Guides › Nomenclature

HLA null alleles: what the N suffix means, causes and counts

Updated 2026-10-30 · allele examples checked against IPD-IMGT/HLA 3.66.0 on 2026-10-30

An HLA null allele is an allele that has been shown or inferred not to be expressed as a protein, and its name ends in N, as in HLA-A*24:09N. The gene is present and can be typed, but there is no HLA molecule on the cell surface from that copy.

That matters wherever a typing result is turned into a statement about protein: matching, antibody interpretation and serological equivalents. A null allele read as its expressed relative is a silent error. In IPD-IMGT/HLA release 3.66.0 there are 2,072 named null alleles.

What the N suffix says

The nomenclature at hla.alleles.org uses optional suffixes for expression status. N marks a null allele. The others are L (low cell-surface expression), S (secreted, not on the cell surface), C (cytoplasmic), A (aberrant) and Q (questionable). As of September 2026 no alleles have been named with C or A.

The suffix is part of the name, not an annotation beside it. A*24:09N is a different allele from A*24:02, and there is no expressed A*24:09 for it to collapse into. Any process that strips trailing letters to "clean" names will turn it into a name that means something else.

Why alleles are null

The suffix assignment page says the most common cause is a premature stop codon in the domains encoding the peptide-binding groove or the Ig-like domain. That stop can come from a point mutation or from an insertion or deletion that shifts the reading frame.

Two further rules from the same page explain names that look surprising:

• Splice-site changes start as Q. A mutation in the first two or last two bases of an intron is given Q by default, and becomes N only if expression work shows the allele is not expressed.
• The field structure follows the mutation. The page's example is A*01:01:01:02N, named as an intronic variant because the underlying mutation is intronic. A null allele can therefore carry a four-field name, which elsewhere marks a difference outside the coding sequence.

The page also says suffix assignments follow established principles from expression studies and may be revised as new data arrive. A name with N today records the evidence today.

How many null alleles there are

Counted from the 3.66.0 Allelelist.txt, 2,072 current names end in N. By locus:

For comparison, the same file has 460 Q, 7 L and 2 S names. Counts change every quarter, so quote them with the release.

To list the null alleles yourself, take Allelelist.txt for the release you need and select the names ending in N. The WMDA file rel_dna_ser.txt gives every one of them the serology value 0, which its README defines as the marker for a null allele.

Where null alleles hide

A null allele usually looks like an ordinary member of its group until you read the name in full.

Inside G groups. A G group is the set of alleles identical across the exons encoding the peptide-binding domains, exons 2 and 3 for class I. An allele whose defect lies outside those exons can share a G group with expressed alleles. In release 3.66.0, 534 null alleles share a named G group with other alleles, and 162 G groups contain at least one. Examples: A*24:09N is in A*24:02:01G, A*11:21N in A*11:01:01G, B*51:11N in B*51:01:01G, B*15:01:01:02N in B*15:01:01G, and DRB4*01:03:01:02N in DRB4*01:01:01G.

Not inside P groups. P groups are defined by identical protein in those domains. In release 3.66.0 no P group contains a null allele, and A*24:09N has no P group at all. The G and P groups guide explains the difference.

Behind low-resolution results. A first-field result such as A*24 covers both A*24:02 and A*24:09N. A typing method can also miss a null-causing change that lies outside the region it examines: a 2024 review of typing methods notes that focusing on the antigen recognition site hinders "detection of null alleles carrying mutations outside the ARS".

Standing alone. Most null alleles are in no named G group at all: 1,538 of the 2,072 in release 3.66.0. B*08:08N is one.

Why the N matters in matching

At allele level, a match compares two-field names. A*24:09N and A*24:02 are different names, so they are a mismatch, even though a serological or first-field view would call both "A24".

Checked with the HLA-Verify match engine at release 3.66.0: a recipient typed A*02:01, A*24:02 and a donor typed A*02:01, A*24:09N, identical at B, C and DRB1, scores 7/8 with the A locus mismatched and a null_allele flag. Reading the null allele as A24 would report 8/8. The 8/8 and 10/10 guide walks through the arithmetic. This is a statement about the count, not about donor selection.

Null status can change

Because suffixes follow evidence, a null allele can be renamed. Two examples from Deleted_alleles.txt in release 3.66.0:

• C*04:09N: "Suffix changed becoming C*04:09L (December 2024)". The allele is now C*04:09L, low expression rather than null.
• A*01:34N: "Sequence shown to be expressed at low levels and renamed A*01:01:38L (March 2011)".

Stored typings that contain either old name need to be read through the name history, not compared as strings. The expression suffixes guide covers the other suffixes and their changes.

Handling null alleles in data

• Keep the suffix in every stored and exchanged name.
• When reducing to two fields, keep N whenever every allele under the name carries it, so A*24:09N stays A*24:09N.
• Flag null alleles on import, so they are visible to whoever reads the typing next.
• Re-check stored null names on each release; suffix changes are recorded in Deleted_alleles.txt.

GET /v1/allele/{name} reports null status, G group and P group for a name in the pinned release, and /v1/verify flags null_allele on every null token it finds in free text.

Frequently asked questions

What does N mean at the end of an HLA allele name?

It marks a null allele, one shown or inferred not to be expressed. The suffix is part of the name and must not be removed.

What is the most common cause of an HLA null allele?

According to the nomenclature committee's suffix page, a premature stop codon in the domains encoding the peptide-binding groove or Ig-like domain, from a point mutation or a frameshifting insertion or deletion.

How many HLA null alleles are there?

IPD-IMGT/HLA release 3.66.0 names 2,072 alleles ending in N, most at HLA-A, -C and -B. The number changes every quarterly release.

Can a G group contain a null allele?

Yes. A G group only requires identity across the peptide-binding exons, so A*24:09N is in A*24:02:01G. No P group contains a null allele in release 3.66.0.