Parallelized Tools for the Preparation and Curation of Large Libraries for Virtual Screening

Abstract

We introduce a software package, ParaLig, that provides solutions for several workflows occurring repeatedly in computational drug discovery: parameterization of small molecules with partial charges and free energies of solvation, generation of conformers, virtual chemical reactions, creating combinatorial libraries, and molecule editing tasks. Throughout, we emphasize the maintenance/creation of 3D coordinates and better interoperability. The latter is achieved by stable embedding of meta-information into output files and by improving the mutual compatibility of molecular representations (e.g., aromaticity perception). ParaLig wraps around core functionalities provided by a variety of existing software: the two popular cheminformatics packages OpenBabel and RDKit, and the general force field providers CGenFF and AmberTools. Our workflows emphasize scalable bulk processing of large libraries of molecules, and we provide an MPI-based wrapper to simplify deployment to high-performance computing resources. Along with salient descriptions of the methods, we benchmark performance both in terms of throughput and in terms of the quality of some of the results.