Central Amino Acids Of The Polypeptide Fragment-based Assembly Of Protein Structure Prediction

Function of proteins is determined by their structures, and structures are determined bytheir sequences. The three-dimensional structures of proteins are critical useful for proteinfunction analysis. Currently, there are two major methods to study the structures ofproteins directly, X-ray crystallography or nuclear magnetic resonance (NMR)spectroscopy. Because of the limitation of these two methods, it is hard to get allthree-dimensional structures of proteins. In the past several decades, a great progress hasbeen made in protein structure prediction by computer. These prediction methods could bedivided into three categories, homology modeling, fold recognition, and ab initio. Thereare some problems in these methods, such as the requirement of vast computationalresources, manual selection in computational progress, residue replacement, and side chainallocation. These problems make it hard for users to predict protein three-dimensionalstructures.In this research, the structures of central amino acids of peptide fragments wereanalyzed. We attempted to use C' deviation to study the geometry of amino acids inpeptides. Diameters of distribution were used to describe the various atomic structures, andscatter graphs provided visual evaluation. In this experiment, by comparing the atomicdistribution of the central amino acids of five-residue peptide fragments and three-residuepeptide fragments, we found that longer length of peptide fragments could restrict thedistribution of atoms to a greater degree. And the secondary structures of amino acids inthe central position of the peptide fragments were also analyzed. The results showed thatatoms of central amino acids with different secondary structures distributed in distinctareas.According to the result which longer length of peptide fragments could restrict thedistribution of atoms to a greater degree, we established a protein structure predictionmethod named assembly of protein three-dimensional structure with central amino acids ofpeptide fragments. This method cuts target sequences into nine-residue peptide fragments, assembles every central amino acids of nine-residue fragments according to sequences,fulfill the prediction of protein structures. Because the central amino acids of best matchingnine-residue peptide fragments searched from library and target peptide fragments are thesame, residue replacement is avoided in prediction progress, and coordinates of side chainsof best matching nine-residue peptide fragments could be used after transformation. Sidechain allocation is also avoided.At last we evaluated the precision and efficiency of this new protein structureprediction method. Results showed the precision of our method was satisfied. Therequirement of computational resource was low, and the personal computer couldaccomplish the method. The program could run either in Windows or Linux environmentwith Perl compiler. The whole process of prediction was fully automatic, manualinterference was not needed.The major innovation in this research:1. C' deviation was proposed first time and used to analyze the structure of central aminoacids of peptide fragments.2. A new parameter, distribution diameter, was used to describe the distribution of theatoms, and it could compare the distribution of atoms directly.3. We analyzed the distribution of central amino acids of peptide fragments with differentlocal sequence similarities, and found longer length of peptide fragments could restrictthe distribution of atoms to a greater degree.4. We analyzed the secondary structures of amino acids in the central position of thepeptide fragments, found central amino acids with different secondary structuresdistributed in distinct areas.5. We established a protein structure prediction method named assembly of proteinthree-dimensional structure with central amino acids of peptide fragments, resolved theresidue replacement and side chain allocation.6. Prediction of protein structure on PC was achieved.