COMPUTATIONAL CHEMISTRY FOR IMPROVED NATURAL COMPOUNDS-TARGET AFFINITY PREDICTIONS

Abstract

The rapid evolution of pathogens underscores an urgent need for accelerated therapeutic development strategies. With an emphasis on natural compounds, this work expands the field of drug repositioning by employing machine learning(ML) techniques to forecast compound-protein interactions that may have therapeutic consequences. Our method makes use of several pre-trained Drug-Target Affinity (DTA) models, such as GraphDTA, MLT-LE, and DeepDTA, to predict binding affinities between protein targets listed in BindingDB and natural products sourced from the COCONUT database. This integration aims to create a robust database facilitating the repurposing of naturally occurring compounds, which are often overlooked in traditional synthetic drug pipelines.