Project lead: Vitalina Komashko, Xia Yang

Project team: Bin Zhang, Jun Zhu, Justin Guinney

Abstract: Ovarian adenocarcinoma is the ninth most common cancer among women and it accounts for more death than any other types of cancer of the female reproductive system. Most of the women are usually asymptotic or present with non-specific syndromes, making early diagnosis impossible. Despite surgical resection and aggressive chemotherapy the majority of patients will suffer from recurrence of the disease and the 5-year survival rate is less than 40%. Currently our understanding of the pathogenesis is still limited and early diagnostic markers are yet to be identified. The cancer genome contains many irreversible (on the level of nucleotides, copy number variations etc) and reversible aberrations (epigenetic changes that include changes in the pattern of histone code, DNA methylation as well as regulation of microRNA expression). Epigenetic changes in the genome are affected by genetic variations and in turn have an effect on gene and microRNA expression. To our knowledge, the role of epigenetic mechanisms such as DNA methylation and microRNA has not been fully explored for ovarian cancer. We hypothesize that changes in DNA methylation status play an important role in the development of ovarian cancer as well as patient survival by altering gene expression and the associated gene network states. The ovarian cancer dataset from The Cancer Genome Atlas (TCGA) is comprised of genetic, epigenetic, and clinical traits, providing a unique opportunity to address this hypothesis. Identifying ovarian cancer-specific DNA methylation aberrations as well as their relationship with other types of molecular traits and clinical traits will not only help to understand the underlying molecular mechanism of ovarian cancer but also help uncover novel biomarkers.

Specific aims:

Aim 1: - Investigate the role of DNA methylation in ovarian cancer by correlating methylation changes with cancer status and other relevant clinical traits.

Identify differential methylation loci between ovarian cancer cases and controls
Identify methylation loci that are correlated with traits related to cancer severity, survival, and response to therapy

Aim 2: - Determine the relationships between DNA methylation and other types of molecular traits.

Identify genes whose expression levels are correlated with differential DNA methylation loci identified in Aim 1
Identify microRNAs whose expression levels are correlated with the differential DNA methylation loci
Identify single nucleotide polymorphisms (SNPs) that are associated with DNA methylation (mSNPs), gene expression (eSNPs), and microRNA (miSNPs)
Establish causal relationships between DNA methylation and gene/microRNA expression using a conditional correlation-based causality test

Aim 3: - Construction of co-expression and co-methylation networks to identify regulatory mechanisms responsible for changes in gene expression and DNA methylation profiles.

Construct co-expression network and identify co-expression network modules that are correlated with DNA methylation, candidate regulatory genes, and clinical traits
Construct co-methylation network and identify co-methylation network modules that are correlated with expression of epigenes and clinical traits

Aim 4: - Identify composite predictive markers of clinical phenotypes of ovarian cancer by simultaneously considering expression, genetic and epigenetic data.

Construct predictive models of survival using DNA methylation, microRNAs, SNPs, and copy number variations (CNVs)
Experimental validation of the markers using tissue samples.

Complete description of the research plan with background and proposed methods can be found here. Also the latest brainstorming ideas about each specific aim of the project are outlined on the slides here.

Table of contents:

Data description and collection
Normalization of DNA methylation data
Analysis of clinical traits
1. Survival
Analysis of gene expression data, smaller set of patients (~384)
1. Correlation with clinical traits
  1. Survival
2. And DNA methylation
  1. Correlation
3. Gene coexpression network
Analysis of gene expression data, larger set of patients (522)
1. Data retrieval and normalization
2. Building a coexpression network
DNA methylation: correlation with clinical traits.
1. Survival
DNA methylation network analysis (comethylation)
DNA methylation analysis: expression methylation as M or beta value

Summary of DNA methylation data analyses from TCGA ovarian cancer paper (PMID:21720365)

This project was discontinued in December 2011 because of the strong batch effect in DNA methylation data which was also highly correlated with clinical variables.

Browser not supported