COMPASITES - Computer-aided active site analysis of protein structures

The thesis in hand comprises the general part of my cumulative dissertation, sub-mitted to the University of Hamburg in September 2012. It describes my scientificwork on computer strategies for structure-based active site analysis in the group forComputational Molecular Design of the Center for Bioinformatics from June 2008 tillSeptember 2012. Furthermore, this manuscript contains six scientific journal publica-tions, listed in a separate bibliography at the end of this dissertation and cited in thetext with D1 D6.Molecular recognition and binding of small molecules to a protein is the foundation forthe maintenance of biological systems. In this context, the 3D structure of the protein,more precisely, the structure of the active site plays a fundamental role. Understandingand modifying the mechanism of ligand binding is of high practical interest in pharma-ceutical and biotechnological research. Due to the growing number of available threedimensional protein structures, efficient computational methods are needed to assistexperimental approaches.In my work, I designed several strategies addressing different parts of the structure-based computer-aided active site analysis cycle. Since the active site of a protein isthe key to its function, the first step in my work was the development of a novel algo-rithm for binding site detection. For predicted sites, several global and local descriptorscan subsequently be calculated and used for protein assessment. For protein classifica-tion, the descriptors are incorporated into hierarchical clustering, machine learning andnearest neighbor search techniques. These classifiers can be used to prioritize poten-tial disease modifying proteins in drug development processes, and to annotate proteinfunction. In a subsequent study, the shape complementarity requirement for molecularrecognition has been explored and numerically registered.Since many approaches suffer from the rigid modeling of the naturally flexible structureof proteins, I participated in establishing an approach for active site comparison basedon triangle descriptors of the active site, accounting for small changes in the bindingsite.