v2]),$([O,S;-]),$([N&v3;H1,H2]-[!$(*=[O,N,P,S])]),$([N;v3;H0]),$([n,o,s;+0]),F]
DefineFeature Acceptor [$([O,S;H1;v2]-[!$(*=[O,N,P,S])]),$([O,S;H0;v2]),$([O,S;-]),$([N&v3;H1,H2]-[!$(*=[O,N,P,S])]),$([N;v3;H0]),$([n;D2;H0;+0]),$([o;+0]]

Cheers,

Konrad

After modifying Gobbi_Pharm2D.py, the following test script gives:

from rdkit import Chem, DataStructs
from rdkit.Chem import ChemicalFeatures
from rdkit.Chem.Pharm2D.SigFactory import SigFactory
from rdkit.Chem.Pharm2D import Gobbi_Pharm2D,Generate

mol = Chem.MolFromSmiles('Cn1ccc2ccccc21’) # N-methylindole

bi = {}
fp = Generate.Gen2DFingerprint(mol,Gobbi_Pharm2D.factory,bitInfo=bi)

onbits = fp.GetOnBits()
for onbit in onbits:
    print onbit, Gobbi_Pharm2D.factory.GetBitDescription(onbit), bi[onbit]

output:

77 AR HD |0 0|0 0| [[(3,), (0,)]]

mol = Chem.MolFromSmiles('c1ncccc1C’) # 2-methylpyridine

output:

63 AR BG |0 0|0 0| [[(5,), (1,)]]
70 AR HA |0 0|0 0| [[(5,), (1,)]]