A catalog of HLA region SNPs with functional annotations, disease associations and correlations with HLA types

The HLA complex is the most gene dense and polymorphic region in the human genome, and not surprisingly, shows a large number of disease associations. The latest genome assembly contains u003e 250 K SNPs in the HLA region (chr6:28.5 to 34.5 Mb). The emphasis in disease association studies has been on the HLA gene polymorphisms, despite the fact that non-HLA genes vastly outnumber HLA genes. We adopted a comprehensive approach to systematically analyze all HLA region SNPs by functionally annotating, and mapping them to conserved extended haplotypes by genotyping 95 reference cell lines using the Illumina ImmunoChip. We also measured expression levels of classical HLA genes in eight HLA homozygous reference cell lines at baseline and after three treatments (heat shock, LPS, and IFN-gamma). We screened the GWAS catalog, dbGAP and GRASP database for disease associations of SNPs, and assigned CADD, RegulomeDB, and PolyPhen/SIFT scores to all SNPs. SNPs were assessed for alterations they cause in transcription factor binding sites, or microRNA/microRNA binding site sequences. The overall data revealed SNPs representing known HLA associations, suggested mechanisms for some HLA associations via linkage disequilibrium between HLA alleles and functional SNPs, explained the source of the extra amount of DNA specific to the DRB4 family of haplotypes as copy number variation, and unraveled areas of the HLA region with insufficient coverage in GWAS arrays. The SNPs with the most deleterious effects were mainly in non-HLA genes of the region, and most have not been included in GWAS arrays. We also noted that there are more disease associations in the HLA region than listed in the GWAS catalog. Given the insufficient coverage in GWAS arrays due to difficulties in assay design and structural variation, sequencing-based studies are needed to unravel even more associations missed by GWAS. The catalog produced in this work combines HLA and SNP data with functional effects and induced expression levels, and should be of great value in the design and interpretation of HLA and disease association studies.