Nature 433, 633–638 (2005)

We reported a universal trend of amino-acid gain and loss observed for recent evolutionary history among a diverse set of 15 taxa, with amino acids of declining frequencies being the first to be incorporated into the genetic code and those of increasing frequencies being late recruits. We have since discovered that a similar scenario for protein evolution was proposed by Zuckerkandl and colleagues more than thirty years ago1. Their analysis of a far smaller vertebrate-specific data set of two protein families also revealed asymmetric patterns of amino-acid substitution, and they went on to speculate that “extrapolation to zero occurrence of the rare amino acids might define the time at which they were introduced into the genetic code.”