As an alternative to virtual screening, the program DOCK6 can be used for de novo design (DOCK_DN) whereby new molecules are constructed from scratch in a protein binding site, in a layer-by-layer fashion, using libraries of drug-like fragments. To help maintain "drug likeness", the current DOCK_DN version allows users to specify several parameters, such as ligand molecular weight, number of rotatable bonds, and number of hydrogen bond donors/acceptors, which are evaluated during growth. In this work, we explore integration of DOCK6 with the widely used cheminformatics program RDKit so that additional drug-based descriptors, such as logP, logS, topological polar surface area, PAINS, quantitative estimate of druglikeness, and synthetic accessibility, among others, can be computed. The objective is to enable greater control, based on user-definable boundary conditions, over compounds as they are assembled and passed to the next layer with the ultimate goal being construction of ensembles with more favorable drug-like properties. In the present work, we outline how RDKit was integrated and validated, which for some routines required translation from Python into C++. Considerable testing and comparisons between the new DOCK6/RDKit implementation (MOL2 format as input) and the standard Python/RDKit distribution (SMILES format as input) suggest that the integration is robust and will be a useful tool to help guide de novo design. Progress to date will be presented.

Descriptor driven de novo design for DOCK6 using RDKIT

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.