Network-based inference of cancer progression from microarray data

Authors:
Yongjin Park;Stanley Shackney;Russell Schwartz
Affiliations:
Department of Biological Sciences, Carnegie Mellon University, Pittsburgh, PA;Departments of Human Oncology and Human Genetics, Drexel University School of Medicine, Pittsburgh, PA;Department of Biological Sciences, Carnegie Mellon University, Pittsburgh, PA
Venue:
ISBRA'08 Proceedings of the 4th international conference on Bioinformatics research and applications
Year:
2008

Citing 4
Cited 1

Introduction to algorithms

Introduction to algorithms
BiNGO: a Cytoscape plugin to assess overrepresentation of Gene Ontology categories in Biological Networks

Bioinformatics
Clustering microarray gene expression data using weighted Chinese restaurant process

Bioinformatics
Learning graphical model structure using L1-regularization paths

AAAI'07 Proceedings of the 22nd national conference on Artificial intelligence - Volume 2

Network-Based Inference of Cancer Progression from Microarray Data

IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB)

Quantified Score

Hi-index	0.00

Visualization

Abstract

Cancer cells exhibit a common phenotype of uncontrolledcell growth, but this phenotype may arise from many different combinationsof mutations. By inferring how cells evolve in individual tumors, aprocess called cancer progression, we may be able to identify importantmutational events for different tumor types, potentially leading to newtherapeutics and diagnostics. Prior work has shown that it is possibleto infer frequent progression pathways by using gene expression profilesto estimate "distances" between tumors. Individual mutations can,however, result in large shifts in expression levels, making it difficult toaccurately identify evolutionary distance from differences in expression.Here, we apply gene network models in order to improve our ability toestimate evolutionary distances from expression data by controlling forcorrelations among co-regulated genes. We test two variants of this approach,one using full regulatory networks inferred from a candidate geneset and the other using simplified modular networks inferred from clustersof similarly expressed genes. Application to a set of E2F-responsivegenes from a lung cancer microarray data set shows a small improvementin phylogenies when correcting from the full network but a more substantialimprovement when correcting from the modular network. Theseresults suggest that a network correction approach can lead to betteridentification of tumor similarity, but that sophisticated network modelsare needed to control for the large hypothesis space and sparse datacurrently available.