QC analyses of SNP array data: experience from a large population of dairy sires with 23.8 million data points

Abstract

The use of a high throughput SNP genotyping platform with 15,380 bovine SNP assays, across 1546 dairy bulls resulted in a data set of approximately 23.8 M SNP data points. Stringent control measures based around low polymorphic content, sample failure, deviation from HWE, low call rate, non-Mendelian inheritance, tri-allelic SNP, and incompatible clustering of data, resulted in removal of 4321 SNPs. The majority (2973) were due to low polymorphic content (MAF<0.01), with the remaining features consistent with assay quality. Despite the need for removal of some SNP, repeatability of SNP call rate was extremely high (>99%) across repeat samples, and between platforms. SNP technology has now matured where comprehensive genome-wide analyses can be conducted in cattle with a high degree of robustness.