Study On The Correlation Of Gene Expression Profiles Of Breast Cancer

Idientifing differentially expressed genes associated with cancer initiation and progression from high-throughput protein and gene expression profiles is an important task of cancer research. Although related researches have made great progress, we have found the problem: Only a few differential genes found in gene expression profiles of different studies on the same cancer are overlapping. That means reproducibility is low. The problem of low repeatability is urgently needed to be resolved. Otherwise it will make the high-throughput technology questioned, and further clinical applications will also be affected.Analyzing the reasons, we found that one of the reasons for low reproducibility is caused by perturbation of the error data. The perturbation of gene expression profiles consists of experimental error and biological error. The experimental error can be controlled through experimental means, technology and equipment improvement. The biological error comes from single nucleotide polymorphism and diversity in gene expression regulation, and it can not be avoided. It is the main part of the perturbation error. Therefore, a new statistical method for screening differentially expressed genes to improve the repeatability to found the different cancer related genes is very significant. Some studies found that although the overlap rate of differentially expressed genes screened in different studies is low, they are consistent in the sense of correlation. This suggests that we may study in correlation to improve the overlap rate of screened cancer differentially expressed genes.This paper selects the expression profile data of breast cancer as the research object. By studying the correlation between data, the improved method for screening differentially expressed genes is proposed.1. Nowadays, False Discovery Rate(FDR) is widely used in choosing differentially expressed genes from expression profile data. And it usually uses Adaptive Linear Step-Up(ALSU) to calculate and control program. In this paper, various kinds of correlation coefficient were defined and calculated in breast cancer expression profile. The P value adjusted by FDR was increased, and the overall correlation coefficient of the genes was decreased. The differentially gene function enrichment analysis showed that, the average correlation coefficient of differentially genes in significant enrichment pathway is higher than the average correlation coefficient of the whole differentially gene.2. Based on the FDR method, the improved method screening differentially expressed gene in breast cancer is put forward in this parper. The results show that gene duplication rate of the improved method is higher than gene duplication rate of the FDR method. And the known breast cancer associated genes in these repeated genes are also more than those by FDR. At the same time, the result of function enrichment is also better than the result by FDR in screening differential gene. Thus, improved method based on correlation is robust and stable, the effect is better than the traditional FDR method.In summary, this paper studies the correlation of breast cancer gene expression data. A improved method on FDR for screening differentially expressed genes combined with the method of correlation is proposed. The repeatability of different cancer expression data is improved. It has a positive significance to study cancer initiation and progression and solve clinical problems through high-throughput technology.