Machine Learning-Based Quantitative Structure-Activity Relationship and ADMET Prediction Models for ERα Activity of Anti-Breast Cancer Drug Candidates

Breast cancer is presently one of the most common malignancies worldwide, with a higher fatality rate. In this study, quantitative structure-activity relationship (QSAR) model compound biological activity and ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) properties prediction were performed using estrogen receptor alpha (ERα) antagonist information collected from samples. We first utilized grey relation analysis (GRA) in conjunction random forest (RF) algorithm to identify top 20 molecular descriptor variables that have greatest influence on activity, then we used Spearman correlation 16 independent variables. Second, QSAR developed based BP neural network (BPNN), genetic optimized (GA-BPNN), support vector regression (SVR). The BPNN, SVR, logistic (LR) models predict substances, impacts each compared assessed. results reveal SVR was prediction, classification properties: predicts Caco-2 hERG(human Ether-a-go-go Related Gene) properties, LR cytochrome P450 enzyme 3A4 subtype (CYP3A4) Micronucleus (MN) BPNN Human Oral Bioavailability (HOB) properties. Finally, entropy theory validate rationality variable screening, sensitivity demonstrates constructed has high accuracy stability, which can be as reference for screening probable active compounds drug discovery.