Kissinger C R, Gehlhaar D K, Smith B A, Bouzida D
Pfizer Inc., 10350 North Torrey Pines Road, San Diego, CA 92037, USA.
Acta Crystallogr D Biol Crystallogr. 2001 Oct;57(Pt 10):1474-9. doi: 10.1107/s0907444901012458. Epub 2001 Sep 21.
Stochastic search algorithms can be used to perform rapid six-dimensional molecular-replacement searches. A molecular-replacement procedure has been developed that uses an evolutionary algorithm to simultaneously optimize the orientation and position of a search model in a unit cell. Here, the performance of this algorithm and its dependence on search model quality and choice of target function are examined. Although the evolutionary search procedure is capable of finding solutions with search models that represent only a small fraction of the total scattering matter of the target molecule, the efficiency of the search procedure is highly dependent on the quality of the search model. Polyalanine models frequently provide better search efficiency than all-atom models, even in cases where the side-chain positions are known with high accuracy. Although the success of the search procedure is not highly dependent on the statistic used as the target function, the correlation coefficient between observed and calculated structure-factor amplitudes generally results in better search efficiency than does the R factor. An alternative stochastic search procedure, simulated annealing, provides similar overall performance to evolutionary search. Methods of extending the evolutionary search algorithm to include internal optimization, selection and construction of the search model are now beginning to be investigated.
随机搜索算法可用于执行快速的六维分子置换搜索。已经开发出一种分子置换程序，该程序使用进化算法同时优化搜索模型在晶胞中的方向和位置。在此，研究了该算法的性能及其对搜索模型质量和目标函数选择的依赖性。尽管进化搜索程序能够使用仅代表目标分子总散射物质一小部分的搜索模型找到解决方案，但搜索程序的效率高度依赖于搜索模型的质量。即使在侧链位置已知精度很高的情况下，聚丙氨酸模型通常也比全原子模型提供更好的搜索效率。尽管搜索程序的成功并不高度依赖于用作目标函数的统计量，但观察到的和计算出的结构因子振幅之间的相关系数通常比R因子产生更好的搜索效率。另一种随机搜索程序，模拟退火，提供了与进化搜索相似的整体性能。现在开始研究将进化搜索算法扩展到包括内部优化、搜索模型的选择和构建的方法。
