Classification and Screening of Recognition Protein Complexes with Genetic Programming

A protein complex is either transient or obligate. In the organisms, proteins recognize each other with their binding sites and become transient protein complexes. Proteins perform their functions through protein-protein recognition. In this paper, we train a model by genetic programming with physicochemical properties information of the binding sites. The model classifies and screens the proteins of recognition interactions. The model achieves an average classification accuracy of 75%. For screening, given a protein with concave binding site and a database of proteins with convex binding site, from the database we retrieve a small set of proteins in which one of them recognizes with the query protein. In average, 65% of proteins in the database are screened out.