Koretke K K, Luthey-Schulten Z, Wolynes P G
School of Chemical Sciences, University of Illinois, Urbana, IL 61801, USA.
Proc Natl Acad Sci U S A. 1998 Mar 17;95(6):2932-7. doi: 10.1073/pnas.95.6.2932.
The protein energy landscape theory is used to obtain optimal energy functions for protein structure prediction via simulated annealing. The analysis here takes advantage of a more complete statistical characterization of the protein energy landscape and thereby improves on previous approximations. This schema partially takes into account correlations in the energy landscape. It also incorporates the relationships between folding dynamics and characteristic energy scales that control the collapse of the proteins and modulate rigidity of short-range interactions. Simulated annealing for the optimal energy functions, which are associative memory hamiltonians using a database of folding patterns, generally leads to quantitatively correct structures. In some cases the algorithm achieves "creativity," i.e., structures result that are better than any homolog in the database.
蛋白质能量景观理论用于通过模拟退火获得用于蛋白质结构预测的最优能量函数。此处的分析利用了对蛋白质能量景观更完整的统计特征描述,从而改进了先前的近似方法。该模式部分考虑了能量景观中的相关性。它还纳入了折叠动力学与控制蛋白质折叠和调节短程相互作用刚性的特征能量尺度之间的关系。使用折叠模式数据库的关联记忆哈密顿量对最优能量函数进行模拟退火,通常会得到定量正确的结构。在某些情况下,该算法实现了“创造性”,即得到的结构比数据库中的任何同源物都更好。