Activity Prediction of Small Molecule Inhibitors for Antirheumatoid Arthritis Targets Based on Artificial Intelligence

• DOI: 10.1021/acscombsci.0c00169
• Publication Date: 2020-11-04T16:14:11Z
• Authors: Guomeng Xing; Li Liang; Chenglong Deng; Yi Hua; Xingye Chen; Yan Yang; Haichun Liu; Tao Lu; Yadong Chen; Yanmin Zhang
• Keywords: Drug target

Rheumatoid arthritis (RA) is a chronic autoimmune disease, which compared to "immortal cancer" in industry. Currently, SYK, BTK, and JAK are the three major targets of protein tyrosine kinase for this disease. According existing research, marketed research drugs RA mostly based on single target, limits their efficacy. Therefore, designing multitarget or dual-target inhibitors provide new insights treatment regarding specific association between from two signal transduction pathways. In study, machine learning (XGBoost, SVM) deep (DNN) models were combined first time build powerful integrated model JAK. The predictive power was proved be superior that classifier. order accurately assess generalization ability model, comprehensive similarity analysis performed training test set, prediction accuracy specifically analyzed under different thresholds. External validation conducted using single-target inhibitors, respectively. Results showed our not only obtained high recall rate (97%) prediction, but also achieved favorable yield (54.4%) prediction. Furthermore, by clustering performance various classes proved, evaluating applicability domain drug screening. summary, proposed promising screen SYK/JAK BTK/JAK as drugs, beneficial clinical rheumatoid arthritis.