ALGORITHMS: Collaborative Research:Development of Vector Space based Methods for Protein Structure Prediction

University Of Texas At Dallas, Richardson TX

Investigators

Abstract

What is novel and unique in this proposed research is the proposed design of knowledge-based prediction systems based on optimal vector space representations of proteins that have previously been represented by character strings. If an optimal representation of a character string can be found by a numerical sequence, then a great variety of methodologies from disciplines such as optimization, pattern discovery, and machine learning can be readily applied to new understanding of protein tertiary structure and function. For this, kernel based nonlinear classifiers and nonlinear dimension reduction as well as visualization methods will be developed to provide scalable and elective prediction systems.The prediction systems will be specially tailored for several problems related to protein structure discovery such as protein secondary structure, relative solvent accessibility and disulfide connectivity, as well as prediction of protein-protein interaction. In this proposal the P.I.s describe how they intend to accomplish this, so that their preliminary results can be extended to the more general structure and protein-protein interaction prediction problem. All results obtained will be made available to the research community in order to facilitate further research activity. Using existing web servers, the results will be made available to teaching faculty and graduate and undergraduate students in a suitable tutorial form. This will allow those interested to tailor the material for use in graduate and undergraduate research and class projects. The authors will incorporate the results into current and future course material as well. Special efforts will be made by the two women PIs to provide opportunities for women graduate students to participate in the proposed research and for development of the related software, which has long range social impact beyond the scientific results.