Protein Backbone Reconstruction with Tool Preference Classification for Standard and Nonstandard Proteins

Given a protein sequence and the Cα coordinates on its backbone, the all-atom protein backbone reconstruction problem (PBRP) is to reconstruct the backbone by its 3D coordinates of N, C and O atoms. In the past few decades, many methods have been proposed for solving PBRP. Related research reveals that if proper prediction tools are selected to build the 3D coordinates of the desired atoms, the RMSD may be improved. In this paper, we propose a method for solving PBRP, performing tool preference classification on each atom of the residue, where the classification model is generated by support vector machine (SVM). We rebuild the backbone by combining the prediction results of all atoms in all residues. The data sets used in our experiments are CASP7, CASP8, and CASP9, which contain 65, 52, and 63 proteins, respectively. These data sets contain nonstandard amino acids along with standard ones. The RMSDs we achieve are 0.3496 in CASP7, 0.3084 in CASP8, and 0.3286 in CASP9. This research was partially supported by the National Science Council of Taiwan under contract NSC 100-2221E-242-003. This research was also partially supported by the “Project Digital Convergence Service Open Platform” of the Institute for Information Industry which is subsidized by the Ministry of Economy Affairs of the Republic of China. Corresponding author. E-mail: [email protected] (Chang-Biau Yang).