PyFingerprint: A Python repository from hcji

PyFingerprint

There are many types of chemical fingerprint for describing the molecule provided by different tools, such as RDKit, CDK and OpenBabel. This package aims to summarize them all.

Dependencies

Install

Clone the repo and navigate to it.
Create a predefined Python conda environment by conda env create -f env/pyfingerprint_env.yml.

Activate conda environment and install with pip.

  conda activate pyfingerprint
  pip install git+https://github.com/hcji/PyFingerprint.git

Usage

Fingerprints for single molecule

import numpy as np
from PyFingerprint.fingerprint import get_fingerprint, get_fingerprints

cdktypes = ['standard', 'extended', 'graph', 'maccs', 'pubchem', 'estate', 'hybridization', 'lingo', 
            'klekota-roth', 'shortestpath', 'cdk-substructure', 'circular', 'cdk-atompairs']
rdktypes = ['rdkit', 'morgan', 'rdk-maccs', 'topological-torsion', 'avalon', 'atom-pair', 'rdk-descriptor']
babeltypes = ['fp2', 'fp3', 'fp4', 'spectrophore']
vectypes = ['mol2vec']

smi = 'CCCCN'
output = {}
for f in cdktypes:
    output[f] = get_fingerprint(smi, f)

for f in rdktypes:
    output[f] = get_fingerprint(smi, f)
    
for f in babeltypes:
    output[f] = get_fingerprint(smi, f)
    
for f in vectypes:
    output[f] = get_fingerprint(smi, f)

output_np = output.copy()
for k, fp in output.items():
    output_np[k] = fp.to_numpy()

Fingerprints for multi molecules

smlist = ['CCCCC', 'CCCCN', 'CCCCO']    
output = {}
for f in cdktypes:
    output[f] = get_fingerprints(smlist, f)

for f in rdktypes:
    output[f] = get_fingerprints(smlist, f)
    
for f in babeltypes:
    output[f] = get_fingerprints(smlist, f)
    
for f in vectypes:
    output[f] = get_fingerprints(smlist, f)

output_np = output.copy()
for k, fps in output.items():
    output_np[k] = np.array([fp.to_numpy() for fp in fps])

Cite

@article{doi:10.1021/acs.analchem.0c01450,
author = {Ji, Hongchao and Deng, Hanzi and Lu, Hongmei and Zhang, Zhimin},
title = {Predicting a Molecular Fingerprint from an Electron Ionization Mass Spectrum with Deep Neural Networks},
journal = {Analytical Chemistry},
volume = {92},
number = {13},
pages = {8649-8653},
year = {2020},
doi = {10.1021/acs.analchem.0c01450},
    note ={PMID: 32584545},
URL = { 
        https://doi.org/10.1021/acs.analchem.0c01450
    
}}

Support fingerprint types:

**standard**: Considers paths of a given length. These are hashed fingerprints, with a default length of 1024.
**extended**: Similar to the standard type, but takes rings and atomic properties into account into account.
**graph**: Similar to the standard type by simply considers connectivity.
**hybridization**: Similar to the standard type, but only consider hybridization state.
**estate**: 79 bit fingerprints corresponding to the E-State atom types described by Hall and Kier.
**cdk-atompairs**: CDK's implementation of the atompairs fingerprint
**cdk-substructure**: CDK substructure fingerprint, basically identical to openbabel's fp4.
**pubchem**: 881 bit fingerprints defined by PubChem.
**klekota-roth**: 4860 bit fingerprint defined by Klekota and Roth.
**shortestpath**: A fingerprint based on the shortest paths between pairs of atoms and takes into account ring systems, charges etc.
**rdk-descriptor**: Various molecular descriptors implemented and calculated by RDKit.
**circular**: An implementation of the ECFP6 fingerprint.
**lingo**: An implementation of the LINGO fingerprint.
**rdkit**: Another implementation of a Daylight-like fingerprint by RDKit.
**maccs**: The popular 166 bit MACCS keys described by MDL.
**avalon**: Substructure or similarity Avalon fingerprint.
**atom-pair**: RDKit Atom-Pair fingerprint.
**topological-torsion**: RDKit Topological-Torsion Fingerprint.
**morgan**: RDKit Morgan fingerprint.
**fp2**: OpenBabel FP2 fingerprint, which indexes small molecule fragments based on linear segments of up to 7 atoms in length.
**fp3**: OpenBabel FP3 fingerprint, which is a fingerprint method created from a set of SMARTS patterns defining functional groups.
**fp4**: OpenBabel FP4 fingerprint, which is a fingerprint method created from a set of SMARTS patterns defining functional groups.
**spectrophore** Openbabel implementation of the spectrophore fingerprint (https://github.com/silicos-it/spectrophore).
**mol2vec**: Unsupervised machine learning approach for mulecule representation.

WeChat public account: Chemocoder