Virtual Screening of HMG-CoA reductase inhibitors of West Bali National Park natural compounds database using machine learning

Abstract

Atherosclerosis is one of the causes of cardiovascular disease (CVD). The high level of cholesterol which is controlled by 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase plays an essential role in the pathogenesis of atherosclerosis. By inhibiting the activity of HMG-CoA reductase, the biosynthesis of cholesterol may be limited and therefore contribute to the reduction of blood cholesterol. This research aims to identify the hit compounds of HMG-CoA reductase inhibitors from the natural compounds database of West Bali National Park from the Internal PRBBOT BRIN database (Indonesia's natural compounds data base). We conducted a virtual screening workflow using a quantitative structure-activity relationship (QSAR) strategy based on artificial intelligence approach to allow faster screening of HMG-CoA reductase inhibitor from 2608 compounds. Eight classifications and five regressions in machine learning algorithms were applied to build a virtual screening workflow using the 1173 compounds dataset from the ChEMBL database. The classification QSAR model used the Random Forest and Fuzzy Rule algorithm with a tied score of accuracy were 0.972 and the regression QSAR model used the Tree Ensemble algorithm with the R2 pred = 0.88. Virtual screening results identified three hit compounds as HMG-CoA reductase inhibitors from Calophyllum inophyllum L., including Inocalophyllin B, Brasiliensic acid, and Inophylloidic acid. These results indicated the benefit of the machine learning approaches for potential screening compounds as an inhibitor for the HMG-CoA reductase enzyme, and it may be useful to screen various drug candidates for other target diseases.