Uncovering mechanisms of transcriptional regulations by systematic mining of cis regulatory elements with gene expression profiles

Qicheng Ma,Joseph D Szustakowski,Daniel Kemp,Nanguneri Nirmala,Gung-Wei Chirn,Adel Bakhtiarova,Penelope A Kosinski

doi:10.1186/1756-0381-1-4

Qicheng Ma, Joseph D Szustakowski + Show 5 more

Open Access

https://doi.org/10.1186/1756-0381-1-4

Copy DOI

Abstract

BackgroundContrary to the traditional biology approach, where the expression patterns of a handful of genes are studied at a time, microarray experiments enable biologists to study the expression patterns of many genes simultaneously from gene expression profile data and decipher the underlying hidden biological mechanism from the observed gene expression changes. While the statistical significance of the gene expression data can be deduced by various methods, the biological interpretation of the data presents a challenge.ResultsA method, called CisTransMine, is proposed to help infer the underlying biological mechanisms for the observed gene expression changes in microarray experiments. Specifically, this method will predict potential cis-regulatory elements in promoter regions which could regulate gene expression changes. This approach builds on the MotifADE method published in 2004 and extends it with two modifications: up-regulated genes and down-regulated genes are tested separately and in addition, tests have been implemented to identify combinations of transcription factors that work synergistically. The method has been applied to a genome wide expression dataset intended to study myogenesis in a mouse C2C12 cell differentiation model. The results shown here both confirm the prior biological knowledge and facilitate the discovery of new biological insights.ConclusionThe results validate that the CisTransMine approach is a robust method to uncover the hidden transcriptional regulatory mechanisms that can facilitate the discovery of mechanisms of transcriptional regulation.

Highlights

Contrary to the traditional biology approach, where the expression patterns of a handful of genes are studied at a time, microarray experiments enable biologists to study the expression patterns of many genes simultaneously from gene expression profile data and decipher the underlying hidden biological mechanism from the observed gene expression changes
The MotifADE algorithm works in three steps: (1) rank genes based on differential expression between two conditions using the signal-to-noise ratio as the difference metric in descending order; (2) For each motif, identify the group of genes whose promoter regions contains the motif; and (3) apply the two-tailed non-parametric Mann-Whitney rank sum test to determine if these genes tend to be enriched toward the top or bottom of the ranked list or tend to be randomly distributed on the list
Results for known transcriptional factors We use the mouse C2C12 cell differentiation dataset as a test case [6]

Summary

Introduction

Contrary to the traditional biology approach, where the expression patterns of a handful of genes are studied at a time, microarray experiments enable biologists to study the expression patterns of many genes simultaneously from gene expression profile data and decipher the underlying hidden biological mechanism from the observed gene expression changes. The MotifADE method [5] assumes that if up-regulated or down-regulated genes which contain certain transcriptional factor binding sites are co-ordinately regulated, changes in their expression levels could be explained by those transcriptional factors. The MotifADE algorithm works in three steps: (1) rank genes based on differential expression between two conditions using the signal-to-noise ratio as the difference metric in descending order (the signal-to-noise ratio is used as opposed to the fold change value based on the expression level since the former takes into account the standard deviation); (2) For each motif, identify the group of genes whose promoter regions contains the motif; and (3) apply the two-tailed non-parametric Mann-Whitney rank sum test to determine if these genes tend to be enriched toward the top or bottom of the ranked list (indicating association) or tend to be randomly distributed on the list (indicating no association)

Methods

Results

Discussion

Conclusion