Evolutionary replay of duplicate-gene retention across independent whole-genome duplications

Whole-genome duplication presents ancient gene lineages with a choice: to retain both copies or revert to a single-copy state. It is uncertain whether the lineage's identity holds any information about this choice across separate duplications. Researchers have now placed the post-duplication retention within a common ancestral-lineage coordinate and compared the percentile ranks within each event, without the need for every lineage to be observed in every event.

Three separate angiosperm whole-genome duplications exhibited a reproducible replay of gene-lineage retention, with a global fixed-rank effect (T_replay = 0.210) and a bootstrap 95% confidence interval of 0.172-0.248. The signal also transferred among events that were not included in the analysis, suggesting that the information is transferable.

A reference-panel score, frozen before the observation of the target outcome, was able to predict retention in Apple/Pear and YeastWGD. However, this correlation was not consistently observed in teleost genome duplications, indicating some heterogeneity among events. Despite this, a stable-complex proxy for conserved dosage constraint did not detect stronger transfer, leaving post-duplication evolution with a reproducible and potentially transferable gene-lineage-associated component within a highly variable event-specific framework.