Fig. 1
From: Large-scale comparison of machine learning methods for profiling prediction of kinase inhibitors
The scheme and workflow of this work. A Dataset collection. B Models construction and selecting the optimal model for the kinase profiling prediction task. C ML methods for the construction of fingerprint- and RDKitDes-based models. D DL methods for the construction of graph-based models. (E) Combined-features- and fusion-based models construction