Examples
Every example below is a real record in this database. Together, these examples illustrate a central principle: allele names can be renamed, retired, or no longer present in a current source release; the same name can denote different sequences in different species; and one sequence can carry several official names. The sequence defines identity, and names are source-attributed labels attached to that identity.
Names are not unique in either direction.
One sequence, two official IMGT names
IGHV1-69 and its duplicated paralogue IGHV1-69D carry the identical sequence at *01: this one piece of DNA is recorded as both IGHV1-69*01 and IGHV1-69D*01. Search either name and AIRBabel resolves to this same record. The converse also occurs: the string IGHV1-69*01 is also used for entirely different sequences in chicken, gorilla, mouse and rhesus. One sequence, many names; one name, many sequences. Only the sequence identifies the allele.
Names are unstable; the sequence is not.
A former IMGT name still finds its sequence
IMGT called this sequence IGKV1D-17*02 in the 2013-01-20 release and has called it IGKV1-17*03 from 2015-03-19 onwards, a different gene and a different allele number for the identical DNA. Searching the old name still lands here, because the release snapshots attach every former designation to the sequence's hash rather than to a name. One sequence, two official names, held one after the other.
A reference record is not fixed, even when its name is.
Same name, three different sequences
IMGT has published mouse IGHV1-2*01 over three sequence extents: 297 nt up to 2015-03-19, 294 nt up to 2022-01-17, and the 321 nt record shown here, which contains both earlier ones. The name never changed and the allele was never withdrawn; only how much sequence IMGT publishes under it did. Because identity is the hash, each extent is its own record. The chain is listed below under Earlier records, so it remains visible rather than looking like three unrelated alleles. Note also that IGHV1-2*01 is a live name in human too, on an entirely different sequence.
Different sources use different naming grammars.
The same allele under two naming conventions
IGHV1-2*09 (an IMGT designation) and IGHV1-2*04_S3434 (a source's mutation-suffix name) denote the identical sequence. AIRBabel resolves them to one record by hashing the sequence, not by parsing the names.
Nucleotide identity and amino-acid identity are different questions.
One silent substitution, identical amino-acid sequence
IGHV3-25*05 and IGHV3-25*03 differ at a single nucleotide (position 279, C to G). The substitution is silent, so their V-REGION amino-acid sequences are identical and both carry the same protein UID. Nucleotide-level and amino-acid-level identity answer different questions, so the record keeps both.
Open the record → HUMAN-VOST4CB5RNot every citation is equal.
Evidence: sequence-level literature support and OGRDB curation
IGHV1-2*02 is the germline behind VRC01-class HIV antibodies. Its references separate papers whose text or supplement contains this exact sequence (a hash match, robust across species) from papers that merely mention the name (a lead, marked with an asterisk). External records link to OGRDB's curated, evidence-backed entry for the same allele.
A citation can be used as a query, not only as a record annotation.
Start from a paper, not from a name
Ramesh et al. 2017 (Front Immunol 8:1407) assembled the rhesus macaque immunoglobulin loci de novo. Searching its PubMed ID returns AIRBabel records linked to that publication, each labelled with its evidence tier; here as a curator-verified citation carried by the MUSA rhesus set. That is the reverse of an allele page's reference list: given a paper, which AIRBabel allele records are linked to it? Archived records are returned on purpose, because a paper citing a name that IMGT has since retired is exactly the case where the citation still has to lead somewhere.
Run this search →User-supplied repertoire sequences may be partial.
Paste a fragment, not the whole allele
This is a 57 nt substring of IGHV1-18*01. With partial matching on, it resolves at 100% local identity. You do not need a full-length sequence, and you do not need to know any name.