| Summary: | 碩士 === 國防醫學院 === 公共衛生學研究所 === 103 === Background：Recent studies present many candidate genes of colorectal cancer by microarray data analysis. These genes are association between category, predict, prognosis and targeted therapy of colorectal cancer. These studies present many genes but some genes will overlap which maybe these genes in the same pathway.
Objective：Find the candidate genes which upstream or downstream in pathway by coefficient of variation, principal component analysis, logistic regression and Pearson correlation.
Methods：We collect microarray data of colorectal cancer from GEO database. Pooled difference microarray data to two dataset. Using Prediction Analysis of Microarray select candidate genes of colorectal cancer and establish prediction model. Association of candidate genes would effect significance of genes in logistic regression. Selected important candidate genes by logistic regression model. These important candidate genes and other genes in dataset analyze by Pearson correlation. We select some the relative gene in same pathway with candidate genes possibly by two difference microarray dataset. We utilize principal component analysis to check these candidate genes and the relative genes which whether in same pathway. Sorted these genes in same pathway and confirmed genes place in upstream or downstream.
Results： We utilize statistics analysis to find CWH43, GUCA2B, SCNN1B, AQP8 and SPIB in same pathway. IL6R, GUCA2B, SLC30A10, SCNN1B and AQP8 are in same pathway. AQP8 and CLDN8 are in same pathway. SPP1 and TNFAIP6 are in same pathway. SPIB, BANK1 and MS4A1 are in same pathway.
Conclusion：Using these analysis can find these genes in same pathway. We also confirm difference candidate gene in difference studies since these genes are in same pathway possibly.
|
|---|
