Gene expression analysis of microarray data : a case study of papilllary thyroid carcinoma data
Microarray technology allows researchers to monitor the mRNA transcription levels of thousands of genes in parallel which opens the door for more advanced cancer research. This thesis focuses on a case study of papillary thyroid carcinoma data. Fourteen publicly available Affymetrix microarray data sets were used where seven samples were collected from normal thyroid tissue and the remaining seven were collected from papillary thyroid carcinoma tissue. The present study compared the results obtained from three different normalization processes: MAS5.0, RMA and GCRMA in detecting differentially expressed genes under two conditions. Internal consistencies within the methods as well as the results across three methods were compared. Statistical packages 82.5.1 and Bioconductor 2.08 are used to perform the data analysis. Each step of normalization with MAS5.0 and RMA is described. Statistical package Limma is used to fit a linear model. Finally an empirical Bayes method is used to detect the significantly differentially expressed genes. First, considering all genes a comparison is made among the three normalization methods where RMA and GCRMA showed the maximum agreement in detecting differentially expressed genes. Then using unspecified filtering process a set of genes was selected and the whole process was replicated where the top fifty differentially expressed genes did not show any overlap with each other.