Automated determination of protein structures from NMR data by iterative analysis of self-consistent contact patterns.

摘要

This dissertation describes a prototype ruled-based expert system, AutoStructure, which automatically determines protein structures from NOESY peak lists and other NMR data using various analysis rules generalized from basic procedures developed by human experts. AutoStructure uses a two-step approach. In the first step, AutoStructure generates a reliable initial protein fold using intelligent analysis methods based on spectrum specific properties and the identification of self-consistent NOE contact patterns, without using any 3D structure model. In particular, the software identifies secondary structures, including alignments between β-strands, based upon the combined pattern analysis of secondary structure specific NOE contacts, chemical shift, scalar coupling constant, and slow amide proton exchange data. In the second step, the software derives and generates conformational constraints (e.g. distance, dihedral angle and hydrogen bond constraints) automatically and submits parallel DYANA structure calculations to an array of Pentium III processors. The resulting structures are then automatically refined by iterative cycles of assigning self-consistent NOESY cross peaks and regeneration of the protein structure with DYANA calculations.; AutoStructure has been tested on three real protein NMR data sets. These three proteins are from three different folding families, and range in sizes between 38 and 155 amino acid residues. One is a mixed α/β protein, the second is mainly β-sheet, and the third one is mainly α-helix. Two are monomeric proteins, and the third is a homodimer. With these data sets, AutoStructure could provide high-quality automated analysis of NOESY cross peak assignments and determine high-quality 3D protein structures in hours. These results support the view that rapid and fully automatic analysis of 3D protein structures from NMR data can play a potentially significant role in the emerging area of structural genomics.