作者: Chris Colovos , Todd O. Yeates
关键词: Pairwise comparison 、 Function (mathematics) 、 Atom 、 Protein structure 、 Crystallography 、 Hydrogen bond 、 MODELLER 、 Biological system 、 Chemistry 、 Crystal structure 、 Set (abstract data type)
摘要: A novel method for differentiating between correctly and incorrectly determined regions of protein structures based on characteristic atomic interaction is described. Different types atoms are distributed nonrandomly with respect to each other in proteins. Errors model building lead more randomized distributions the different atom types, which can be distinguished from correct by statistical methods. Atoms classified one three categories: carbon (C), nitrogen (N), oxygen (O). This leads six combinations pairwise noncovalently bonded interactions (CC, CN, CO, NN, NO, OO). quadratic error function used characterize set nine-residue sliding windows a database 96 reliable structures. Regions candidate that mistraced or misregistered then identified analysis pattern nonbonded window.