Lee Audrey, Streinu Ileana, Brock Oliver
Department of Computer Science, University of Massachusetts Amherst, Amherst, MA, USA.
Phys Biol. 2005 Nov 9;2(4):S108-15. doi: 10.1088/1478-3975/2/4/S05.
Motivated by recently developed computational techniques for studying protein flexibility, and their potential applications in docking, we propose an efficient method for sampling the conformational space of complex molecular structures. We focus on the loop closure problem, identified in the work of Thorpe and Lei (2004 Phil. Mag. 84 1323-31) as a primary bottleneck in the fast simulation of molecular motions. By modeling a molecular structure as a branching robot, we use an intuitive method in which the robot holds onto itself for maintaining loop constraints. New conformations are generated by applying random external forces, while internal, attractive forces pull the loops closed. Our implementation, tested on several model molecules with low number of degrees of freedom but many interconnected loops, gives promising results that show an almost four times speed-up on the benchmark cube-molecule of Thorpe and Lei.
受近期开发的用于研究蛋白质灵活性的计算技术及其在对接方面潜在应用的启发，我们提出了一种有效方法来对复杂分子结构的构象空间进行采样。我们关注环闭合问题，该问题在Thorpe和Lei（2004年，《哲学杂志》84卷，1323 - 31页）的工作中被确定为分子运动快速模拟中的主要瓶颈。通过将分子结构建模为分支机器人，我们使用一种直观的方法，即机器人抓住自身以维持环约束。通过施加随机外力生成新的构象，而内部吸引力则将环拉闭。我们的实现方法在几个自由度数量较少但有许多相互连接环的模型分子上进行了测试，得到了很有前景的结果，在Thorpe和Lei的基准立方体分子上显示出几乎四倍的加速。
