Keywords: KIR Netherlands WGS immunologyThe killer cell immunoglobulin-like receptor (KIR) proteins evolve to fight pathogens and mediate the body's reaction to pregnancy. They also play roles in autoimmune diseases, cancer, transplantation, and immunotherapy. KIR genes are under selection pressure for variation at both the structural/haplotype and allele/base levels. Consequently, current reference alleles number in the thousands of exon-level variants, hundreds of full-genes, and dozens of full haplotypes. These multi-faceted variations make it impossible to interpret KIR haplotypes from abundant short-read genome sequencing data using existing methods. Here, we developed an efficient computational approach to accurately interpret KIR haplotypes with short reads from whole genome sequencing (WGS). We designed synthetic candidate sequence probes of size 25 base pairs (25mers) by analyzing known allele sequences. We then queried whole genome sequences of The Genome of the Netherlands (GoNL) with the designed probes to infer genotypes at the gene level, and imputed the haplotype structures using family relationships and expectation-maximization to resolve the ambiguities in haplotype inference. From the results, we report individual haplotype pair predictions, haplotype frequencies, and novel markers for 19 genes, 24 intergene regions, 18 haplotypes, and 8 half-haplotypes. 92% of GoNL cohort can be explained in 18 haplotypes and the frequencies are similar to expectations from previous studies. We applied the combination of gene, intergene, and haplotype associations to leverage new information from short-read sequences.These methods can also be used to predict KIR haplotypes and discover KIR markers in other populations, and the interpretation algorithm is deployable as in an easy-to-use container.