Did evolution select a nonrandom "alphabet" of amino acids?

The last universal common ancestor of contemporary biology (LUCA) used a precise set of 20 amino acids as a standard alphabet with which to build genetically encoded protein polymers. Considerable evidence indicates that some of these amino acids were present through nonbiological syntheses prior to the origin of life, while the rest evolved as inventions of early metabolism. However, the same evidence indicates that many alternatives were also available, which highlights the question: what factors led biological evolution on our planet to define its standard alphabet? One possibility is that natural selection favored a set of amino acids that exhibits clear, nonrandom properties-a set of especially useful building blocks. However, previous analysis that tested whether the standard alphabet comprises amino acids with unusually high variance in size, charge, and hydrophobicity (properties that govern what protein structures and functions can be constructed) failed to clearly distinguish evolution's choice from a sample of randomly chosen alternatives. Here, we demonstrate unambiguous support for a refined hypothesis: that an optimal set of amino acids would spread evenly across a broad range of values for each fundamental property. Specifically, we show that the standard set of 20 amino acids represents the possible spectra of size, charge, and hydrophobicity more broadly and more evenly than can be explained by chance alone.     

Testing the potential for computational chemistry to quantify biophysical properties of the non-proteinaceous amino acids

Although most proteins of most living organisms are constructed from the same set of 20 amino acids, all indications are that this standard alphabet represents a mere subset of what was available to life during early evolution. However, we currently lack an appropriate quantitative framework with which to test the qualitative hypotheses that have been offered to date as explanations for nature's "choices." Specifically, although many indices have been developed to describe the 20 standard amino acids, few or no comparable data extend to prebiotically plausible alternatives because of the costly and time-consuming bench experiments that would be required. Computational chemistry (specifically quantitative structure property relationship methods) offers a potentially fast, cost-effective remedy for this knowledge gap by predicting such molecular properties in silico. Thus, we investigated the use of various freely accessible programs to predict three key amino acid properties (hydrophobicity, charge, and size). We assessed the accuracy of these predictions by comparisons with experimentally determined counterparts for appropriate test data sets. In light of these results, and factors of software accessibility and transparency, we suggest a method for further computational assessments of prebiotically plausible amino acids. The results serve as a starting point for future quantitative analysis of amino acid alphabet evolution.