rdkit-cpp-utils

Description

rdkit is the goto toolkit for anything related to cheminformatics and data science applications in Chemistry. It is mostly used in python which shouldn't be uderestimated performance-wise. Still for large amounts of data getting some help from C++ which is the native language of rdkit makes sense. A couple of good sources to start are: https://www.rdkit.org/docs/GettingStartedInC%2B%2B.html and https://github.com/iwatobipen/rdkit_cpp/tree/main. The aim of this repo is to collect some standard functionality in C++ when increased efficiency is required.

Installation

This is the hard and less fun part. This is what worked for me:

1. Compile rdkit from source following https://greglandrum.github.io/rdkit-blog/posts/2023-03-17-setting-up-a-cxx-dev-env2.html One can also look at https://github.com/rdkit/rdkit/tree/master/.azure-pipelines. After a successful set-up you should have a conda environment and all the environmental variables set as described in the links.
2. Clone the repository, and activate the conda environment that was created in Step 1

   git clone https://github.com/Arhs99/rdkit-cpp-utils.git
   cd rdkit-cpp-utils
   conda activate py310_rdkit_build
   

3. Environmental variables: At least RDBASE and LD_LIBRARY_PATH should be set, if not:

export LD_LIBRARY_PATH=${RDBASE}/lib:${CONDA_PREFIX}/lib:${LD_LIBRARY_PATH}

5. Create a build directory and run to compile and link:

cmake ..
make

This should create the executable morganfp

Usage

One can use either morganfp or fingerprint.py to calculate Morgan fingerprints from an sdf file and store them as numpy array files *.npy. Additional arguments are the fingerprint radius and the number of bits

C++ example

Run as:

./morganfp ../data/ChEMBL_set.sdf  ../data/arr_cpp.npy 3 2048

python example

python fingerprints.py data/ChEMBL_set.sdf  data/arr_py.npy 3 2048

Acknowledgment

The files cnpy.cpp and cnpy.h used for loading/saving C++ data as numpy .npy files were copied from the cnpy library https://github.com/rogersce/cnpy