Synonymous codons are not used equally
Eighteen of the twenty amino acids are coded by more than one codon, and every organism uses its synonymous codons in its own proportions. Leucine has six codons; in human genes CTG codes about 40% of leucines, while in E. coli CTG is also common but CTA is rare. This preference is codon bias, and a codon usage table measures it.
The columns
A table lists every codon with the amino acid it codes and two numbers:
- Frequency per thousand: how many times the codon appears in every 1,000 codons of the sequences counted. The 64 values add up to 1,000.
- Fraction: the share of that amino acid coded by this codon. The fractions for one amino acid add up to 1. A fraction of 0.05 marks a rare codon.
Some tables also give the raw count. Tables for a species are built from thousands of its genes; the table from one gene or one plasmid only describes that sequence.
Why it matters
Rare codons match rare tRNAs. A gene full of codons that are rare in the host translates slowly, ribosomes stall, and yield falls; stretches of rare arginine codons (AGG, AGA, CGA in E. coli) are a classic cause of poor expression. The codon adaptation index, CAI, summarises how well a gene's codons match the host's preferred ones, from 0 to 1; genes above about 0.8 usually express well.
Codon optimization
Optimizing a gene for a new host means replacing rare codons with ones the host prefers while leaving the protein unchanged. Good optimization also avoids very high GC content, long repeats, unwanted restriction sites and strong RNA secondary structure at the start of the gene. Over-optimizing every codon to the single most frequent one can hurt folding, so many tools match the host's distribution instead.
Tools
The Codon Usage Calculator builds a table from any coding sequences you paste. Codon Optimization rewrites a sequence for a chosen host, and Reverse Translate turns a protein back into DNA using the most frequent codons.