Molecular Fingerprinting

author: shiyu

Desription

Molecular Fingerprinting encodes a Simplified Molecular Input Line Entry Specification (SMILES) as a fingerprint. The fingerprint can represent elements, atom pairs, or functional groups, etc., and are often used for substructure searches and similarity searches in drug discovery.

This operator uses RDKit to generate the molecular fingerprint.

Code Example

Before running the following code, you need to install rdkit, refer to https://www.rdkit.org/docs/Install.html.

# install rdkit with conda
$ conda install -c conda-forge rdkit

An example that use the Morgan algorithm to generate a fingerprint of the molecular formula 'Cc1ccc(cc1)S(=O)(=O)N'.

Write the pipeline in simplified style:

import towhee

towhee.dc(['Cc1ccc(cc1)S(=O)(=O)N']) \
  .molecular_fingerprinting.rdkit() \
  .show()

Write a same pipeline with explicit inputs/outputs name specifications:

import towhee

towhee.dc['smiles'](['Cc1ccc(cc1)S(=O)(=O)N']) \
  .molecular_fingerprinting.rdkit['smiles', 'fingerprint']() \
  .show()

Factory Constructor

Create the operator via the following factory method:

molecular_fingerprinting.rdkit( algorithm: str = 'morgan', size: int = 2048)

Parameters:

algorithm: str

Which algorithm to use for fingerprinting, including 'morgan', 'daylight', 'ap', 'maccs', defaluts to 'morgan', and there is the list of available fingerprints.

size: int

The bit vector size just for morgan and daylight algorithm, defaults to 2048.

Interface

An molecular fingerprinting operator takes a SMILES as input. It uses the RDKit specified by algorithm name to generate a SMILES fingerprint.

Parameters:

smiles: str

A Simplified Molecular Input Line Entry Specification (SMILES).

Returns: bytes

The molecular fingerprint.