项目名称: 基于L1范数约束稀疏矩阵分解的基因表达谱数据分析
项目编号: No.61272339
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 郑春厚
作者单位: 安徽大学
项目金额: 80万元
中文摘要: 基因表达谱数据分析能够帮助人们建立新的疾病诊疗方法,深入了解复杂的基因调控关系。为充分挖掘基因表达谱数据中蕴含的信息,本项目中,我们拟开展以下研究:(1) 结合基因表达数据特点,利用基因只被一小部分调控因子调控的"稀疏性",形成L1-范数约束条件,设计出适合基因表达谱数据的稀疏矩阵分解算法;(2)利用肿瘤基因表达谱数据矩阵的稀疏分解因子,结合基因调控的先验信息,对肿瘤样本进行聚类分析,为肿瘤临床诊断和治疗提供科学依据;(3)利用模式生物基因表达数据矩阵的稀疏因子,对基因进行功能聚类分析,发现具有特定功能的转录模块;(4)利用ChIP-chip数据,形成基因表达谱数据稀疏矩阵分解的初始化条件,对模式生物基因表达谱数据矩阵进行分解,并根据所得到连接权值因子矩阵初步构建一些简单模式生物的基因调控网络。
中文关键词: 基因表达谱;稀疏矩阵分解;差异表达模块;基因突变;
英文摘要: Gene expression data analysis is helpful for finding new method of disease diagnosis and treatment, and helpful for deeply understanding the complicated gene regulation. In order to sufficiently mine the knowledge in gene expression data, in this research, we will study the following terms: (1)Using the "sparse attribute" of gene regulation, integrating the characteristic of gene expression data, to form the L1-norm constraint. Based on which, the novel sparse matrix decomposition algorithm suit for gene expression data will be proposed.(2)Using the sparse factor of tumor gene expression data matrix, combining the prior information of gene regulation, to cluster the tumor samples and provide the basis for tumor diagnosis and treatment. (3) To identifying the transcriptional modules by clustering the genes based on the sparse factor of model organism gene expression data matrix. (4) Using ChIP-chip data to initialize the sparse matrix decomposition for model organism gene expression data, then using the connectivity strength factor matrix of which to reconstruct the gene regulatory network.
英文关键词: Gene expression profile;Sparse matrix;Differentially expression modules;Gene mutation;