乳腺癌组织学分级下目标基因提取及转录调控网络构建

Aimed genes' extraction and construction of transcription regulatory network under different grading levels of breast cancer

目的乳腺癌类型和分级多样性导致其预后差别显著,探寻乳腺癌不同分级情况下的基因表达差异及调控关系能够为乳腺癌致病机制的发现提供重要依据。方法对不同分级下的乳腺癌基因表达数据利用快速独立成分分析(FastICA)方法提取特征基因,并结合人类蛋白质相互作用(PPI)数据选取目标基因。在此基础上,结合转录因子对靶基因调控的先验信息,利用网络成分分析(NCA)方法对与乳腺癌发病有密切关系的转录因子及其靶基因构建转录调控网络。结果筛选出的基因经过数据库验证与乳腺癌相关的占48.15%,构建的调控网络发现了多个转录因子及靶基因在不同分级情况下的活性变化趋势。结论 FastICA算法结合PPI数据提取目标基因的方法较为有效,通过NCA算法构建的转录调控网络为研究乳腺癌发生发展机制提供了新的方法。

Objective The diversities of breast cancer types and grading levels lead to distinct difference for breast cancer prognosis. Studying the gene difference expression and regulatory relationship among genes under different grading levels of breast cancer could provide an important basis for finding breast cancer pathogenesis. MethodsUsing fast independent component analysis （ FastICA ） method to extract feature genes of gene expression data of breast cancer, and then selected the aimed genes by combining with human protein-protein interaction data （ PPI） . On this basis, introducing prior information which described regulatory relationships about how transcription factors regulated their target genes, we continued to analize transcription factors and their target genes, which were closely associated with the incidence of breast cancer, by using network components analysis method （ NCA） , and then constructed a transcriptional regulatory network. Results Selected aimed gene which was closely associated with breast cancer is about 48.15%, that had been validated by breast cancer database. And from the built regulatory network, found out the activity change trend of multiple transcription factors and their target genes under different grading levels. Conclusion FastICA algorithm combined with PPI data for extracting aimed gene is a relatively ef-fective method. Simultaneously, constructing transcription regulatory network with NCA method provides a novel way for studying progression mechanism of breast cancer.