Phylogenetic expression profiling reveals widespread coordinated evolution of gene expression

Phylogenetic profiling, which infers functional relationships between genes based on patterns of gene presence/absence across species, has proven to be highly effective. Here we introduce a complementary approach, phylogenetic expression profiling (PEP), which detects gene sets with correlated expression levels across a phylogeny. Applying PEP to RNA-seq data consisting of 657 samples from 309 diverse unicellular eukaryotes, we found several hundred gene sets evolving in a coordinated fashion. These allowed us to predict a role of the Golgi apparatus in Alzheimer's disease, as well as novel genes related to diabetes pathways. We also detected adaptive evolution of tRNA ligase levels to match genome-wide codon usage. In sum, we found that PEP is an effective method for inferring functional relationships - especially among core cellular components that are never lost, to which phylogenetic profiling cannot be applied - and that many subunits of the most conserved molecular machines are coexpressed across eukaryotes.