Structural- and cheminformatics of protein kinases
A huge amount of data is available regarding protein kinase inhibitors. ChEMBL lists 1000 "kinase" targets with up to 30000 "bioactivity" data points per target. The Protein Data Bank lists som 6500 crystal structures of protein kinases. Advanced informatics methods are needed to mine this data.
The diversity of structural forms of a single enzyme complicates the understanding of structure-function relationships (left image: green tubes depict the activation loops of the protein kinase Met; upper right image: different activation loop forms have different inhibitor binding cavities). Protein kinases have active and inactive forms, and multivariate statistical methods assist classification (lower right image: PLS or "Projection onto Latent Spaces" analysis of activation structural parameters highlighted for three lung cancer target protein kinases.
Principal Component Analysis of inhibitor-kinase profiling data identifies protein kinase similarities, and highlights especially the unusual nature of the lung cancer target EGFR.