THEORETICAL PROTEIN STRUCTURE PREDICTION BY HOMOLGY

描述：Abagyan博士先前已经证明， 能量函数，结合强大的构象搜索程序， 可以准确预测小分子的详细（全原子）结构 和多肽片段。 建议将这些进展应用于 开发一个全面和准确的同源性预测程序。 将该方法应用于同源建模测试实例表明， 允许在理想化条件下成功预测环和侧链 条件 然而，变形而僵硬的结构环境， 模型环或其他骨架片段被发现是最大的 成功的同源性建模的障碍。 建议：调整 回路建模方法，以允许在回路的周围重新排列; 自动确定这种“软”环境;应用 多肽末端、移动的二级结构元件的预测方法 和肽平面方向误差;以测试和完善的所有组成部分 该方法，使用已知同源对的最佳测试集， 标准数据库。

DESCRIPTION: Dr. Abagyan has previously demonstrated that an improved energy function, combined with a powerful conformational search procedure, can accurately predict the detailed (all-atom) structure of small molecules and polypeptide fragments. It is proposed to apply these advances to the development of a comprehensive and accurate homology prediction procedure. Application of this method to homology modeling test cases showed that it permits successful prediction of loops and side chains under idealized conditions. However, the deformed yet rigid structural environment of modeled loops or other backbone fragments was found to be the largest obstacle to successful homology modeling. It is proposed: to adapt the loop modeling method to permit rearrangements in the loop's surroundings; to automate the determination of this "soft" environment; to apply the prediction method to polypeptide ends, mobile secondary structure elements and peptide plane orientation errors; to test and refine all components of the method, using an optimal test set of known homologous pairs from the standard database.