ANALYSIS OF DNA METHYLATION AND RNA-SEQ DATA FOR PROSTATE ADENOCARCINOMA:AN INTEGRATIVE APPROACH

Abstract

Epigenetics is rapidly gaining recognition as it accounts for the change in phenotype without any change in the genotype or the DNA sequence. These are the heritable changes observed in gene expression without any change observed in the coding sequence. DNA methylation is the most extensively studied epigenetic mechanism which adds a methyl group to DNA at a cytosine base almost every time accompanied by a guanine base. These sites of methylation are also called “CpG” islands known to harbour promoter regions. Hence methylation at promoter regions directly affects the binding of DNA-binding protein thus inhibiting transcription and gene expression. DNA methylation is observed to play major roles in gene regulating mechanisms and gene silencing mechanisms. Therefore understanding the relationship between DNA methylation and gene expression becomes very important. As a proof of concept, Prostate Adenocarcinoma (PRAD) was chosen as the cancer to be studied as it is the second leading cause of death in men. DNA methylation (level 1) and Gene expression data (level 3) from The Cancer Genome Atlas (TCGA) were downloaded for 18 normal matched with tumor and 18 tumor matched with normal samples from batch 184. “R” programming language was used to integrate and analyse the data. “R – Bioconductor” packages “minfi” and “COHCAP” were used to find 453 differentially methylated regions with with p-value < 0.05, fdr < 0.05 and beta value (methylation) > 0.2. The gene expression data was integrated with matched TCGA IDs and Pearson correlation analysis was carried out. 180 significant correlations were identified, out of which 112 correlations were chosen by applying stringent rules like correlation < -0.5, p value < 0.001 and false discovery rate < 0.001. Upon visual inspection of the results, 74 correlations were finally filtered and functional enrichment was carried out. It was discovered that genes "GSTP1" and "FGFR2" are already known to be involved in prostate cancer pathway and progression and these genes were present in the final filtered significant correlations. This approach may indicate the involvement of other novel genes in the prostate cancer pathway for which experimental validation must be carried out.