Structural genomics is a word-wide initiative aimed at producing a comprehensive mapping of protein structure space. The resulting knowledge of the protein structures will be vitally important for understanding and manipulating their biochemical and cellular functions. One of the main challenges this initiative faces it that only about 2-10% of pursued protein targets yield high-resolution protein structures. A significant bottleneck in acquiring the structures is the ability to obtain diffraction-quality crystals. The application of current protocols yields crystals for approximately 30% of the input proteins and well-diffracting crystals for even a smaller fraction. This motivated the development of models that can be used to either support or directly predict protein crystallization. Several in-silico methods, including SECRET (Proteins 62:343-55, 2006), OB-score (FEBS Lett. 580:4005-9, 2006), CRYSTALP (BBRC 355:764-9, 2007), and ParCrys (Bioinformatics 24:901-7, 2008), which predict crystallization propensity using the protein sequence as the input have recently been proposed. These methods account only for intra-molecular factors that are encoded in the protein chain, while ignoring inter-molecular factors such as protein-protein, protein-ligand, and/or protein-precipitant interactions, buffer composition, precipitant diffusion method, gravity, etc. In spite of this significant limitation the above methods were shown to succeed in providing useful predictions. In this talk, we will overview the current state-of-the-art in sequence-based prediction of protein crystallization propensity. We will describe our current findings that support the claim that the crystallization propensity can be predicted directly from the protein chain and we will also introduce our new predictor, CRYSTALP2.

Dr. Lukasz Kurgan received the M.Sc. degree (with honors) from the AGH University of Science and Technology, Krakow, Poland in 1999 and the Ph.D. degree from the University of Colorado at Boulder, U.S.A. in 2003. Dr. Kurgan joined the Department of Electrical and Computer Engineering at the University of Alberta in Edmonton in 2003, where he was promoted to the Associate Professor rank in 2007. He is a founder and director of the biomine lab (http://biomine.ece.ualberta.ca/). His research group specializes in the development and application of modern data mining methods in bioinformatics, with focus on analysis of sequence, structure, and function of biologically interesting macromolecules. Dr. Kurgan is involved, as a steering committee member, in organization of the International Conference on Machine Learning and Applications. He serves on the editorial board of several journals including Neurocomputing, Open Proteomics Journal, Journal of Biomedical Science and Engineering, and Open Bioinformatics Journal. For more information about Professor Lukasz Kurgan, visit http://biomine.ece.ualberta.ca/.