Additional file 2 of Large-scale comparison of machine learning methods for profiling prediction of kinase inhibitors

Additional file 2: Table S1. Details on benchmark dataset for kinase profiling prediction task used in this study. Table S2. Structural diversity and chemical space analysis of the compounds in each kinase. Table S3. Detailed performance results of different ML methods. Table S4. Detailed individual kinases where the GCN models outperform the RF::RDKitDes models. Table S5. Detailed individual kinases where the FP-GNN models outperform the RF::RDKitDes models. Table S6. The optimal in silico predictive models for each kinase in terms of AUC metric. Table S7. Comparison performance of models based on combined features and single feature in terms of F1 score. Table S8. Ranking of all single models by AUC values. Table S9. Comparison of our models with the reported in silico prediction models for kinase profiling prediction task. Table S10. The predicted activity probability and experimental % activity of CHMFL-BMX-078.