Discordant SNP Genotyping Between the Illumina Equine54K and 75KSNP Platforms Causing False Positive Associations

We performed two equine genome-wide association studies aimed at the identification of genetic loci associated with the development of (1) neuroxonal dystrophy (NAD) and (2) immune mediated myositis (IMM). Both studies involved the merging of genotype data from samples that had been genotyped on either the Illumina Equine 54K (1^st generation) or 75K (2^nd generation) SNP platforms and included unbalanced phenotypes from each array type. For NAD, 23 affected (genotyped on the 75K platform) and 63 controls (n=38 on 54K and n=25 on 75K) were genotyped over a shared 42,890 SNPs that passed quality control (geno>90% and maf >1%). A standard allelic case-control association test (λ=1.06) revealed seven genome-wide significant associations (P_unadj<1.16x10^-6) at chr13:27912038, chr19:57161987, chr3:38568320, chr20:19701770, chr18:31444786, chr27:21578541 and chr14:18380414. For the IMM study, 36 affected (n=1 on 54K and n=35 on 75K) and 50 control (n=20 54K and n=31 on 75K) QHs were genotyped over a shared 41,840 single nucleotide polymorphisms SNPs that passed quality control. Population clustering was corrected for using a Cochran-Mantel-Haesnzel (CMH) association analysis. The most highly associated SNP was on chr13:27912038 (P_unadj =1.26x10^-11); confirmed using GEMMA (P_unadj=6.53x10^-7) and GenABEL (P_unadj=5.68x10^-7). Additional genome-wide significant regions identified on the CMH test included chr20:19701770, chr19:57161987 and chr3: 38568320. Genotype analysis at these loci revealed that the SNPs were polymorphic on the 54K but homozygous at the minor allele on the 75K array, leading to false positive associations. These SNPs should be excluded prior to association testing on samples genotyped with the 1^st and 2^nd generation EquineSNP arrays.