复杂网络中子网提取的排名—重构方法

发布时间：2018-02-08 10:25

本文关键词： 复杂网络蛋白质相互作用基因芯片 Fisher判别　出处：《清华大学》2011年硕士论文　论文类型：学位论文

【摘要】：基因芯片技术是近年来迅速发展起来的高通量技术,可以一次测得上万个基因的相对表达值.基因芯片已经成为生物信息学研究的重要手段.通过比较正常人与癌症病人的芯片数据,有望找到差异表达的基因,从而从基因层次了解疾病的发生发展.由于人体的大部分基因并不差异表达,如何有效地从大量的噪声中提取出差异表达的部分一直是生物学家关注的问题.人们已经发现,通过加入先验信息,可以有效地去除噪声.蛋白质-蛋白质相互作用网络(PPI)描述了蛋白质之间的关系.寻找PPI网络中在不同条件下有差异表达的子网实际上正是融合了网络这一先验信息.目前流行的做法是根据不同条件下的基因表达值对所有可能的子网定义一个分数,然后用模拟退火或遗传算法寻找得分最高的子网.本文提出了一种与现有文献不同的提取子网的方法,并进行了理论上的证明和大量的模拟试验.这种方法的基本思路是对网络中的边给出一个排名,排名的高低反映了这条边在局部网络中的贡献大小.然后利用这个排名将网络重新构建起来,从中提取出差异表达的子网.与现有的方法相比,作者提出的算法将网络的节点和边放在局部的网络中考虑,而不是单独拿出来分析,所设计的算法速度更快,在从PPI网络中提取子网时考虑了网络的拓扑性质.初步的模拟试验显示该算法的查全率和查准率比已有的算法要高,更全面的结论还需要进一步验证.
[Abstract]:Gene chip technology is a high-throughput technology developed rapidly in recent years. The relative expression of tens of thousands of genes can be measured at a time. Gene chips have become an important means of bioinformatics research. By comparing the microarray data of normal people with cancer patients, it is expected to find differentially expressed genes. So as to understand the occurrence and development of disease from the gene level. Since most genes in the human body are not differentially expressed, how to effectively extract the differential expression part from a large amount of noise has always been a concern of biologists. By adding prior information, The protein-protein interaction network describes the relationship between proteins. Looking for subnets with different expressions under different conditions in PPI networks is actually a priori fusion of networks. Information. The prevailing practice is to define a score for all possible subnets based on gene expression values under different conditions, Then simulated annealing or genetic algorithm is used to find the subnet with the highest score. The basic idea of this method is to give a ranking of the edges in the network. The rank level reflects the contribution of the edge to the local network. Then the ranking is used to reconstruct the network to extract subnets of differential expressions. The algorithm proposed by the author takes the nodes and edges of the network into account in the local network, instead of being taken out and analyzed separately, and the algorithm is designed more quickly. The topological properties of the network are taken into account in extracting subnets from PPI networks. The preliminary simulation results show that the recall and precision rates of the proposed algorithm are higher than those of the existing algorithms, and the more comprehensive conclusions need to be further verified.
【学位授予单位】：清华大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：R346;O157.5

【共引文献】