DYNAMIC STRUCTURE OF BIOMOLECULES IN SOLUTION

许多生物分子在溶液中具有柔性。 这可以是 通过核磁共振（NMR）方法评估。 确实，J 标量耦合常数和核奥弗豪瑟效应 (NOE) 数据为 当构象异构体存在时，会受到不同类型的平均 解决方案。 在这种情况下，可能无法拟合所有观察到的 具有单一分子构象的 NMR 参数。 我们有 开发了一个程序，PDQPRO（二次概率分布） 编程），确定最优概率分布 通过寻找最适合的潜在构象异构体的预定义库 计算和实验 NMR 参数之间的关系。 这种方法需要一种独立的采样方法来产生 一组潜在的构象异构体。 多项灵活测试分析 具有模拟 NMR 数据的分子已证明 PDQPRO 能够 识别正确的构象异构体和 计算正确的概率，前提是这些概率是正确的 构象异构体存在于潜在结构池中。 目前 我们正在研究 PDQPRO 结合的性能 各种采样方法。 特别是，我们使用了 MDtar Torda 等人开发的方法。生成 MD 轨迹 灵活的RNA环。 MDtar 轨迹作为一个整体解释了 现有核磁共振数据；然而，MD轨迹不太适合 由于其尺寸而进行全面的结构分析。 应用 PDQPRO 对此轨迹的影响减少了一个数量级 解释实验数据所需的构象异构体数量。 的 计算机图形实验室资源对于我们的项目是必要的 用于结构系综的图形表示，这是 这是一个不平凡的问题。 特别是，我们正在使用一些 Eric Pettersen 为此特殊目的编写的 MidasPlus 代表 和大卫·科纳丁。

Many biological molecules are flexible in solution. This can be assessed by nuclear magnetic resonance (NMR) methods. Indeed, J scalar coupling constants and nuclear Overhauser effect (NOE) data are subject to different types of averaging when conformers are present in solution. In suc a case, it may not be possible to fit all observed NMR parameters with a single molecular conformation. We have developed a program, PDQPRO (Probability Distribution by Quadratic PROgramming), which determines the optimal probability distribution for a predefined pool of potential conformers by finding the best fit between calculated and experimental NMR parameters. Such an approach requires an independent sampling method producing a set of potential conformers. Analysis of a number of flexible test molecules with simulated NMR data has demonstrated that PDQPRO is able to recognize the correct conformers and calculate the correct probabilities, provided that those correct conformers are present in the pool of potential structures. Presently we are investigating the performance of PDQPRO in combination with various sampling methods. In particular, we have used the MDtar method developed by Torda et al. to generate an MD trajectory for a flexible RNA loop. The MDtar trajectory as a whole explained the existing NMR data; however, MD trajectories are not very suitable for a comprehensive structural analysis due to its size. The application of the PDQPRO to this trajectory reduced by an order of magnitude the number of conformers required to explain the experimental data. The Computer Graphics Laboratory resources are necessary for our project for graphical representation of structural ensembles, which is a non-trivial problem. In particular, we are using a number of MidasPlus delegates written for this special purpose by Eric Pettersen and David Konerding.