Tanimoto Similarities: Python script to perform fingerprinting and calculate Tanimoto similarities on multiple compounds using RDKit.

Introduction

tanimoto_similarities.py script calculates Tanimoto similarities of given molecules in the form of smiles.

Provide the input in the form a CSV file consisting of all smiles provided in a column under the name, "SMILES". The file can contain other information also.

This script will calculate similarities and save them in the form of text files and heatmaps. Generated heatmaps will help you visualize the matrix. Sample smiles are provided in the 'smiles.csv' file.

Another script tanimoto_similarities_one_vs_all.py calculates tanimoto similarities of one compound with the other compounds. The query compound is supplied in the form of SMILES and the rest of the target compounds are inputted in the form of sdf provided as a single sdf file.

sdftosmi.py script will convert molecules in sdf format into their corresponding SMILES. All molecules are provided in a single SDF file as input.

smitostr.py will convert smiles to structure.

Requirements

It requires Python3. This script uses RDKit and some additional packages. Install them using the following commands.

conda create -c conda-forge -n my-rdkit-env rdkit

pip3 install seaborn

sudo apt-get install python3-matplotlib

conda install pandas

pip3 install numpy

Usage

This script consists of two functions. One function calculates the similarity matrix and shows the usual heatmap and saves the output file as 'similarities.txt'. The other function calculates the similarity matrix as a lower triangular matrix and saves the output file as 'similarities_lower_tri.txt'. Run the scripts as: $ python3 tanimoto_similarities.py

OR

$ python3 tanimoto_similarities_one_vs_all.py

For more information on this script, read this article: https://bioinformaticsreview.com/20220608/tanimoto_similarities-py-a-python-script-to-calculate-tanimoto-similarities-of-multiple-compounds-using-rdkit/