Epigenetic Variability Confounds Transcriptome but not Proteome Profiling for Coexpression-based Gene Function Prediction

Genes are often co-expressed with their genomic neighbors, even if these are functionally unrelated. For small expression changes driven by genetic variation within the same cell type, non-functional mRNA co-expression is not propagated to the protein level. However, it is unclear if protein levels are also buffered against any non-functional mRNA co-expression accompanying large, regulated changes in the gene expression program, such as those occurring during cell differentiation. Here, we address this question by analyzing mRNA and protein expression changes for housekeeping genes across 20 mouse tissues. We find that a large proportion of mRNA co-expression is indeed non-functional and does not lead to co-expressed proteins. Chromosomal proximity of genes explains a proportion of this non-functional mRNA co-expression. However, the main driver of non-functional mRNA co-expression across mouse tissues is epigenetic similarity. Both factors together provide an explanation for why monitoring protein co-expression outperforms mRNA co-expression data in gene function prediction. Furthermore, this suggests that housekeeping genes translocating during evolution within genomic subcompartments might maintain their broad expression pattern.