The study and prediction of kinase function (kinomics) is of major importance for proteome research dueto the widespread distribution of kinases. However, theprediction of protein function based on the similaritybetween a functionally annotated 3D template and a querystructure may fail, for instance, if a similar proteinstructure cannot be identified. Alternatively, function canbe assigned using 3D-structural empirical parameters. Inprevious studies, we introduced parameters based onelectrostatic entropy (
Proteins 2004,
56, 715) and molecular vibration entropy (
Bioinformatics 2003,
19, 2079) butignored other important factors such as van der Waals(vdw) interactions. In the work described here, we define3D-vdw entropies (
k) and use them for the first time toderive a classifier for protein kinases. The model classifiescorrectly 88.0% of proteins in training and more than85.0% of proteins in validation studies. Principal components analysis of heterogeneous proteins demonstratedthat
k codify information that is different to that described by other bulk or folding parameters. In additionalvalidation experiments, the model recognized 129 out of142 kinases (90.8%) and 592 out of 677 non-kinases(87.4%) not used above. This study provides a basis forfurther consideration of
k as parameters for the empirical search for structure-function relationships.