A Hybrid Structure-Based Machine Learning Approach for Predicting Kinase Inhibition by Small Molecules

• DOI: 10.1021/acs.jcim.3c00347
• Publication Date: 2023-08-18T18:26:08Z
• Authors: Changchang Liu; Peter Kutchukian; Nhan D. Nguyen; Mohammed AlQuraishi; Peter K. Sorger
• Keywords: Kinome

Kinases have been the focus of drug discovery programs for three decades leading to over 70 therapeutic kinase inhibitors and biophysical affinity measurements 130,000 kinase-compound pairs. Nonetheless, precise target spectrum many kinases remains only partly understood. In this study, we describe a computational approach unlocking qualitative quantitative kinome-wide binding structure-based machine learning. Our study has components: (i) Kinase Inhibitor Complex (KinCo) data set comprising in silico predicted structures paired with experimental constants, (ii) learning loss function that integrates model training, (iii) trained on KinCo. We show our outperforms methods crystal alone predicting binary interaction affinities; relative structure-free methods, also captures known biochemistry more successfully generalizes distant sequences compound scaffolds.