Algorithmic and Artificial Intelligence Methods for Protein Bioinformatics

Chapter 26

Biological networks–based analysis of gene expression signatures*

GANG CHEN and JIANXIN WANG

26.1 Introduction

Diagnostic and prognostic gene signatures for complex diseases is a major step toward better personal medicine. A gene signature is a group of genes whose expression pattern represents the status of a gene expression disease [1]. Identification of the gene signatures of disease subtypes, risk stratification, pathologic parameters, and clinical outcomes has the potential to help physicians and surgeons find a personal optimized treatment, avoid unnecessary medication, and reduce costs [2, 3].

Various gene signatures are developed for various complex diseases, especially cancer. Since researchers found that gene expression signatures were able to predict clinical¹ outcome of breast cancer in 2002 [4, 5], this method has become a hot topic and attracted the attention of both biologists and oncologists. Signatures for various phenotypes, such as poor prognosis [5], invasiveness [6], recurrence [7], and metastasis [8, 9], have been experimentally derived from patient groups and biological hypotheses. MammaPrint, a fully commercialized microarray-based 70-gene signature for breast cancer that was developed by Agendia (http://www.agendia.com/), was approved by FDA [10] in 2009. Because of rapid development of high-throughput tecniques, such as microarry, the number of gene signatures has grown quite rapidly. The first release of GeneSigDB [11], published in August ...