DANGLE: A Bayesian inferential method for predicting protein backbone dihedral angles and secondary structure

Abstract

This paper introduces DANGLE, a new algorithm that employs Bayesian inference to estimate the likelihood of all possible values of the backbone dihedral angles φ{symbol} and ψ for each residue in a query protein, based on observed chemical shifts and the conformational preferences of each amino acid type. The method provides robust estimates of φ{symbol} and ψ within realistic boundary ranges, an indication of the degeneracy in the relationship between shift measurements and conformation at each site, and faithful secondary structure state assignments. When a simple degeneracy-based filtering procedure is applied, DANGLE offers an ideal compromise between accuracy and coverage when compared with other shift-based dihedral angle prediction methods. In addition, per residue analysis of shift/structure degeneracy has potential to be a useful new approach for studying the properties of unfolded proteins, with sufficient sensitivity to identify regions of residual structure in the acid denatured state of apomyoglobin. © 2009 Elsevier Inc. All rights reserved.