HLA & immunogenetics

KIR genes

T1K

KIR (killer-cell immunoglobulin-like receptors) sit on natural killer cells and probe the HLA class I molecules of other cells. The KIR gene complex on chromosome 19 is highly variable: people differ in the number and type of their KIR genes. Genome genotypes KIR from sequence data.

Key points

  • Inhibitory KIR keep NK cells in check as long as they recognise self HLA class I.
  • If the HLA class I signal is lost (missing self), for example in tumour or viral infection, the NK cell attacks.
  • The KIR repertoire is highly variable and influences infection defence, pregnancy and transplant outcome.
KIR RECEPTOR · NK CELL HLA LIGAND · TARGET CELL KIR2DL1 inhibitory HLA-C group 2 C2 · lysine 80 KIR2DL2 / 2DL3 inhibitory HLA-C group 1 C1 · asparagine 80 KIR3DL1 inhibitory HLA-B Bw4 motif KIR3DL2 inhibitory HLA-A A*03 / A*11 If self HLA-I is missing (missing self), inhibition is lost and the NK cell is activated.

A variable gene complex

The KIR locus on chromosome 19 spans up to 14 genes that are not all present in every person. Two haplotype groups are distinguished: A haplotypes carry a largely fixed, mostly inhibitory gene set, while B haplotypes additionally carry more activating KIR and vary more in content and arrangement. This diversity arises from gene duplication and deletion together with high allelic polymorphism and makes KIR as demanding as HLA. Every person inherits one KIR haplotype from each parent, so the individual repertoire varies widely.

How Genome genotypes

Genome uses T1K (Song et al., 2023), which determines KIR and HLA alleles directly from short reads by matching the reads against the IPD-KIR and the IPD-IMGT/HLA reference databases. T1K first gathers candidate reads through short sequence matches (seeds) and assigns them to the alleles with the most matching nucleotides. A weighted expectation-maximization algorithm then estimates allele abundances for all genes at once, so reads that fit several highly similar KIR genes are handled jointly rather than in isolation. Per locus T1K selects the allele pair that explains the most reads, filters weak candidates below a fraction of the dominant allele and assigns a quality score to each call. Genome presents these calls as independent technical evidence, separate from any clinical interpretation.

Why KIR is hard from short reads

KIR genotyping cannot be solved with the usual variant pipelines that assume a single reference genome. First, KIR genes are very similar to each other, so short reads often fit several genes equally well, for example the closely related KIR2DL5A and KIR2DL5B. Second, individual KIR genes can be entirely absent from a chromosome, so the tool must infer presence or absence from read coverage rather than only distinguishing alleles. Third, allelic diversity is high and grows with every IPD-KIR update. T1K addresses this by modelling all genes jointly: low estimated abundance points to a missing gene, and the simultaneous estimation separates homologous genes that could not be resolved on their own.

What Genome measures. Which KIR genes are present and in which alleles, each with a per-call quality score. The complex is shown as independent technical evidence alongside HLA typing.

Related topics

Sources

  1. 1Song et al., 2023 Efficient and accurate KIR and HLA genotyping with massively parallel sequencing data. Genome Research 33:923-931. doi.org/10.1101/gr.277585.122
  2. 2Parham & Moffett, 2013 Variable NK cell receptors and their MHC class I ligands in immunity, reproduction and human evolution. Nature Reviews Immunology 13:133-144. doi.org/10.1038/nri3370
  3. 3Robinson et al., 2015 The IPD and IMGT/HLA database: allele variant databases. Nucleic Acids Research 43:D423-D431. doi.org/10.1093/nar/gku1161