Modeling of synthetic accessibility of potential drug molecules containing five-membered aromatic heterocycles

Abstract

A new cheminformatics approach was proposed for modeling synthetic accessibility (SA), which is one of the most critical problems in computer-aided design of novel drugs. The approach is based on filtering molecular structures that contain synthetically irrelevant substructures using the hierarchically organized library of SMARTS substructures, which considers the fragment’s local environment. It compensates for the drawbacks of the existing SA modeling methods. The proposed approach was validated using five-membered aromatic heterocycles. Applying this algorithm significantly reduces the number of false positive predictions commonly encountered when employing traditional SA modeling methods. In its current form, the proposed algorithm can serve as an efficient additional filtering layer. A hierarchical substructure filtering algorithm has been described for the first time, along with a novel method for the automated merging of SMARTS patterns based on atomic primitives.