Towards the Quantitative Study of Polyploid Genome Content

Abstract

Gene duplication is a key feature of genome content evolution in floweringplants, constantly producing novelty through small scale duplications and through large influxes from whole-genome duplications (WGD). These saltational changes may fuel adaptation, as the redundancy of duplicate gene copies allows for increased diversity and the evolution of novel functions. Such diversity may explain why polyploids have faster niche differentiation than their diploid relatives and why paralogs—the result of gene duplication—have been instrumental to repeated global adaptation to abiotic stresses. However, it is currently unknown how significant a role WGDs play in long term plant evolution, and if that role is mediated through the gene content turnover it produces. Two key roadblocks to answering this question are the limitations of currently available methods for both inferring WGDs within the history of a lineage and identifying the gene content they produced. To-date few studies have acknowledged that the species-tree itself is a hypothesis that inherently imparts uncertainty onto any WGD inference, and even fewer studies assess how sensitive inference methods are to such uncertainty. To fill this gap in our understanding I utilized several popular WGD inference tools under two alternative species tree topologies to infer WGDs across over 400 angiosperm species. From these analyses I uncovered that each plant lineage has experienced on average ~3.5 WGDs in their history, but their precise placement can vary by ~20% depending on the species tree used. Following initial inference, identifying the gene content these WGDs produce is a tradeoff between accuracy and computational times that are infeasible for phylogeny wide studies. To circumvent this limitation I developed the machine learning tool Frackify (Fractionation Classify) which quickly and accurately identifies paleologs—paralogs formed by WGD—within genomes. To begin assessing how paleologs have shaped plant adaptation, I used Frackify to identify paleologs in 22 crop species with prior reported candidate domestication gene lists. Paleologs were enriched in candidate domestication gene lists of over half of the species tested, and was the only class of genes to show this repeated pattern. The body of my work suggests that a more nuanced understanding of how paralogs respond to selection is necessary to address how WGDs have shaped flowering plant evolution. Classic models of dosage sensitivity, sub- and neofunctionalization cannot sufficiently explain how paralogs respond to positive selection and contribute to adaptation. Future efforts are needed to disentangle what properties make paleologs unique relative to other paralogs in their propensity to respond to positive selection.