2014.6.13 Development of knowledge-based energy functions by series expansions and parameter optimization

2019-07-07 00:07:05 5




题目:Development of knowledge-based energy functions by series expansions and  parameter optimization



报告人梁世德  博士








We expanded atomic interaction energy functions in a protein as series. Thousands of parameters were optimized by maximizing the gap between the native and non-native conformations in single side chain prediction for all of available protein structures.  When the resulting energy functions (the OSCAR force field) were used to model side chain conformations of a whole protein, the prediction accuracies were 88.8% for χ1, 79.7% for χ1 + 2, 1.24 Å overall root mean square deviation (RMSD), and 0.62 Å RMSD for core residues, respectively, compared with the next-best performing side-chain modeling program which achieved 86.6% for χ1, 75.7% for χ1 + 2, 1.40 Å overall RMSD, and 0.86 Å RMSD for core residues.  The high accuracy of our program was confirmed by independent third party research. In addition, the OSCAR force field showed significantly better performances than classical physics-based force fields or other knowledge–based energy functions in protein loop modeling, protein design, and protein-protein interaction.