Multi-variate Minimization Techniques and Molecular Dynamics Simulations for Predicting 3-D Structures of DNA and RNA on Supercomputers

这里提出的研究目标是开发用于计算大型生物分子三维结构的数学工具。特别是，建模和大规模最小化技术将被设计来预测单链和双链RNA和DNA分子的构象。计算工具包括半经验势能最小化和分子动力学，在Cray X-MP上执行，并辅以Evans和Sutherland PS-330上的交互式计算机图形。该计划将在以下四个方面进行程序开发：(1)全局优化策略；验证Levy和Gomez的确定隧道法和随机模拟退火的一种变体：(2)更快地处理能量函数中的成对相互作用；研究Greengard和Rokhlin最近提出的新方法，该方法可以将N -粒子系统的O（N2）阶计算复杂度降低到O(N)阶：(3)分子动力学的实现：动力学模拟将能够检查从最小化中获得的结构的动态轨迹；(4)溶剂和金属离子加入分子模型；这将提供一个更真实的势场来表示在自然环境中影响RNA和DNA的分子力。开发的计算工具将产生研究三维分子结构的一些有趣问题的能力。这些问题包括DNA构象对核酸碱基序列的依赖以及单链RNA的结构和复杂折叠模式。最小化和动力学结果将用于制作16毫米的电影和录像带，这些电影和录像带将揭示结构的动态范围，并允许我们将构象特征与已知的生物功能联系起来。本提案计划开发数学工具，使用最先进的优化算法，用于检查生物重要分子RNA和DNA中溶剂和金属离子相互作用的动力学模拟。计算科学与工程项目主任建议资助69,370美元，为期一年。

The goal for the research proposed here is to develop mathematical tools for computing three-dimensional structures of large biological molecules. In particular, modeling and large- scale minimization techniques will be devised to predict the conformations of single and doublestranded RNA and DNA molecules. The computational tools involve semi-empirical potential energy minimization and molecular dynamics, executed on the Cray X-MP and aided by interactive computer graphics on the Evans and Sutherland PS-330. The plan is to work on the following four areas of program development: (1) Global optimization strategies; test the determination Tunneling method of Levy and Gomez and a variant of the stochastic Simulated Annealing: (2) Faster manipulation of the pairwise interactions in the energy function; investigate the new method recently proposed by Greengard and Rokhlin which could reduce the order O(N2) computational complexity for an N - particle system to order O(N): (3) Implementation of molecular dynamics: dynamics simulations will enable an examination of the dynamic trajectories of the structures obtained from minimization; and: (4) Incorporation of solvent and metalions into the molecular model; this will provide a more realistic potential field to represent the molecular forces influencing the RNA and DNA in its natural milieu. The developed computational tools will yield the ability to investigate some intriquing problems of three-dimensional molecular structures. These problems include the dependence of DNA conformation on the nucleic acid base sequence and the structure and the complex folding patterns of single-stranded RNA. The minimizations and dynamics results will be used to make 16 mm films and videotapes that will reveal the dynamic range of structures and permit us to relate conformational features with known biological functions. This proposal projects to develop mathematical tools, using state-of-the-art optimization algorithms, for examination of dynamics simulation of solvent and metal-ion interaction in the biologically important molecular RNA and DNA. The Program Director for Computational Science and Engineering recommends support in the sum of $69,370 for a period of one year.