VERIFICATION OF PROTEIN STRUCTURES - PATTERNS OF NONBONDED ATOMIC INTERACTIONS

A novel method for differentiating between correctly and incorrectly d etermined regions of protein structures based on characteristic atomic interactions is described. Different types of atoms are distributed n onrandomly with respect to each other in proteins. Errors in model bui lding lead to more randomized distributions of the different atom type s, which can be distinguished from correct distributions by statistica l methods. Atoms are classified in one of three categories: carbon (C) , nitrogen (N), and oxygen (0). This leads to six different combinatio ns of pairwise noncovalently bonded interactions (CC, CN, CO, NN, NO, and 00). A quadratic error function is used to characterize the set of pairwise interactions from nine-residue sliding windows in a database of 96 reliable protein structures. Regions of candidate protein struc tures that are mistraced or misregistered can then be identified by an alysis of the pattern of nonbonded interactions from each window.