Allele-Drift Watch: HLA Release Impact Monitor for Donor Registries

Concept

Stem-cell registries hold millions of donor HLA typings, each interpreted against the allele catalog as it existed on the day of typing. Every IPD-IMGT/HLA release can silently change those interpretations: an ambiguous typing may resolve into a newly named allele, a 10/10 match may shift, or a donor's search ranking may move. This product is a release-diff and impact-scoring service: it connects to the IPD-IMGT/HLA programmatic API, ingests each release, and re-scores the registry's stored typings against the updated allele set, returning a prioritized worklist covering only the donors and donor–patient pairs whose allele calls, ambiguity, or match grades actually changed.

Why now

The source paper reports 8,148 new alleles added between 2023 and July 2025 — a 23% jump to more than 43,000 alleles across 47 genes — plus a 33% reduction in submission processing time (to 22 working days), meaning the catalog will keep growing faster than manual audit cycles can absorb. The database's new APIs enable programmatic queries, and the share of alleles with full genomic sequences rose from 51.2% to 57.6%, so re-interpretation increasingly happens at genomic resolution.

What the new evidence adds

Three studies corroborate the typing landscape that makes release-diff scoring necessary rather than optional:

• PacBio SMRT sequencing now spans entire HLA class I genes, producing accurate allele calls and surfacing novel alleles in routine workflows [1] — the exact mechanism that keeps the catalog growing and that makes full-genomic-resolution records sensitive to each release, since many new alleles are defined by variation outside the classical antigen recognition site.
• NMDP itself typed 10,063 recruitment donors by full-length class I NGS on buccal-swab DNA [2] — direct evidence that the largest registries are generating high-resolution records at scale, so the population of stored typings whose meaning can shift with a release keeps growing with every recruitment cycle.
• NGS is now a proven, high-throughput alternative to Sanger (up to 380 samples per run for HLA-A, -B, and -DRB1) [3], which makes confirmatory re-typing of flagged donors affordable enough that registries can actually act on the service's output.

How the approach adapts

The evidence also sharpens how the service should segment a registry. Because clonal NGS avoids ambiguous typing results in newly typed donors [3], the 'release resolves an old ambiguity' event will increasingly concentrate in legacy records typed by SSO/SSP/Sanger methods. The service should therefore stratify the registry: weight legacy ambiguous typings highest for ambiguity-resolution events, and treat genomic-resolution records (full-gene SMRT/NGS) as the stratum where newly named alleles most often change a call outright. Each worklist item should carry a recommended action — re-interpret in software where the stored data suffices, or queue confirmatory high-resolution re-typing where it does not — since [3] shows re-typing is now a routine, batched operation rather than a bespoke lab effort.

Honest limits

• As high-resolution typing becomes the default at recruitment [1][2][3], the stock of ambiguous legacy typings will shrink over time. The ambiguity-resolution value is largest during the transition years and in registries with long-standing donor pools; the match-grade and search-ranking value persists regardless of typing method.
• The open questions from the prior iteration remain unresolved by this evidence: whether regulators require manual verification of record updates, whether the IPD-IMGT/HLA API exposes split/merge diffs natively or the service must re-process the registry, and how liability for match shifts between typing and transplant is allocated. The design assumes software re-interpretation is permissible where stored data supports it, and degrades gracefully to a re-typing queue where it is not.