基于植物胁迫响应基因表达数据和GO术语结合的特征选择及调控网络研究

项目名称： 基于植物胁迫响应基因表达数据和GO术语结合的特征选择及调控网络研究

项目编号： No.61472061

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 计算机科学学科

项目作者： 孟军

作者单位： 大连理工大学

项目金额： 80万元

中文摘要： 植物在生长发育过程中，因各种逆境、病虫害造成的损失不计其数，探寻主要的抗性基因，掌握它们之间的调控机制并加以利用，一直是人们的夙愿。高通量测序技术的飞速发展，使得基因表达数据量与日俱增，但计算分析技术相对滞后，而且这些数据中存在大量的噪声和冗余。本项目拟根据粗糙集理论，针对植物胁迫响应基因表达数据，结合GO术语建立邻域系统粗糙集模型，设计基于启发式信息的属性约简算法，选择重要的基本知识单元，从中探寻关键基因；建立能体现个体差异的分类器，提出自适应的集成分类器构建策略，运用基于权重的基分类器结果融合方法，获得具有高分类性能的关键基因；构建基因调控网络，再通过生物学实验验证，揭示各基因之间的调控关系，为植物抗性基因资源的有效利用奠定基础。

中文关键词： 基因表达数据；GO术语；特征选择；基因调控网络

英文摘要： In the process of plant growth and development, the losses caused by various adversities and disease are countless. To seek the important resistant gene, grasp their regulatory mechanisms and use them are still our long-cherished wish. The development of high-throughput sequencing technology brings us more and more gene expression data. But the calculation and analysis of biological data lag far behind the amount of data growth. There are a lot of noise and redundant in the large amount of plant stress response gene expression data. This project will apply rough set theory to integrate gene expression data with GO terms and build neighborhood system based rough set model. To design attribute reduction algorithm based on heuristic information. First, we select significant elemental knowledge units and then select the significant genes. Then, classifiers are constructed according to the difference between individual gene information and propose a self-adapting ensemble classifier building strategy and a weight based basic classifier results fusion strategy in order to find the minimum gene subset with high classification performance. Last, we construct gene regulatory networks and verify the result by biology experiment. It will reveal the regulation relationship between the genes related to plant stress response. This project will lay the foundation for the efficient use of plant resistant gene.

英文关键词： Gene expression data;GO terms;Feature selection;Gene regulatory network