Fig. 1

Performance and classification features. a Performance of the classifier respect to the number of trees in the random forest; b idem, close-up on the region around the performance values; c Number of variants in each kinase group; d log odds-ratio of the number of variants in each kinase group; e Number of variants in each kinase domain; f log odds-ratio of the number of variants in each kinase domain; g changes in Cbeta-branching caused by pathogenic and neutral variants; h number of pathogenic and neutral variants affecting catalytic sites as defined by UniProt, FireDB and Phospho.ELM. i Distribution of SIFT scores; j Changes in volume caused by disease-associated and neutral variants; k Changes in hydrophobicity caused by disease-associated and neutral variants; l Accumulated Gene Ontology (GO) log odds-ratio. Note that, where relevant, disease-associated variants were represented in dark red whereas ochre was used for their neutral counterparts