A topology-constrained distance network algorithm for protein structure determination from NOESY data
A topology-constrained distance network algorithm for protein structure determination from NOESY data
复制标题
DOI：
10.1002/prot.20820
复制
发表时间：
2006-02-15
影响因子：
2.9
通讯作者：
Montelione, GT
中科院分区：
文献类型：
--
作者：
Huang, YJ;Tejero, R;Montelione, GT
本文利用图论阐述了多维核Overhauser效应光谱（NOESY）解释问题，并提出了一种新颖的、自下而上的、拓扑约束的距离网络分析算法，用于使用指定共振进行NOESY交叉峰解释。AutoStructure是一个软件套件，实现了这种拓扑约束距离网络分析算法，并使用三维（3D）蛋白质结构计算程序XPLOR/CNS或DYANA迭代生成结构。AutoStructure的最小输入包括氨基酸序列，共振分配列表，以及2D， 3D和/或4D-NOESY交叉峰列表。AutoStructure也可以分析同型二聚体蛋白质，当x -过滤NOESY实验可用。输入数据和最终3D结构的质量使用召回率，精度和F-measure （RPF）分数来评估，F-measure是与输入数据拟合优度的统计度量。AutoStructure已经在三个蛋白质NMR数据集上进行了测试，其中高质量的结构以前已经由专家解决，并在数小时内产生可比的高质量距离约束列表和3D蛋白质结构。我们还将AutoStructure测定的几种蛋白结构与其他独立方法测定的相应同源蛋白进行了比较。该程序已用于二十多个蛋白质结构的测定，其中一些已经发表。
This article formulates the multidimensional nuclear Overhauser effect spectroscopy (NOESY) interpretation problem using graph theory and presents a novel, bottom-up, topology-constrained distance network analysis algorithm for NOESY cross peak interpretation using assigned resonances. AutoStructure is a software suite that implements this topology-constrained distance network analysis algorithm and iteratively generates structures using the three-dimensional (3D) protein structure calculation programs XPLOR/CNS or DYANA. The minimum input for AutoStructure includes the amino acid sequence, a list of resonance assignments, and lists of 2D, 3D, and/or 4D-NOESY cross peaks. AutoStructure can also analyze homodimeric proteins when X-filtered NOESY experiments are available. The quality of input data and final 3D structures is evaluated using recall, precision, and F-measure (RPF) scores, a statistical measure of goodness of fit with the input data. AutoStructure has been tested on three protein NMR data sets for which high-quality structures have previously been solved by an expert, and yields comparable high-quality distance constraint lists and 3D protein structures in hours. We also compare several protein structures determined using AutoStructure with corresponding homologous proteins determined with other independent methods. The program has been used in more than two dozen protein structure determinations, several of which have already been published.