Mammalian Genes Preferentially Co-Retained in Radiation Hybrid Panels Tend to Avoid Coexpression

Ben-Yang Liao,Andrew Ying-Fei Chang

doi:10.1371/journal.pone.0032284

Abstract

Coexpression has been frequently used to explore modules of functionally related genes in eukaryotic genomes. However, we found that genetically interacting mammalian genes identified through radiation hybrid (RH) genotypes tend not to be coexpressed across tissues. This pattern remained unchanged after controlling for potential confounding factors, including chromosomal linkage, chromosomal distance, and gene duplication. Because >99.9% of the genetically interacting genes were identified according to the higher co-retention frequencies, our observation implies that coexpression is not necessarily an indication of the need for the co-presence of two genes in the genome, which is a prerequisite for cofunctionality of their coding proteins in the cell. Therefore, coexpression information must be applied cautiously to the exploration of the functional relatedness of genes in a genome.

Highlights

Coexpression refers to the coherent transcription of genes in spatial, temporal, or environmental dimensions [1,2,3]
Analyses in mammals [10], nematodes [11], and flies [12] indicate that many coexpressed gene clusters are unlikely to have originated to optimize gene regulation. It remains elusive whether the requirement for the co-presence of two genes in a genome is reflected by an increased level of coexpression and, whether coexpression predicts the cofunctionality of genes
To understand the biological implications of gene coexpression, we examined whether elevated coexpression predicts the need for the co-presence of genes in the genome, which is a prerequisite for the cofunctionality of their protein products

Summary

Introduction

Coexpression refers to the coherent transcription of genes in spatial, temporal, or environmental dimensions [1,2,3]. Genes that were preferentially co-retained in the genome consistently showed lower coexpression compared to other gene pairs. We measured coexpression between two genes from expression profile similarities across 63 human or 58 mouse tissues, using the equation ln[(1+CoExp)/(12CoExp)] (see Materials and Methods).

Results

Conclusion