Identification of secondary structures in globular proteins--a new algorithm.

A new algorithm has been developed for identifying helices, extended structures, and bends from the positions of the alpha-carbon atoms using the virtual bond approach. The parameters used are two virtual bond angles (delta 1 and delta 2), the virtual dihedral angle (theta), and the distance (D) between the terminal alpha-carbon atoms of the tripeptide. The criteria for classification have been worked out by model building as well as from proteins whose complete secondary structures are known. These criteria are as follows: (i) magnitude of theta less than or equal to 60 degrees and (delta 1 + delta 2) less than or equal to 230 degrees for a bend, (ii) for a helix, successive thetas should not differ by more than 30 degrees, and (iii) for an extended structure, the cumulative deviation of the above parameters should not vary by more than 20% from the ideal extended chain. The method developed has been applied successfully to three proteins wherein the coordinates of alpha-carbon atoms alone are known and a complete mapping of the secondary structures has now been obtained. One interesting observation is that the percentage of residues not taking part in helices, extended structures, and bends is very small--of the order of 4%.