Hi Payal,
The function you are looking for is DataStructs.ConvertToNumpyArray. Here's
a short demo:
In [3]: from rdkit import DataStructs
In [4]: from rdkit.Chem import rdMolDescriptors
In [5]: fpLen = 2048
In [6]: fp =
rdMolDescriptors.GetHashedAtomPairFingerprintAsBitVect(Chem.MolFromSmiles('CCCC'),fpLen)
In [7]: import numpy as np
In [9]: v = np.array(fpLen,np.short)
In [10]: DataStructs.ConvertToNumpyArray(fp,v)
In [11]: v
Out[11]: array([0, 0, 0, ..., 0, 0, 0], dtype=int16)
In [12]: sum(v)
Out[12]: 6
I hope this helps,
-greg
On Mon, Oct 22, 2018 at 4:29 PM Payal Chatterjee <
payalchatterjee...@gmail.com> wrote:
> Hi
>
> I am using rdkit to fetch atom pair information as AtomPairFingerprint. I
> am reading from a list of molecules as individual SDF. Calling
> GetHashedAtomPairFingerprintAsBitVect on the object "m" (where m is read as
> each molecule), object strings are returned.
>
> trainResi=open(rootpath + 'trialResi.list').readlines()
> trainResi = [x.strip() for x in trainResi]
> print len(trainResi)
>
> for name in trainResi:
> for ligname in glob.glob(os.path.join(modelDir+name+'.sdf'))[:]:
> m = Chem.MolFromMolFile(ligname, sanitize=False, removeHs=False)
>
> for atm in m.GetAtoms():
> aid = atm.GetIdx()
> m.UpdatePropertyCache(strict=False) ##Required because I
> want to include charged molecules too!
> fp =
> AllChem.GetHashedAtomPairFingerprintAsBitVect(m,maxLength=APLength, \
>
> fromAtoms=[aid])
> print fp
>
> * "fp" are returned as (including only one example) :*
> <rdkit.DataStructs.cDataStructs.ExplicitBitVect object at 0x11bfc4d70>
>
> For my code to work further, I am converting my fp into a numpy array,
> which returns [0 0 0 ..., 0 0 0] (zero values)
> What am I doing wrong?
>
> I am new to rdkit and machine learning, and I am using jupyter notebook.
>
> Thanks
> Payal
> _______________________________________________
> Rdkit-discuss mailing list
> Rdkit-discuss@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
>
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
