Secondary structure assignment for alpha/beta proteins by a combinatorial approach

We describe an algorithm for assigning the secondary structure of alpha/beta proteins. Turns are identified very accurately (98%) by simultaneously considering hydrophilicity and the ideal spacing of turns throughout the sequence. The segments bounded by these turns are labeled by a pattern-recognition scheme based on the physical properties of alpha-helices and beta-strands, in this class of proteins. Long-range, as well as local, information is incorporated to enhance the quality of the assignments. Although the assignment for any one sequence is not unique, at least one of the assignments bears a close resemblance to the native structure. The algorithm successfully divides protein sequences into two classes: alpha/beta and non-alpha/beta. The accuracy of the secondary-structure assignments in the alpha/beta class is sufficient to provide useful input for tertiary-structure assignments.