Evaluation of Methods for Calling, Phasing, and Imputing Genotypes in Dairy Cows

• Activity Number: 557
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Host Chapter-Puget Sound
• Author(s): David Whitney* and Michael Keehan and Brian Browning
• Companies: University of Washington and Livestock Improvement Incorporation and University of Washington
• Keywords: phasing ; imputation ; genotype calling

Abstract

To make decisions about the breeding of dairy cows, scientists make use of genetic information from sequence data. However, high quality sequencing of every animal is financially prohibitive. It is much less expensive to sequence a subset of cows, genotype the remaining cows for a subset of markers and impute the ungenotyped markers using the sequenced animals as a reference panel. Beagle is a software package for calling, phasing, and imputing genotypes. Beagle's methods are based on a hidden Markov model for inter-marker correlation. We evaluate Beagle's methods for calling genotypes from low-coverage (11x) sequence data for 556 dairy cows, 40 of which come from families having independent genotype calls from SNP array data. We measure genotype call accuracy by the discordance between the sequence genotype calls and the SNP array genotypes. We find that calling genotypes in the sequence data using Beagle results in a 7-fold reduction in discordant genotypes compared to the original genotype calls which were made without accounting for inter-marker correlation. The accuracy of Beagle's methods for phasing and imputation are currently being evaluated and will be reported.