HLA–KIR Typing: Investigating Natural Killer Cell Genetics in Transplantation and Disease Research

Meta Intent: A practical guide to selecting KIR genotyping resolution, combining KIR results with HLA ligand groups, and interpreting transplantation and disease associations within reproducible evidence boundaries.

A report that says "KIR typed" may describe four very different results: gene presence or absence, gene copy number, allele-level calls, or phased multi-locus haplotypes. Those outputs are not interchangeable. A study can therefore use a technically successful assay and still produce evidence that is too coarse for its stated biological question.

HLA–KIR typing adds another layer. KIR genes are located in the leukocyte receptor complex on chromosome 19, while their principal HLA class I ligands are encoded on chromosome 6. The two systems must be genotyped separately and then joined through an explicit ligand-mapping and analysis model. This is why a combined KIR genotyping and HLA typing project should begin with the intended analytical variable, not simply with a platform name.

This guide follows a resolution-first chain:

Research question → KIR output level → assay design → ambiguity-aware calling → HLA ligand mapping → predefined interaction model → qualified evidence

Define the KIR Result Before Selecting the Assay

The first specification should state what the final KIR variable must represent. Presence/absence typing asks whether a target gene was detected. Copy-number typing asks how many genomic copies are supported. Allele-level typing distinguishes sequence variants within a gene. Haplotype resolution asks which genes and alleles occur together on the same chromosome. Each step adds information that cannot be reconstructed reliably from the preceding level alone.

Why KIR Gene Counts Appear to Disagree

Published methods and service panels do not always count the KIR family in the same way. A laboratory panel may describe 14 KIR genes plus two pseudogenes. IPD-KIR organizes the family as 15 gene loci, including two pseudogene loci, while some reviews enumerate more named genes when closely related forms such as KIR2DL5A/KIR2DL5B or KIR3DL1/KIR3DS1 are listed separately. This is a counting-convention issue, not necessarily a biological disagreement.

A defensible project therefore lists every reported target rather than relying on a headline number. The assay manifest should show whether paired or paralogous names are treated as one locus, separate targets, allele lineages, or unresolved groups. This small documentation step prevents apparent discordance when results are compared across methods or databases.

A/B Classification Is a Summary, Not a Complete Haplotype

KIR haplotypes are commonly grouped as A or B according to gene content. Group A has a relatively conserved organization and predominantly inhibitory content, whereas group B is more variable and includes different combinations of activating and inhibitory genes. Four framework genes—KIR3DL3, KIR3DP1, KIR2DL4, and KIR3DL2—usually delimit the variable centromeric and telomeric regions.

However, "A/A," "A/B," or "B/B" does not report exact copy number, allele identity, centromeric and telomeric motif phase, or unusual structural configurations. It is useful as a broad variable only when that is the prespecified level of analysis.

Figure 1: KIR Typing Resolution Layers. Gene presence, copy number, allele identity, and phased haplotype are nested but distinct outputs. A/B classification compresses gene-content information and should not be mistaken for full structural or allele-level resolution.

Why the KIR Locus Defeats Generic Genotyping Assumptions

The KIR region combines three difficult properties: high sequence similarity among genes, extensive gene-content variation, and copy-number variation. Reads from homologous exons may align to several loci. A deletion can resemble failed amplification. A duplicated segment can distort allele balance. Recombinant or hybrid genes may match different references in different regions. These conditions violate the simple diploid assumption behind many standard variant-calling workflows.

Even framework genes are not an excuse to assume two copies in every sample. Large-cohort studies have documented deletions, duplications, uncommon gene-content haplotypes, and novel alleles. A project interested in structural diversity may therefore need dedicated CNV sequencing services or another quantitative layer rather than a binary positive/negative call.

Genome-wide data can provide supporting evidence, but a nominal whole genome sequencing data set is not automatically KIR-resolved. Read length, insert size, depth, reference representation, multi-mapping policy, and KIR-aware software determine what can be recovered. Whole-exome data are especially vulnerable because capture targets and paralogous regions may be incomplete or uneven.

A second distinction is equally important: genotype is not receptor expression. A KIR gene may be present but not expressed on every NK cell. Alleles can also differ in surface expression or function. HLA–KIR genotyping defines inherited potential; it does not directly measure the cellular repertoire, receptor abundance, education state, or target-cell response.

Figure 2: Structural Sources of KIR Ambiguity. High homology, variable gene content, segmental duplication, deletion, and recombination can generate identical-looking local signals from different underlying haplotypes.

Choose the Minimum Sufficient Resolution for the Research Question

More resolution is valuable only when it changes the study variable or prevents a known ambiguity. The practical aim is the minimum sufficient resolution: enough evidence to support the intended comparison, plus an escalation rule for samples that cannot be resolved at that level.

For example, a cohort study of common gene-content frequencies may not benefit from full-locus phasing. Conversely, a project testing whether a specific KIR allele behaves differently in the presence of a particular HLA allotype cannot collapse results to A/B groups. When the HLA component itself requires full-gene or phased resolution, the companion guide on high-resolution HLA typing by NGS defines that separate requirement.

Compare KIR Genotyping Methods by Deliverable, Not by Platform

A method should be selected according to the output it has been validated to produce. The phrase "NGS-based KIR typing" is incomplete unless it is accompanied by target regions, copy-number logic, allele-calling scope, phasing capability, quality gates, and an ambiguity policy.

PCR-SSP and PCR-SSO: Efficient Gene-Content Screens

Sequence-specific priming and oligonucleotide-probe approaches can efficiently determine whether expected KIR targets are present. They remain useful when the endpoint is a validated gene-content pattern. Their limitation is conceptual as well as technical: a positive reaction does not identify every allele, quantify copy number, or reveal how genes are phased across two haplotypes. Allelic variation in primer or probe sites can also create apparent absence if the assay was not designed for the relevant diversity.

qPCR, dPCR, and MLPA: Quantitative Structural Evidence

Quantitative PCR and digital PCR can estimate locus copy number when target and reference assays are carefully calibrated. An MLPA assay can provide an orthogonal targeted view of dosage across selected regions. These approaches are particularly useful when a binary assay produces an unexpected motif or when sequencing read depth suggests a duplication or deletion.

Copy number still does not identify the copied allele or place extra copies on a specific chromosome. A value of three, for example, is compatible with more than one haplotype arrangement. Family segregation, long-range information, or another phasing strategy may be needed when chromosome-specific structure is the endpoint.

Targeted Short-Read NGS: Joint Content, Copy Number, and Allele Evidence

A purpose-built targeted region sequencing design can concentrate coverage on informative KIR regions. Amplicon sequencing may support high-throughput analysis, while multiplex PCR sequencing can organize many gene-specific targets into a limited number of reactions.

The resulting resolution depends on what was amplified. Exon-focused panels may distinguish many alleles but leave phase unresolved across distant variants. Read-depth-based copy-number estimates require stable normalization, reference samples spanning relevant copy states, and explicit behavior for ambiguous clusters. Novel variation can reduce assignment confidence when the caller is restricted to known references.

Long-Range and Long-Read Strategies: Better Connectivity, Not Automatic Certainty

Long amplicon analysis can connect variants across a larger portion of a KIR gene. Targeted long-read strategies, including Nanopore target sequencing, can improve whole-gene continuity and help investigate complex structures. The trend toward graph-based and haplotype-aware analysis reflects a real need: linear references are often insufficient for highly polymorphic immune loci.

Long reads do not remove the need for quality control. Amplicon dropout, chimeric molecules, uneven coverage, homopolymer or context-specific errors, and under-supported structural paths can still generate false certainty. A long-read result should state molecule support, strand balance where relevant, target completeness, and whether phase is observed directly or inferred computationally.

Orthogonal Confirmation Should Be Triggered by Risk

Confirmation is most useful for novel alleles, unexpected copy number, discordant replicate calls, unusual framework-gene patterns, and variants that drive the primary analysis. Local sequence questions may be checked with Sanger sequencing; dosage questions require a quantitative method; full structural questions require evidence that spans the relevant junction. Repeating the same assay does not resolve a method-specific blind spot.

Figure 3: Method-to-Output Map for KIR Genotyping. Assays overlap, but their validated deliverables differ. The method should be chosen from the required output backward, with escalation rules for unresolved copy number, allele, or phase.

The Control Set Defines the Resolution You Can Defend

KIR assay validation should challenge the exact output the project intends to report. A presence/absence panel needs positive and negative controls for variable genes and evidence that common primer-site variation does not create false absence. A copy-number workflow needs reference samples representing more than the common two-copy state. An allele-level workflow needs samples with known heterozygosity, homologous gene combinations, and alleles that differ within the actually covered regions. A phasing workflow needs truth material in which phase was established independently.

Controls should also challenge the sample process. Genomic DNA from blood, saliva, buccal material, cell pellets, or archived specimens can differ in integrity, inhibitors, human DNA fraction, and compatibility with long-range amplification. The acceptance criteria should therefore be tied to the selected assay architecture. High total DNA concentration cannot compensate for fragmentation when the method depends on intact long amplicons.

Include replicate strategy where it resolves a known source of variation. Technical library replicates can reveal unstable amplification or borderline copy-number clustering, but they do not replace independent biological samples. Re-extraction can test matrix-related failure. An orthogonal method can test a different measurement principle. These three forms of repetition answer different questions and should not be grouped under a single "replicated" label.

A useful validation panel contains common genotypes, rare or structurally unusual configurations, zero-to-multiple-copy states where available, and no-template or contamination controls. The panel should be versioned along with the assay because an apparently stable workflow can become under-specified as new KIR alleles are added to the reference database.

Before cohort processing begins, run the complete laboratory and analysis path on the validation panel and define which discrepancies trigger repeat testing, orthogonal confirmation, manual review, or a deliberately unresolved report.

Use a KIR-Aware Analysis Pipeline That Preserves Uncertainty

A generic pipeline that aligns reads to one reference and calls diploid variants can produce plausible-looking but biologically invalid KIR results. A KIR-aware workflow should resolve the problem in stages.

1. Confirm sample identity and DNA suitability. Track concentration, integrity, contamination indicators, plate position, and any donor–recipient pairing identifiers.
2. Assess target-level coverage. Evaluate covered regions, uniformity, strand or molecule support, and dropout patterns before assigning absence.
3. Estimate gene content and copy number. Separate zero-copy evidence from low coverage, then normalize depth or quantitative signals against validated controls.
4. Perform copy-number-aware allele inference. The number of expected allele copies should constrain genotype decomposition. Multi-mapping and shared sequence must be modeled rather than silently discarded.
5. Retain novel and ambiguous states. Report compatible allele sets, unresolved gene groups, and candidate novel variants instead of forcing the closest database match.
6. Infer phase only when supported. Read-backed phase, family segregation, population inference, and database matching are different evidence types and should be labeled separately.
7. Apply orthogonal confirmation rules. Trigger confirmation according to analytical risk and the role of the call in the primary hypothesis.

Reference context is part of the result. IPD-KIR releases change as new alleles and haplotypes are curated. A call generated against one release may be renamed, split, or resolved differently later. The pipeline version, reference release, target definition, and parameter set should therefore travel with the genotype table.

Figure 4: KIR-Aware Calling and Ambiguity Gates. Copy number constrains allele inference, while unresolved or novel states remain visible. A forced single call is not a quality improvement.

Build a Report That Can Be Reinterpreted

A reusable KIR report should contain more than a final genotype label. At minimum, each locus-level row should include:

• reported target and nomenclature convention;
• gene presence status and supporting coverage;
• estimated copy number with confidence or review flag;
• allele call or compatible allele set;
• phase status and the evidence used to establish it;
• covered and uncovered regions;
• novel-variant or recombinant-structure flag;
• reference database and release;
• software, workflow, and parameter version;
• orthogonal confirmation status.

A machine-readable table should be accompanied by a concise interpretation note explaining what the assay did not resolve. This is especially important when results are shared between population studies or reanalyzed after a database update. "No call" and "not assayed" should never be converted to "absent."

Convert HLA Alleles to Ligand Groups as a Separate Step

KIR and HLA results should remain separate until both have passed their own quality checks. The HLA layer is then transformed into ligand categories required by the prespecified model. Common abstractions include C1/C2 based mainly on HLA-C and Bw4/Bw6 based mainly on HLA-B. Cross-locus exceptions matter: selected HLA-B allotypes can contribute a C1 epitope, while selected HLA-A allotypes can carry Bw4. More conditional, peptide-dependent relationships—such as HLA-A3/A11 with KIR3DL2—should not be merged into the same rule table without an explicit evidence label.

This transformation is not a lookup that should remain hidden inside an analysis script. The project should record:

• which HLA loci and allele fields were used;
• the ligand-definition source and version;
• how ambiguous HLA calls were handled;
• whether cross-locus C1/Bw4 assignments or conditional HLA-A relationships were included;
• which KIR–ligand relationships were considered established, conditional, or unknown;
• whether the analysis used ligand presence, ligand dose, or allele-specific interaction terms.

Three boundaries must remain visible. First, KIR gene presence does not prove receptor surface expression. Second, an HLA ligand genotype does not show expression on the relevant target cell. Third, a compatible receptor–ligand pair does not by itself demonstrate NK-cell education, inhibition, activation, or cytotoxicity. The matrix overview, HLA Typing and Immune Repertoire Sequencing Services, places this genetic layer in the broader immune-profiling context.

Figure 5: From HLA Alleles to Ligand-Aware Variables. Ligand mapping is a documented transformation between two validated genotype layers, not proof that a functional receptor–ligand interaction occurred.

Predefine the Donor–Recipient Model Before Testing Transplant Outcomes

"KIR mismatch" is not one variable. Transplantation studies have used ligand–ligand mismatch, receptor–ligand mismatch, missing-ligand or missing-self models, education models, A/B haplotype categories, centromeric and telomeric motifs, B-content scores, and allele-specific KIR–HLA pairs. These models can classify the same donor–recipient pair differently.

The direction of the data must be explicit. A donor-KIR/recipient-HLA model is not interchangeable with recipient-KIR/donor-HLA or donor-HLA/recipient-HLA matching. The research record should identify the donor and recipient variable in every interaction term.

Context also changes the biological interpretation. Disease, graft source, HLA matching level, conditioning intensity, T-cell depletion, post-transplant cyclophosphamide, serostatus, donor age, and endpoint definition can all influence an observed association. A result from one transplant platform should not be presented as a universal property of a KIR gene or B haplotype.

Evidence remains model- and cohort-dependent. A 2024 validation study of 5,017 unrelated-donor transplants did not confirm several previously proposed donor KIR classifications. A 2026 meta-analysis reported pooled associations between donor B/X genotypes and some outcomes, but its evidence base largely used low-resolution AA/B/X typing, showed heterogeneity, and could not resolve refined centromeric/telomeric subtypes or donor–recipient KIR combinations. These findings are not inherently contradictory: a pooled association and a prospectively useful selection rule are different evidentiary standards.

HLA matching remains primary. No generally applicable KIR-genotype algorithm currently replaces established donor-selection criteria. KIR variables are better framed as research stratifiers or candidate auxiliary variables whose value must be tested in the relevant transplant setting and against independently defined endpoints.

For projects focused specifically on unit identity, confirmatory typing, and HLA matching logic, see HLA Typing for Cord Blood Banking and Transplantation. The present article addresses only the additional KIR and ligand-model layer.

Figure 6: KIR–HLA Models Are Not Interchangeable. Each model consumes a different subset and direction of donor and recipient information. The model definition must be fixed before outcome testing.

Design Disease-Association Studies for Replication, Not Just Discovery

KIR–HLA combinations have been investigated in viral infection, autoimmune and inflammatory phenotypes, cancer research, and reproductive immunology. These fields are biologically important, but they are also vulnerable to population structure, underpowered interaction tests, inconsistent phenotype definitions, and selective reporting.

KIR and HLA frequencies vary across populations and show evidence of coevolution. A disease-association study should therefore model ancestry and recruitment structure rather than treating a pooled cohort as genetically homogeneous. A population genetics framework can help separate the primary biological comparison from background frequency differences and linkage patterns.

The analytical hierarchy should also be prespecified. Gene content, copy number, allele, motif, phased haplotype, HLA ligand, and KIR–HLA interaction terms are different hypothesis families. Testing all of them without a multiplicity plan produces unstable "best" associations. Useful safeguards include:

• a primary resolution and primary interaction model;
• minimum counts for rare genotype combinations;
• ancestry-aware covariates or stratified analysis;
• correction for the tested hypothesis family;
• sensitivity analysis across plausible ambiguity resolutions;
• an independent replication cohort;
• functional follow-up for associations that imply receptor behavior.

Most importantly, a cohort-level association is not an individual diagnostic result. KIR–HLA genotypes alone should not be translated into personal infection risk, autoimmune prognosis, fertility advice, or treatment decisions. The defensible claim is limited to the sampled population, measured phenotype, selected model, and validated analytical resolution.

Use a One-Page HLA–KIR Project Specification

Before sample processing, the study team should be able to complete the following specification without using "high resolution" as an undefined shortcut:

1. Primary question: What biological comparison or association will be tested?
2. Unit of inference: Individual, donor, recipient, donor–recipient pair, family, or population?
3. KIR output: Presence, copy number, allele, motif, or phased haplotype?
4. HLA output: Which loci, which allele fields, and which phase requirements?
5. Ligand mapping: Which C1/C2, Bw4/Bw6, HLA-A, or allele-specific rules?
6. Interaction model: Which direction and which formal definition?
7. Assay architecture: Targets, covered regions, expected blind spots, and controls?
8. Ambiguity policy: What remains unresolved, what triggers review, and what is excluded?
9. Confirmation policy: Which calls require an independent method?
10. Analysis plan: Covariates, multiplicity control, sensitivity analysis, and replication?
11. Deliverables: Genotype table, QC evidence, ligand derivation, model output, and limitations?

When a project includes many loci or custom immune-genetic targets, a gene panel sequencing service can be evaluated against this specification. The panel should be accepted only if its target coverage and analysis model match the required KIR output—not because it carries a broad "immune" label.

Figure 7: HLA–KIR Project Specification. A decision-ready project connects the biological question to an explicit genotype resolution, ligand transformation, interaction model, uncertainty policy, and final deliverable.

What a Decision-Ready Deliverable Should Contain

A final project package should allow another analyst to reproduce both the genotype interpretation and the derived KIR–HLA variables. Recommended components include the sample manifest; assay target manifest; per-locus coverage and copy-number evidence; genotype and compatible-call sets; phase evidence; HLA allele calls; ligand-group derivation table; model definitions; versioned software and reference databases; excluded or unresolved samples; orthogonal confirmation results; and a limitations statement.

CD Genomics can support research teams in defining the required KIR and HLA resolution, selecting targeted or long-range sequencing strategies, establishing quantitative and orthogonal confirmation routes, and organizing results for downstream population or donor–recipient research. Project support should be selected according to the study question, sample architecture, and required evidence level.

Conclusion

HLA–KIR typing is most useful when it produces a precisely defined analytical variable. Gene content, copy number, allele identity, and phased haplotype answer different questions. HLA ligand groups add a second inference layer, and transplantation or disease models add a third. Each layer needs its own quality checks and uncertainty boundaries.

The strongest project is not the one with the most methods. It is the one in which the required resolution is specified before testing, the assay is validated for that output, ambiguity is retained, the KIR–HLA model is explicit, and the conclusion does not travel farther than the evidence.

Frequently Asked Questions

Research Use Only. Not for use in diagnostic procedures.

References:

1. IPD-KIR Database. Introduction to KIR gene organization, nomenclature, and haplotypes.
2. Downing J, D'Orsogna L. High-resolution human KIR genotyping. Immunogenetics. 2022;74:369–379.
3. Wagner I, et al. Allele-Level KIR Genotyping of More Than a Million Samples: Workflow, Algorithm, and Observations. Frontiers in Immunology. 2018;9:2843.
4. Jiang W, et al. qKAT: a high-throughput qPCR method for KIR gene copy number and haplotype determination. Genome Medicine. 2016;8:99.
5. Amorim LM, et al. High-Resolution Characterization of KIR Genes in a Large North American Cohort Reveals Novel Details of Structural and Sequence Diversity. Frontiers in Immunology. 2021;12:674778.
6. Marin WM, et al. High-throughput Interpretation of KIR Short-read Sequencing Data with PING. PLOS Computational Biology. 2021;17:e1008904.
7. Dhuyser A, et al. KIR in Allogeneic Hematopoietic Stem Cell Transplantation: Need for a Unified Paradigm for Donor Selection. Frontiers in Immunology. 2022;13:821533.
8. Dhuyser A, et al. Comparison of NK alloreactivity prediction models based on KIR-MHC interactions in haematopoietic stem cell transplantation. Frontiers in Immunology. 2023;14:1028162.
9. Schetelig J, et al. Donor KIR genotype based outcome prediction after allogeneic stem cell transplantation: no land in sight. Frontiers in Immunology. 2024;15:1350470.
10. Huang Y, et al. Effects of donor killer-cell immunoglobulin-like receptor genotypes on clinical outcome after allogeneic hematopoietic stem cell transplantation—a systematic review and meta-analysis. Frontiers in Immunology. 2026;17:1856878.