dN/dS ratio
/ dee-EN over dee-ESS /
Suppose you want to know whether an editor cares about a particular paragraph in a manuscript. You can't read the editor's mind, but you can watch the edits. If the editor freely accepts changes that alter wording but quietly rejects changes that alter the meaning, you can tell the meaning matters to them. The dN/dS ratio is exactly this trick applied to a protein-coding gene: it compares the changes that alter the protein with the changes that don't, to read what selection has been doing.
It rests on the genetic code. A mutation in a coding region is either synonymous (it changes the DNA but, thanks to the code's redundancy, spells the same amino acid — the protein is unchanged) or nonsynonymous (it changes the amino acid — the protein is altered). dS is the rate of synonymous changes per available synonymous site; dN is the rate of nonsynonymous changes per available nonsynonymous site. Because synonymous changes are nearly invisible to selection, dS estimates the neutral background rate. Comparing dN to that background reveals the pressure on the protein: dN/dS near 1 means changes to the protein are accumulating about as freely as neutral ones (no strong selection); dN/dS well below 1 means protein-altering changes are being removed (purifying selection — the usual case); dN/dS above 1 is the rarer, exciting signal of positive selection actively favoring change.
This makes dN/dS one of the workhorse tools for detecting selection from sequence alone, without any experiment. It is used to find genes under positive selection — immune genes racing against pathogens, surface proteins of viruses dodging antibodies, genes shaped during human evolution. Two honest cautions: dN/dS averaged over a whole gene can wash out a few sites under strong positive selection sitting in a sea of conserved ones (so site-specific methods are often needed), and a value near 1 is ambiguous — it can mean genuine neutrality or a mix of opposing pressures.
Flu virus surface proteins often show dN/dS above 1 at the sites antibodies grab onto: changing those amino acids helps the virus escape immunity, so positive selection drives protein-altering changes faster than silent ones.
dN/dS below 1 = purifying selection; near 1 = neutral; above 1 = positive selection.
A whole-gene dN/dS averages over every codon, so a handful of positively selected sites can hide inside a conserved gene and pull the average below 1 — finding real positive selection often needs methods that test site by site.