Integrating molecular docking, CoMFA analysis, and machine-learning classification with virtual screening toward identification of novel scaffolds as Plasmodium falciparum enoyl acyl carrier protein reductase inhibitor

Abstract

In the present study, an integrated application of various in silico methods including molecular docking, CoMFA analyses, and machine-learning classification methods were used on a set of known triclosan and rhodanine inhibitors of Plasmodium falciparum enoyl acyl carrier protein reductase (PfENR) and Maybridge compound database with the prime objective of implementation of knowledge-based synergistic approach toward prioritized screening and identification of novel scaffolds as PfENR inhibitors. Ensemble of CoMFA models were build with excellent values of statistical matrices where genetic algorithm was applied in a conformation selection step on a dataset clustered over chemical space to ensure high degree of structural variability with correlation. Two-dimensional and three-dimensional descriptors were used vigorously to classify actives from inactives by extracting useful correlation among selected descriptors. Nevertheless, the entire hypothesis was pipelined sequentially in the pursuit of proposing probable actives from Maybridge database on the basis of docking, machine learning classification, and CoMFA studies. After ADME filtering, a set of 26 compounds from the Maybridge database were finally predicted to be plausible inhibitors of PfENR satisfying multiple computational validation models.