Codon Usage Table & Rare Codon Scanner (with CAI Calculation)

Codon usage frequencies for nine commonly used organisms, plus a rare codon scanner: paste a CDS sequence, select your expression host, and immediately see which positions will slow down translation.

Organisms and sample sizes

Codon usage: E. coli W3110

Codon usage: Homo sapiens

Check the sample size before trusting the numbers

This is the most important point on this page. Codon usage tables are derived by counting codons across a set of genes — the fewer genes in that set, the less trustworthy the table. Most codon usage pages give you numbers without telling you the sample size.

The Kazusa database entry for “Escherichia coli K12” is based on only 14 CDS. E. coli is by far the most commonly used host for codon optimization, yet using a table derived from 14 genes to optimize a sequence is like inferring a country’s height distribution from three people. This page uses W3110 (4,332 CDS), the standard K-12 laboratory strain with genome-level coverage.

Every organism in the table above shows its CDS count; entries with fewer than 1,000 CDS are flagged as “small sample” — CHO cells have only 331 CDS, which matters even though CHO is a widely used protein expression host.

Why rare codons matter

Synonymous codons differ dramatically in their corresponding tRNA abundance. When a sequence contains a run of codons that are rare in the host, the ribosome stalls there, potentially reducing yield, causing premature termination, or inducing frameshifting. One of the most common causes of failed heterologous expression is inserting a eukaryotic gene rich in AGA/AGG (rare arginine codons in E. coli) directly into E. coli.

The scanner flags codons whose relative frequency is below 0.10 and shows the optimal codon for that amino acid in the selected host.

About CAI

The Codon Adaptation Index (CAI) was introduced by Sharp and Li (Nucleic Acids Res. 1987;15(3):1281–1295). It is defined as the geometric mean of the relative adaptiveness w of each codon, where w is the usage frequency of that codon divided by the frequency of the most common codon for the same amino acid family. A CAI closer to 1 indicates that the sequence codon usage more closely matches host preference.

One important distinction must be made clear: Sharp and Li’s original formulation uses a high-expression gene reference set to compute w, whereas this page uses the genome-wide average. This is a common simplification, but it introduces a systematic offset relative to CAI values computed from high-expression reference sets, so the numbers here cannot be directly compared with CAI values reported in the literature. They are appropriate for comparing different sequence variants against the same host.

Methionine, tryptophan (each encoded by a single codon, so w is always 1), and stop codons are excluded from the calculation.