Structure-Based\nQSAR Modeling of RET Kinase Inhibitors\nfrom 49 Different 5,6-Fused Bicyclic Heteroaromatic Cores to Patent-Driven\nValidation

RET receptor tyrosine kinase is crucial for nerve and\ntissue development\nbut can be an important oncogenic driver. This study focuses on exploring\nthe design principles of potent RET inhibitors through molecular docking\nand 3D-QSAR modeling of 5,6-fused bicyclic heteroaromatic derivatives.\nFirst of all, RET inhibitors of 49 different bicyclic substructures\nwere collected from five different data sources and selected through\nmolecular docking simulations. QSAR models were built from the 3399\nconformers of 952 RET inhibitors using the partial least-squares method\nand statistically evaluated. The optimal QSAR model exhibited high predictive performance,\nwith <i>R</i><sup>2</sup> (of training data) and <i>Q</i><sup>2</sup> (of test data) values of 0.801 and 0.794,\nrespectively, effectively predicting known inhibitors. The optimal\nmodel was doubly verified by patent-filed RET inhibitors as the out-of-set\ndata to demonstrate acceptable residual analysis results. Moreover,\nfeature importance analysis of the QSAR model outlined the impact\nof substituent characteristics on the inhibitory activity within the\n5,6-fused bicyclic heteroaromatic core structures. Furthermore, the\nrelationship between structure and inhibitory activity was successfully\napplied to the RET screening of known clinical and nonclinical kinase\ninhibitors to afford accurate off-target prediction.