During the period of this Small Grant for Exploratory Research award, the principal investigator will complete an initial investigation of the applicability of a unique mathematical technique based on the methods of stochastic state-space modeling and Bayesian filtering to the prediction of protein structural class. This research will complete a pilot study, which so far has demonstrated that our approach is capable of correctly discriminating more than 80% of the time between predominantly all alpha and predominantly all beta proteins. In addition, it will allow for a proper comparison with other methods, and a full statistical evaluation of the predicted classification sensitivity and specificity. One of the fundamental problems in molecular biology is that of relating a protein primary sequence to its function. The function, whether structural or catalytic, is encoded in the protein's structure, often in the details of the three-dimensional arrangement of amino-acid side chains on the surface. The principal investigator is suggesting a method of prediction of protein structural classes from protein sequence data. His method employs stochastic modeling and Bayesian filtering.
