The value of phylogenetic information in post-genomic protein structure prediction and function discovery

Structure-based drug design of inhibitors against targets with characterized function is the basis for pharmaceutical development worldwide. Meanwhile, genomes of multiple species are providing vast sequence data that is sparsely annotated with structural and functional information. Effective computational methods have become imperative in biological discovery in the post-genomic sequence era.; Structural information provides specific advantages in protein system characterizations and drug discovery. I have developed a method based on primary sequence and phylogenetic information, to identify evolutionary sequence correlations of residue, proximal in space. Novel constraints related to conditional probabilities of residue and secondary element contacts are calculated from available structures. I describe an ab initio protein contact map prediction method, which relies on iterative application of structural constraints to the evolutionary correlations. The predicted contact maps have features of protein structures not seen in other prediction methods and can be used in combination with other methods to generate protein models.; The post-genomic era provides opportunities to reinterpret data in context of complete catalogs of genes. On the example of a malarial cysteine protease, I have performed protein family analysis, homology modeling and computational screening. A number of antimalarial compounds with low μM activity are identified in cell culture assays. The compounds were selected for predicted binding in unique specificity sites of the malarial protease relative to human protease homologs. Most prescribed antimalarials have significant side effects and lack of similarity between known antimalarials and the compounds identified in this work represents novel routes for antimalarial drug discovery.; The elucidation of protein function in light of whole genome data requires effective computational methods. I have designed and implemented a computer application, JEvTrace, to manipulate and analyze protein family sequences, structures and phylogeny. Based on the ideas of the Evolutionary Trace (Lichtarge, Bourne et al. 1996), a number of algorithmic variations provide new insights into protein function in a family context. Differences and similarities between phylogenetic subclades are analyzed using the primary sequence, tertiary structure, and phylogenetic information. Based on examples of protein structures with unknown or predicted function, JEvTrace provides evidence for new functions and specificities.; Ab initio prediction of structure and function using primary sequence data directly addresses the problems of the post-genomic sequence era. Biological function and structure and their evolution are incredibly complex and not easily amenable to mathematical and computational representations. Combinations of different methods and data represent an important strategy to account for biological variations and irregularities.