ITR: Development of Novel Computational Methods for Genome-Wide Discovery of Gene Function and Networks

marcotte, Edward m .德克萨斯大学：基因功能和网络全基因组发现的新计算方法的发展在人类基因组编码的大约4万个基因中，与迄今为止测序的所有其他基因组一样，大约一半是完全未被表征和功能未知的。人们迫切需要找到方法来发现这数千个未被表征的基因的功能，以及这些基因产物如何参与细胞中的网络、途径和系统。该项目的目标是开发新的计算方法，用于在全基因组范围内发现基因的功能，并发现基因是如何组织成系统和途径的。提出了几种能够提供此类信息的新方法。这些方法被称为“非同源”方法，其本质是除了分析基因序列外，还分析基因的环境特性——比如基因出现在哪些生物体中，哪些基因可以被发现融合在一起，以及其他特性。这样的上下文属性为发现基因功能和基因网络的计算方法的能力提供了巨大的增长。这个提议的核心是一种新的基于图形的方法，将在相同细胞途径中运作的基因联系在一起，从本质上产生基因的“功能图”。该方法结合了一种新的计算方法，用于寻找大序列家族中蛋白质的特定物理相互作用伙伴，以及可视化所产生的复杂基因网络的方法。这些方法的应用将提高我们对每个测序基因组中数千个未表征基因功能的理解。

EIA-0219061Marcotte, Edward MUniversity of TexasITR: Development of novel computational methods for genome-wide discovery of gene function and networksOf the roughly 40,000 genes encoded by the human genome, as with every other genome sequenced to date, about half are completely uncharacterized and of unknown function. There is a broad need for methods to discover the functions of these thousands of uncharacterized genes, as well as how the gene products participate in networks, pathways and systems in the cell. This project's goal is to develop novel computational methods for discovering the functions of genes on a genome-wide scale, and to discover how the genes are organized into systems and pathways. Several novel methods are proposed capable of providing such information. The essence of these methods, called "non-homology" methods, is that they analyze contextual properties of genes-such as which organisms the genes appear in, which genes can be found fused together, and other properties-in addition to the sequences of the genes. Such contextual properties turn out to provide a tremendous increase in the ability of computational methods to discover gene function and gene networks. At the core of this proposal is a novel graph-based method for associating genes together that operate in the same cellular pathway, essentially resulting in a "functional map" of the genes. This method is combined with a new computational method for finding specific physical interaction partners for proteins in large sequence families, and with methods to visualize the resultant complex gene networks. Application of these methods should improve our understanding of the functions of the thousands of uncharacterized genes in each sequenced genome.