Support vector machine algorithms in the search of KIR gene associations with disease

Killer-cell immunoglobulin-like receptors (KIR) are membrane proteins expressed by natural killer cells and CD8 lymphocytes. The KIR system consists of 17 genes and 614 alleles, some of which bind human leukocyte antigens (HLA). Both KIR and HLA modulate susceptibility to haematological malignancies, viral infections and autoimmune diseases. Molecular epidemiology studies employ traditional statistical methods to identify links between KIR genes and disease. Here we describe our results at applying artificial intelligence algorithms (support vector machines) to identify associations between KIR genes and disease. We demonstrate that these algorithms are capable of classifying samples into healthy and diseased groups based solely on KIR genotype with potential use in clinical decision support systems.