Imputation efficiency with different low density chips in French dairy and beef breeds

Abstract

Low density chips are appealing alternative tools in order to reduce genotyping costs. Such a chip is already commercially available. The best way to use low density chips is to impute data to a more dense coverage such as the standard 50K genotype. Two alternative in silico chips are presented and include markers selected to optimise Minor Allele Frequency and spacing. The objective of this study was to compare imputation accuracy of these custom low density chips with the commercially available 3K chip. Three French dairy and beef breeds were studied: Holstein, Montbéliarde and Blonde d’Aquitaine with respectively 4,037, 1,219 and 991 50K-genotypes. Markers were masked for the validation population in order to mimic a low density genotype. Imputation was realised with Beagle software. 95% to 99% of alleles were correctly imputed depending on the breed and the low density chip. Custom low density chip gave better results. The gain to use 6K chip was found to be even higher for beef breeds such as Blonde d’Aquitaine. A low density chip with 6,000 markers is a valuable genotyping tool that is suitable for both dairy and beef breeds. Such a tool can be used for pre-selection of young animals or large screening of the female population