Differentially methylated topologically associating domains in brain tumors

Abstract

In recent years, cancer has become one of the main causes of human death all over the world. Cancer is a generation of unhealthy cells that neither grow nor operate normally. Central nervous system (CNS) tumor is a creation of abnormal cells either in the spinal cord or in the brain. Tumor is result of chromosome changes that includes change in chromatin structure and change in genome. In epigenomic modification, change in gene expression happens without changing DNA sequences; it is result of other events such as modification of histone or DNA methylation. There are special regions in the structure of chromosomes called Topologically Associating Domain(s) (TADs). One main role of them is regulating gene expression. Modifying chromatin structure in a TAD regions affects gene expression. It is likely that aberrant gene expression converts a cell into an unhealthy cell. One of the main factors that create epigenomic modification in a chromosome is the methylation of cytosine nucleotides in DNA. Microarray is a technique for measuring DNA methylation using probes. It is possible that aberrant methylation of cytosine changes gene expression. The objective of this project has been to determin differentially methylated TADs. Collecting information of differentially methylated TADs is important because the information helps finding differences between different tumor types and normal samples. In this project, epigenomic modification of CNS tumor has been investigated. Data of beta values (proportion of methylated DNA for each probe) has been imported into R programming language; then, by implementing different packages of R, computational and statistical analyses have been performed.There are three TAD boundaries samples, Neural Progenitor Cells (NPC), Embryonic Stem Cell (ESC) and Cortex. In addition, tumors/controls samples from 91 different methylation classes and a data set of normalized beta values had been provided. At the first step, filtering of TADs based on number and distance of overlapped probes was performed. Then, for visualizing beta value distribution in TAD or tumors/controls level, violin plots and box plots were drawn. Since there are too many TADs or tumors to be shown in a figure, TADs or tumors were randomly selected; therefore, resulted violin plots or box plots are only estimation of beta value distribution. In the next step, three sets heat maps of all chromosomes over all tumors/controls were drawn with three arithmetic calculations: mean, median and proportion of probes with beta value more than 0.6) to decide which of these method is suitable for following analysis. Next, density and heat maps of TADs with larger beta value variation were drawn to detect difference in density of beta values in different chromosome and visualize properties of special group of TADs with larger beta value variations respectively. In the statistical analysis section, Medulloblastoma (MB), Atypical Teratoid/Rhabdoid Tumor (ATRT) and Choroid Plexus tumor( PLEX) tumor samples were compared in pairs. There have been more differentially methylated TADs when MB tumor was compared with ATRT and PLEX than when ATRT with PLEX was compared. Then, by drawing karyotypes, differentially methylated TADs were observed across autosomal chromosomes. In summary, it has been concluded that there are meaningful differences in methylation level of different TADs and these differences also appear as differences between tumor types.