In order to fully understand protein kinase networks, new methods are needed to identify regulators and substrates of kinases, especially for weakly expressed proteins. Here we have developed a hybrid computational search algorithm that combines machine learning and expert knowledge to identify kinase docking sites, and used this algorithm to search the human genome for novel MAP kinase substrates and regulators focused on the JNK family of MAP kinases. Predictions were tested by peptide array followed by rigorous biochemical verification with in vitro binding and kinase assays on wild-type and mutant proteins. Using this procedure, we found new 'D-site' class docking sites in previously known JNK substrates (hnRNP-K, PPM1J/PP2Czeta), as well as new JNK-interacting proteins (MLL4, NEIL1). Finally, we identified new D-site-dependent MAPK substrates, including the hedgehog-regulated transcription factors Gli1 and Gli3, suggesting that a direct connection between MAP kinase and hedgehog signaling may occur at the level of these key regulators. These results demonstrate that a genome-wide search for MAP kinase docking sites can be used to find new docking sites and substrates.

MOTIVATION: Small organic molecules, from nucleotides and amino acids to metabolites and drugs, play a fundamental role in chemistry, biology and medicine. As databases of small molecules continue to grow and become more open, it is important to develop the tools to search them efficiently. In order to develop a BLAST-like tool for small molecules, one must first understand the statistical behavior of molecular similarity scores. RESULTS: We develop a new detailed theory of molecular similarity scores that can be applied to a variety of molecular representations and similarity measures. For concreteness, we focus on the most widely used measure--the Tanimoto measure applied to chemical fingerprints. In both the case of empirical fingerprints and fingerprints generated by several stochastic models, we derive accurate approximations for both the distribution and extreme value distribution of similarity scores. These approximation are derived using a ratio of correlated Gaussians approach. The theory enables the calculation of significance scores, such as Z-scores and P-values, and the estimation of the top hits list size. Empirical results obtained using both the random models and real data from the ChemDB database are given to corroborate the theory and show how it can be applied to mine chemical space. AVAILABILITY: Data and related resources are available through http://cdb.ics.uci.edu.