UC Davis

The evolution of genes de novo from nongenic sequence has become more appreciated as a mechanism of gene origin. Most studies have used few genotypes or distantly related species to infer de novo genes, which may lead to undersampling short-lived or uncommon alleles. Here, we examine the messy beginnings of de novo genes as they arise in the population. We use a comparative transcriptomics approach with an expansive pool of tissue and genotype-specific sequences to capture the characteristics of rarer alleles. To do so, we used RNA-seq in accessory gland tissue from a core set of 29 D. melanogaster genotypes, as well as additional genotypes from two of its close relatives. We show the majority of unannotated genes are lowly transcribed and lack fixed expression within species. Some rare alleles can be found in multiple species, which is hard to reconcile with models of a discrete, single new allele origin and/or fast rates of loss. As a technical consideration, the rarity of many unannotated genes may inflate the de novo classification rate, as evidence of their expression in outgroups is likely to be missed. Next, we explored whether the location of these genes may suggest a mechanism for their origin. Surprisingly, we found within tissue-specific genes, gene age was inversely correlated with proximity to other tissue-specific genes. This does not support the idea that the youngest de novo genes simply reutilize the tissue-specific regulatory elements near annotated genes. Despite this, younger de novo genes showed an enrichment for some tissue-specific binding motifs, which could indicate convergent evolution of regulatory sequence elsewhere in intergenic space.