Structure-Based Prediction of the Interactome

描述（由申请人提供）：蛋白质-蛋白质相互作用（PPIs）在所有生物过程中发挥核心作用。与基因组的完整测序类似，相互作用组的完整描述是深入了解生物过程的基本步骤，并具有影响系统生物学，基因组学，分子生物学和治疗学的巨大潜力。尽管发现ppi的高通量生化方法已被证明非常成功，但目前相互作用组的实验覆盖率仍然不足，并且将受益于计算工具。本提案的广泛、长期目标是利用基于结构的计算方法提供的信息作为潜在的高质量、高覆盖率的数据源，用于大规模集成方法来构建相互作用组。具体而言，本项目旨在：1)开发新的基于结构的预测方法，可应用于基因组尺度；2)将这些预测与其他功能基因组信息相结合，在基因组尺度上预测ppi。该项目还将为实验研究产生可测试的假设。该研究的一个关键产品是LTHREADER程序，这是一个本地化的线程程序，可以同时将查询序列对与蛋白质-蛋白质界面模板对齐。通过利用蛋白质复合物界面中包含的信息，它可以显著提高覆盖范围和预测质量。一些核心的计算方面是开发用于将查询序列对线程化到模板的算法（使用线性规划），学习统计势（svm），以及组合多个蛋白质接口分数用于PPI预测（增强）。这种基于结构的方法的输出将与Struct2Net框架中的其他功能基因组数据相结合，用于预测ppi（使用随机森林）。拟议研究的最终成果将是一个全基因组PPI预测的综合数据库，该数据库来自纯粹基于结构的方法以及综合方法。该数据库还将包括细胞外配体与受体的相互作用。

DESCRIPTION (provided by applicant): Protein-protein interactions (PPIs) play a central role in all biological processes. Akin to the complete sequencing of genomes, complete descriptions of interactomes is a fundamental step towards a deeper understanding of biological processes, and has a vast potential to impact systems biology, genomics, molecular biology and therapeutics. Although high-throughput biochemical approaches for discovering PPIs have proven very successful, the current experimental coverage of the interactome remains inadequate and would benefit from computational tools. The broad, long term goal of this proposal is to harness the information provided by structure-based computational approaches as a potentially high-quality, high-coverage data source for large-scale integrative approaches to interactome construction. Specifically, this project aims to: 1) develop new structure-based prediction methods that can be applied on a genome scale, and 2) integrate these predictions with other functional genomic information to predict PPIs at a genome scale. This project will also generate testable hypotheses for experimental investigations. A key product of the proposed research is the LTHREADER program, a localized threading program that will simultaneously align query sequence-pairs to templates of protein-protein interfaces. By exploiting information contained in the protein complex interfaces, it may significantly improve upon the state-of-the-art in coverage and prediction quality. Some of the core computational aspects are the development of algorithms for threading query sequence-pairs to templates (using linear programming), learning statistical potentials (SVMs), and combining multiple protein interface scores for PPI prediction (boosting). The output from such structure-based approaches will be combined with other functional genomic data in the Struct2Net framework for predicting PPIs (using random forests). A final product of the proposed research will be a comprehensive database of genome-wide PPI predictions derived from purely structure-based as well as integrative approaches. The database will also include extracellular ligand-receptor interactions. The prediction of PPIs will enable better elucidation of extracellular and intracellular signaling networks, which has direct medical implications in terms of drug target identification. For example, a promising public-health application of this research is the rational design of therapeutics which inhibit or interfere with the binding of extracellular ligands to receptors. All the produced computational algorithms, software, and databases will be made publicly available for further studies.Relevance Proteins interact with each other to communicate within and between cells, forming networks (the Interactome) that play fundamental roles in all biomedical processes including the maintenance of cellular integrity, metabolism, transcription/translation, and cell-cell communication. Understanding these interaction networks on a large scale will empower both rational, targeted drug design and more intelligent disease management. In this project, we develop computational methods for structure-based prediction of protein-protein interactions, and integrate these predictions with available high- throughput genomic data to predict the Interactomes of entire species' genomes.