GUIDING PROBABILISTIC SEARCH OF THE PROTEIN CONFORMATIONAL SPACE WITH STRUCTURAL PROFILES

摘要

The roughness of the protein energy surface poses a signi¯cant challenge to search algorithmsnthat seek to obtain a structural characterization of the native state. Recent research seeks to biasnsearch toward near-native conformations through one-dimensional structural pro¯les of thenprotein native state. Here we investigate the e®ectiveness of such pro¯les in a structure predictionnsetting for proteins of various sizes and folds. We pursue two directions. We ¯rst investigate thencontribution of structural pro¯les in comparison to or in conjunction with physics-based energynfunctions in providing an e®ective energy bias. We conduct this investigation in the context ofnMetropolis Monte Carlo with fragment-based assembly. Second, we explore the e®ectiveness ofnstructural pro¯les in providing projection coordinates through which to organize the conformationalnspace. We do so in the context of a robotics-inspired search framework proposed in ournlab that employs projections of the conformational space to guide search. Our ¯ndings indicatenthat structural pro¯les are most e®ective in obtaining physically realistic near-native conformationsnwhen employed in conjunction with physics-based energy functions. Our ¯ndings alsonshow that these pro¯les are very e®ective when employed instead as projection coordinates tonguide probabilistic search toward undersampled regions of the conformational space