谱聚类在多个网络模块识别中的推广及在生物网络中的应用

项目名称： 谱聚类在多个网络模块识别中的推广及在生物网络中的应用

项目编号： No.11471082

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 数理科学和化学

项目作者： 张淑芹

作者单位： 复旦大学

项目金额： 60万元

中文摘要： 网络研究近年来受到很多领域的广泛关注。模块是多类网络的共同性质，目前已有很多计算方法提出来识别单个网络中的模块。随着各类网络数据的大量产生，多个网络中的模块识别问题最近也被提出。通过整合多个网络中的数据，我们可以找到一些单个网络中信号不明显的模块，还可以找到多个网络中的一些共同模块等。目前已有的多个网络中的模块识别方法具有不稳定、计算复杂度高、无理论分析等弱点。为此，我们提出谱聚类在多个网络模块识别中的推广及在生物网络中的应用这一课题。此课题主要是在谱聚类的基础上，对多个网络中的模块识别提出新的模型、算法及理论分析。之后将所提方法应用到多个基因共表达网络中来识别一些癌症的共同致病因素、癌症发展过程中的模块结构变化等。本课题是多学科交叉的课题，包括计算数学、统计学、生物学等各方面。我们期待此课题的立项，以便更有力地开展工作。它的立项对于应用数学、生物信息学、医学等的发展都有重要意义。

中文关键词： 数值方法；矩阵分析；谱聚类；生物网络；模块识别

英文摘要： Network study has attracted much attention from different research fields. Module is a fundamental property of different types of networks, and many computational methods have been proposed to identify the modules. However, most existing methods mainly concentrated on module identification from an individual network. With the rapid accumulation of network data, module identification from multiple networks has been proposd as a problem. By integrating different networks,common modules for the considered networks can be inferred, and more informative modules can be derived from the networks that have weak signals. The proposed methods for module identification from multiple networks have many weaknesses, such as unstability, high computational complexity, no theoretical analysis, etc.. Therefore, we propose the project Module identification from multiple networks by extending spectral clustering with applications in biological networks. We will propose new models, numerical algorithms, and theoretical analysis for the method based on spectral clustering. Then, we will study the common factors for some cancers, and the variation of the modules in the development process of cancer. This project is interdisciplinary, including applied mathematics, statistics, and biology. We hope this proposal can be approved, so that with the support of NSFC, we can make more progress and contribute to development of the related research fields.

英文关键词： numerical methods;matrix analysis;spectral clustering;biological network;module identification