Ligand-based machine learning models to classify active compounds for prostaglandin EP2 receptor

Prostaglandin receptors are pharmacologically validated targets with implications in several medical indications, including glaucoma, cardiac cyanotic disease, pulmonary hypertension, oncology, and various rare diseases. In this study, we developed a ligand-based machine learning (ML) model to classify chemical compounds as either active or inactive against the prostaglandin receptor EP2. From an initial set of 1,826 descriptors, 20 were selected to train random forest algorithms, yielding an area under the curve score (AUC) of > 0.8 for compound classification in the test set. Our resulting ML classifier showed an overall accuracy of 88.9% towards newly experimentally tested EP2 ligands. This adaptable and tractable workflow can be extended to other EP receptors and possibly other similar targets.