Guides › Nomenclature

HLA ambiguity explained: allele, genotype and phase ambiguity

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

An HLA typing result is ambiguous when more than one answer is consistent with the data. There are three distinct kinds: allele ambiguity (which allele is on one chromosome), genotype ambiguity (which pair of alleles the person carries) and phase ambiguity (which alleles at different loci sit on the same chromosome).

They come from different limitations of the typing, they are written differently, and they are not interchangeable. A system that stores all three as "a list of possible alleles" loses information that someone will need later.

The three kinds at a glance

The delimiters come from the GL String grammar published by Milius and colleagues in 2013. The GL String guide covers the full syntax and precedence.

Allele ambiguity

Allele ambiguity means one copy of the gene is known to be one of a list of alleles, and the typing cannot say which. It arises when the alleles are identical over everything the method examined.

The common case is typing that covers the exons encoding the peptide-binding domains. The 2013 review of NGS typing from the 16th International Histocompatibility and Immunogenetics Workshop (De Santis and colleagues) says "only the most polymorphic regions of HLA class I (exon 2 and 3) and II (exon 2) encoding the peptide binding sites are routinely assessed". Alleles that differ only elsewhere cannot be told apart. A 2024 review of typing techniques (Geo and colleagues) puts it the same way for Sanger sequencing: ambiguities occur "when two or more alleles share the same sequences in the targeted exons but differ in the untargeted exons".

In GL String, allele ambiguity is a /-separated list at one position, with every name written in full:

HLA-A*02:01:01:01/HLA-A*02:01:01:02L

That example is the shape the GL String paper recommends: full names, the HLA- prefix on each, never an abbreviated 02:01:01:02L. Note that this particular list includes an L allele; an allele list can mix expression states, which is why it must not be collapsed to a single name.

G and P codes summarise allele ambiguity

The 2010 nomenclature update introduced G and P codes for exactly this situation. A G group (for example A*02:01:01G) is the set of alleles identical in nucleotide sequence across exons 2 and 3 for class I or exon 2 for class II; a P group (for example A*02:01P) is the set encoding the same protein in those exons. A G or P code is a compact way to write a particular, release-defined allele list.

They do not remove the ambiguity. In release 3.66.0, A*02:01:01G has 554 members and A*02:01P has 697, and both are recalculated every release. The G and P groups guide explains the differences, including that G groups can contain null alleles.

Genotype ambiguity

Genotype ambiguity means the typing is consistent with more than one pair of alleles. It is a heterozygous problem: when a person carries two different alleles, a method that reads both copies at once sees a mixture at every position where they differ, and may not be able to say which variant belongs to which copy.

The WMDA README in the IPD-IMGT/HLA repository describes it for sequence-based typing: "Due to the heterozygous nature of the SBT analysis the combinations of many pairs of alleles may give an ambiguous typing result." De Santis and colleagues attribute ambiguous combinations to "cis-trans ambiguities or ... particular allele combinations being identical over the regions commonly analysed". Geo and colleagues note that sequence-specific oligonucleotide typing "does not distinguish between cis and trans polymorphisms" either.

In GL String, + joins the two copies of one genotype, and | separates the alternative genotypes. A string with this shape (the names are illustrative, not a real ambiguity) says "either this pair or that pair":

HLA-A*01:01+HLA-A*24:02|HLA-A*01:02+HLA-A*24:99

Each side of the | is a complete genotype. Splitting it into two allele lists, A*01:01/A*01:02 and A*24:02/A*24:99, would wrongly allow the combination A*01:01 with A*24:99. That is the information a flat list loses.

G and P codes cannot express genotype ambiguity. They describe one copy at a time.

Phase ambiguity

Phase ambiguity is about which alleles at different loci are on the same chromosome, that is, the haplotypes. Typing that resolves each locus separately gives two alleles per locus but does not say which A allele travels with which B allele.

GL String writes known phase with ~ and unknown phase with +. A fully unphased two-locus result is two genotypes joined by ^:

HLA-A*01:01:01:01+HLA-A*24:02:01:01^HLA-B*08:01:01:01+HLA-B*44:02:01:01

The paper defines + as separating alleles "that are detected, but not identified as being on the same chromosome", and says the order of elements in a GL String carries no additional meaning. Writing the A alleles first and the B alleles in the same order does not imply that A*01:01:01:01 and B*08:01:01:01 are on one chromosome. Only ~ says that.

Genotype ambiguity within a locus and phase ambiguity across loci share a root: the method cannot tie two observed variants to the same DNA strand.

What sequencing technology changes

The workshop review summarises what next-generation sequencing adds: "the combination of clonal amplification, which provides phase information, and the ability to sequence larger regions of genes, including introns". Clonal reads attack genotype ambiguity, because each read comes from one copy; longer regions attack allele ambiguity, because more of the gene is seen.

Neither removes ambiguity entirely. Geo and colleagues note that with short reads, "the inability to phase distant polymorphic regions leads to ambiguity". A result from any method should state what it could not resolve, and the SSO, SSP, SBT and NGS comparison covers where each method leaves it.

Keeping ambiguity intact in data

• Store the structure, not just the names. A GL String keeps allele lists, genotype alternatives and phase distinct.
• Store the release. The GL String paper says the IMGT/HLA release is not part of the string and must be conveyed separately. Allele lists and G groups change with each release.
• Do not resolve by picking the first or most common name. Order inside a GL String carries no meaning.
• Validate before storing. POST /v1/glstring checks the grammar, flags problems such as mixed loci in one allele list, and rewrites outdated names to current ones.

Frequently asked questions

What is the difference between allele and genotype ambiguity?

Allele ambiguity leaves open which allele one copy is. Genotype ambiguity leaves open which pair of alleles is present, so the alternatives must be kept as whole pairs, written with | in GL String.

Do G groups remove HLA typing ambiguity?

No. A G group is a release-defined name for a set of alleles identical over the peptide-binding exons. It summarises allele ambiguity but does not resolve it, and it cannot express genotype ambiguity.

What is phase ambiguity in HLA typing?

Not knowing which alleles at different loci lie on the same chromosome. GL String writes unknown phase with + and known phase with ~.

Does NGS eliminate HLA ambiguity?

It reduces it, through clonal reads that carry phase and through sequencing more of each gene. Published reviews note that short reads can still leave distant positions unphased, so results should state any remaining ambiguity.