Integrative network analysis of TCGA data for ovarian cancer

Abstract

BackgroundOver the past years, tremendous efforts have been made to elucidate the molecular basis of the initiation and progression of ovarian cancer. However, most existing studies have been focused on individual genes or a single type of data, which may lack the power to detect the complex mechanisms of cancer formation by overlooking the interactions of different genetic and epigenetic factors.ResultsWe propose an integrative framework to identify genetic and epigenetic features related to ovarian cancer and to quantify the causal relationships among these features using a probabilistic graphical model based on the Cancer Genome Atlas (TCGA) data. In the feature selection, we first defined a set of seed genes by including 48 candidate tumor suppressors or oncogenes and an additional 20 ovarian cancer related genes reported in the literature. The seed genes were then fed into a stepwise correlation-based selector to identify 271 additional features including 177 genes, 82 copy number variation sites, 11 methylation sites and 1 somatic mutation (at gene TP53). We built a Bayesian network model with a logit link function to quantify the causal relationships among these features and discovered a set of 13 hub genes including ARID1A, C19orf53, CSKN2A1 and COL5A2. The directed graph revealed many potential genetic pathways, some of which confirmed the existing results in the literature. Clustering analysis further suggested four gene clusters, three of which correspond to well-defined cellular processes including cell division, tumor invasion and mitochondrial system. In addition, two genes related to glycoprotein synthesis, PSG11 and GALNT10, were found highly predictive for the overall survival time of ovarian cancer patients.ConclusionsThe proposed framework is effective in identifying possible important genetic and epigenetic features that are related to complex cancer diseases. The constructed Bayesian network has identified some new genetic/epigenetic pathways, which may shed new light into the molecular mechanisms of ovarian cancer.Electronic supplementary materialThe online version of this article (doi:10.1186/s12918-014-0136-9) contains supplementary material, which is available to authorized users.

Next-Generation Sequencing Approach in Methylation Analysis of HNF1B and GATA4 Genes: Searching for Biomarkers in Ovarian Cancer

“…Much effort is thus devoted to detecting suitable genes for this role. Many different genes have been identified as being hypermethylated and silenced in ovarian carcinoma [7,8]. …”

New analysis framework incorporating mixed mutual information and scalable Bayesian networks for multimodal high dimensional genomic and epigenomic cancer data

“…For the purposes of this particular analysis, we decided to concentrate on three types of molecular data, one discrete (somatic mutations) and two -continuous (RNA-seq gene expression, and promoter methylation). This selection is reflective of the recent trends in multimodal cancer data analyses [21,26], makes sense in the broad cancer genetics context [27][28][29][30][31][32], and underscores the comparative importance of the methylation molecular data [29].…”

“…While alleviating the scalability issue, this, however, could potentially "throw away the wheat with the chaff", especially if the variable selection process [22,23] is of a simplistic and overly too restrictive kind (e.g., a statistically conservative univariate filter). There are three possible ways to address this, namely: (i) increase the scalability of the BN modeling to genomic data levels (possible, but impractical for frequent/serial analyses), (ii) incorporate higher-order interactions into the variable selection step (thus "upgrading" it from the simple filter to the wrapper [23-25] ---this is the solution implemented in [21]), or (iii) adjust the transition boundary between the variable selection step and the BN modeling step, depending on the investigators' computational resources and the nature (dimensionality, sparseness, heterogeneity) of the actual data. It is the third analytical strategy that we propose in this study, with the goal to achieve the optimal compromise between the computational practicality and modeling exhaustiveness.…”

Coding and non-coding gene regulatory networks underlie the immune response in liver cirrhosis

“…Gene regulatory network analysis is a stable and dynamic hierarchical system model that is increasingly used to study various diseases. This model can combine high-throughput gene expression data, gene interaction information, and relevant analysis and calculation methods[8, 9]. From a macro perspective, the gene regulatory network underlying the development and progression of liver cirrhosis is comprised of many individual regulatory relationships.…”