Inferring non-neutral regulatory change in pathways from transcriptional profiling data
Joshua G. Schraiber, Yulia Mostovoy, Tiffany Y. Hsu, Rachel B. Brem
(Submitted on 19 Apr 2013)
An outstanding question in comparative genomics is the evolutionary importance of gene expression differences between species. Rigorous molecular-evolution methods to infer evidence for natural selection from transcriptional profiling data are at a premium in the field, and to date, phylogenetic approaches have not been well-suited to address the question in the small sets of taxa profiled in standard surveys of gene expression. To meet this challenge, we have developed a strategy to infer evolutionary histories from expression data by analyzing suites of genes of common function. In a manner conceptually similar to molecular-evolution models in which the evolutionary rates of DNA sequence at multiple loci follow a gamma distribution, we modeled expression of the genes of an a priori-defined pathway with rates drawn from an inverse-gamma distribution. We then developed a fitting strategy to infer the parameters of this distribution from expression measurements, and to identify gene groups whose expression patterns were consistent with evolutionary constraint or rapid evolution in particular species. Simulations confirmed the power and accuracy of our inference method. As an experimental testbed for our approach, we generated and analyzed transcriptional profiles of four Saccharomyces yeasts. The results revealed pathways with signatures of constrained and accelerated regulatory evolution in individual yeasts, and across the phylogeny, highlighting the prevalence of pathway- level expression change during the divergence of yeast species. We anticipate that our pathway-based phylogenetic approach will be of broad utility in the search to understand the evolutionary relevance of regulatory change.