Very Low Power to Detect Asymmetric Divergence of Duplicated Genes

Abstract

Asymmetric functional divergence of paralogues is a key aspect of the traditional model of evolution following duplication. If one gene continues to perform the ancestral function while the other copy evolves a new function then we might expect a period of accelerated sequence evolution following duplication in one of the copies. In keeping with this prediction, many individual examples of asymmetric divergence at the level of protein function have been observed that are accompanied by asymmetric divergence at the sequence level. While several large-scale studies suggest that asymmetric divergence is common across a range of different organisms the degree to which they can be considered to provide an accurate estimate of its prevalence and therefore of the importance of this mode of divergence depends on both the accuracy and power of the methods that have been used. We investigated two methods that can be used to detect asymmetric duplicates using simulated data and real data from Arabidopsis thaliana. One of the methods detects departure from a local molecular clock for amino acid sequences and has been used previously. The second method is novel and tests for different selective constraints along the duplicated lineages using codon models of evolution. This approach is less prone to false positive results but has lower power than the molecular clock method. We find that the power to detect asymmetric divergence is low with both methods unless the effect is strong and report a surprising lack of strong evidence for asymmetric divergence in paralogues derived from the most recent round of genome duplication in Arabidopsis.