A comprehensive analysis of structure-function relationships in membrane protein families

尽管膜结合蛋白具有至关重要的作用，但与球状蛋白相比，在全基因组中编码的膜结合蛋白的注释相对较差。本提案的主要目标是建立一个通用的生物信息学平台，用于大规模研究膜蛋白家族。基于我们之前在系统基因组分析方面的工作，我们打算开发一个全面和分层的膜蛋白家族分类，并提供适当的生物信息学工具-多序列比对，隐马尔可夫模型-用于研究其结构和功能方面。特别是，将设计专门考虑到预测结构特征（如疏水跨膜段）的对齐算法。此外，我们计划开发先进的方法来预测膜蛋白的精确功能特异性。相关突变分析将用于探索相互关联的残基网络，确定赋予功能特异性的残基，并确定每个序列家族中的功能亚型。与此同时，特异性决定残基将使用一种基于遗传算法的新型自动方法从多个对齐序列中描绘出来，该方法不需要对给定家族中的功能亚群的先验知识。准确的功能预测是构建能够阻断细胞特定通道和运输系统的药物的重要先决条件。这项工作中产生的方法和数据的另一个潜在应用将是膜蛋白结构基因组学的靶标选择。此外，本项目的结果将有助于构建许多膜蛋白家族的分子模型，研究螺旋填充和寡聚化界面。

In spite of their crucial importance membrane-bound proteins encoded in complete genomes are relatively poorly annotated compared to globular proteins. The main goal of the present proposal is to build a general bioinformatics workbench for investigating membrane protein families at large scale. Based on our previous work in systematic genome analysis we intend to develop a comprehensive and hierarchical membrane protein family classification and provide appropriate bioinformatics instruments ¿ multiple sequence alignments, Hidden Markov Models ¿ for investigating aspects of their structure and function. In particular, alignment algorithms specifically taking into account predicted structural features ¿ such as hydrophobic transmembrane segments ¿ will be designed. Furthermore we plan to develop advanced approaches for predicting the precise functional specificity of membrane proteins. Correlated mutation analysis will be applied to explore the networks of inter-correlated residues, determine the residues that confer functional specificity, and to identify functional subtypes within each sequence family. In parallel, specificity-determining residues will be delineated from multiply aligned sequences using a novel automatic approach based on the genetic algorithm that does not require a priori knowledge on functional sub-groups in a given family. Accurate function prediction is an important pre-requisite for constructing drugs capable of blocking specific channels and transport systems of the cell. Another potential application for the methods and data generated in this work will be target selection for structural genomics of membrane proteins. In addition, the results of this project will be useful for constructing molecular models for numerous families of membrane proteins, investigating helix packing, and oligomerization interfaces.