Kinases play a significant role in diverse signaling pathways and up-regulations of kinases activity has been implicated in multitude of diseases such as cancers. Consequently, understanding kinase selectivity and target-mediated toxicity remains a top priority for kinase inhibitor design.

Here, we present a knowledge-based approach to profile kinase selectivity based on the similarity between drug binding microenvironments. To allow large-scale kinase site similarity profiling, we have created a kinome structure database consisting of 5000 inhibitor-binding pockets from 187 unique human kinase crystal structures.