Inference of gene loss rates after whole genome duplications at early vertebrates through ancient genome reconstructions

The famous 2R hypothesis was first proposed by Susumu Ohno in 1970. It states that the two whole genome duplications had shaped the genome of early vertebrates. The most convincing evidence for 2R hypothesis comes from the 4:1 ratio chromosomal regions that have preserved both gene content and order in vertebrates compared with closely related. However, due to the shortage of such strict evidence, the 2R hypothesis is still under debates. Here, we present a combined perspective of phylogenetic and genomic homology to revisit the hypothesis of 2R whole genome duplications. Ancestral vertebrate genomes as well as ancient duplication events were created from 17 extant vertebrate species. Extant descendants from the duplication events at early vertebrates were extracted and reorganized to partial genomes. We then examined the gene order based synteny, and projected back to phylogenetic gene trees for examination of synteny evidence of the reconstructed early vertebrate genes. We identified 7877 ancestral genes that were created from 3026 duplication events at early vertebrates, and more than 50% of the duplication events show synteny evidence. Thus, our reconstructions provide very strong evidence for the 2R hypothesis. We also reconstructed the genome of early vertebrates, and built a model of the gene gains and losses in early vertebrates. We estimated that there were about 12,000 genes in early vertebrates before 2R, and the probability of a random gene get lost after the first round of whole genome duplication is around 0.45, and the probability of a random gene get lost after the second round of whole genome duplication is around 0.55. This research provides convincing evidence for the 2R hypothesis, and may provide further insights in vertebral evolution. Data availability: https://github.com/haimingt/Ohnologs-and-2R-WGD