Constrained Sequential Monte Carlo and Its Applications

描述(申请人提供)：分子生物学中的一个基本问题是蛋白质的结构-功能关系。要了解结构如何决定蛋白质的功能，关键是：(1)识别蛋白质上重要的功能表面。在基因组和蛋白质组水平上，同样重要的是：(2)确定可能具有不同折叠结构的蛋白质之间的蛋白质表面模式的显著相似性。结构-功能关系的逆问题问：(3)蛋白质功能如何影响蛋白质的折叠和稳定性？一个相关的一般性问题是(4)：几何性质，如包装缺陷，是否会影响蛋白质的稳定性和功能，例如，对于在高温下茁壮成长的嗜热微生物的蛋白质？ 这个项目开发了新的统计模型和计算方法，帮助解决这四个重要的生物学问题。最近在统计学中出现的序贯蒙特卡洛(SMC)方法显示了巨大的前景。该项目开发了约束序贯蒙特卡罗(CSMC)方法，专门为解决这些具有严格约束的高维和复杂的统计推断问题而设计。为成功地实施CSMC，提出了关键部件设计的一般策略和理论。实施的CSMC工具免费分发给研究社区。 该项目的结果使得发现空间表面基序和揭示对药物发现重要的蛋白质的新的功能关系成为可能。当结构信息不可用时，所发现的新模式可以用于搜索功能相关的蛋白质序列。此外，这项研究还为定量评估蛋白质功能如何影响蛋白质折叠和稳定性提供了重要工具。深入了解包装缺陷是如何影响蛋白质稳定性的。

DESCRIPTION (provided by applicant): A fundamental problem in molecular biology is the structure-function relationship of proteins. To understand how structure dictates the function of a protein, it is essential to: (1) Identify functionally important surfaces on protein. At genomic and proteomic scale, it is also critical to: (2) Identify significant similarity of protein surface patterns among proteins which may have different fold structures. The inverse problem of the structure-function relationship asks: (3) How does protein function influence the folding and stability of proteins? A related general question is (4): Do geometric properties such as packing defects influence the stability and functions of proteins, e.g., for proteins from thermophilic microbes that thrive at high temperature? This project develops novel statistical models and computational methods that helps to solve these four important biological problems. The sequential Monte Calo (SMC) methodologies recently emerged in statistics show great promises. This project develops Constrained Sequential Monte Carlo (CSMC) methods specifically designed to solve these high dimensional and complex statistical inference problems with severe constraints. General strategies and theory in designing the key components are developed for successful CSMC implementation. Implemented CSMC tools are disseminated to research community freely. The results of this project enable the discovery of spatial surface motifs and uncover novel functional relations of proteins important for drug discovery. New patterns discovered can be employed to search for functionally related protein sequences, when structural information is not available. In addition, this research provides important tools for quantitatively assessing how protein function influence protein folding and stability. Insights are gained towards understanding how packing defects influence proteins stability.