Utilizing RNA-Seq Data for Cancer Network Inference

TL;DR

This paper presents a probabilistic framework using graphical lasso to infer gene interaction networks from RNA-Seq data, aiding understanding of cancer gene interactions.

Contribution

It introduces a novel application of graphical lasso with maximum entropy principles to construct biologically relevant gene networks from high-dimensional RNA-Seq data.

Findings

Successful network inference from RNA-Seq data

Validation through gene ontology and pathway analysis

Identification of key cancer gene interactions

Abstract

An important challenge in cancer systems biology is to uncover the complex network of interactions between genes (tumor suppressor genes and oncogenes) implicated in cancer. Next generation sequencing provides unparalleled ability to probe the expression levels of the entire set of cancer genes and their transcript isoforms. However, there are onerous statistical and computational issues in interpreting high-dimensional sequencing data and inferring the underlying genetic network. In this study, we analyzed RNA-Seq data from lymphoblastoid cell lines derived from a population of 69 human individuals and implemented a probabilistic framework to construct biologically-relevant genetic networks. In particular, we employed a graphical lasso analysis, motivated by considerations of the maximum entropy formalism, to estimate the sparse inverse covariance matrix of RNA-Seq data. Gene ontology, pathway enrichment and protein-protein path length analysis were all carried out to validate the biological context of the predicted network of interacting cancer gene isoforms.