The amino acid alphabet and the rules for translating 64 nucleotide triplets (codons) into them are two of the relatively few biological phenomena that are nearly universal across life. However, the discoveries of a 21st amino acid selenocysteine, a 22nd amino acid pyrrolysine, and of some non-standard genetic codes in organelle, prokaryotic and eukaryotic genomes, make the dominance of both the standard amino acid alphabet and the standard genetic code more intriguing.

I am interested in the following two questions: what make the proteinaceous amino acids special and why are there exactly 20 of them. To tackle this, I am looking into the link between amino acid properties and the structure of the standard genetic code, since previous evidence shows that the arrangement of codon/amino acid assignments in the standard genetic code efficiently minimizes the phenotypic impact of genetic error. By using different mutation and translation bias models, a series of simulations has been conducted to study the optimized amino acid indices, which maximize the error minimization effect of the standard genetic code. Each amino acid index is a vector of 20 real numbers, one for each amino acid, which represents some quantitative metric of amino acid similarity. Current results demonstrate that the high variety of the standard amino acid alphabet and the high error minimization effect of the standard genetic code could coexist when the mutation bias is high, which is very possible during early genetic code evolution. Future simulation will include a hypothesized incorporation order of amino acids in genetic codes to study the formation of diversity in the standard amino acid alphabet.