Strategy for incorporating newly discovered causative genetic variants into genomic evaluations

Abstract Text:

With sequence data available for an increasing number of dairy cattle, discovery of causative genetic variants is expected to be frequent. Current genomic evaluation systems require genotypes for all markers that contribute to an evaluation. A minimum number of animals with an observation for a new marker is required for accurate imputation. The SNP calls derived from sequence data from the 1000 Bull Genomes Project for 444 Holsteins were combined with SNP genotypes from bulls in the predictor population for US national genetic evaluations to impute candidate variants from the full sequence. From this imputed data, the set of SNP used in genomic evaluation along with the newly discovered causative variants were selected and stored. Those genotypes replaced the original genotypes for the bulls when extracting genotypes for genomic evaluation. The time required for imputation is substantially reduced in routine evaluation by using the haplotype library and assignments from the previous evaluation. To create suitable prior information for the expanded SNP set, genotypes for approximately 100,000 animals (including the predictor bulls and many cows with genotyped progeny) were imputed without priors. This step took about 1 d; if the full set of animals had been used, it would have taken over a week. The accuracy of this approximation was tested using the December 2015 Holstein genomic evaluation of nearly 1 million animals. Genotypes from 978,987 bulls and cows were used to create the priors, which were used to impute the December 2015 Holstein genotypes. Of the nearly 60 billion comparisons, 97.7% were identical, 1% differed by 1 allele, and 1.2% differed by a missing allele. Efficient methods that result in higher concordance may be possible. Adding new highly informative markers to the evaluation process is expected to improve prediction accuracy. In addition, excluding other markers may further increase accuracy if they contribute more noise than value when highly informative markers are included. The procedure developed enables newly discovered causative variants to be added to genomic evaluation almost immediately, which saves the time previously required for a marker to be added to a new genotyping chip as well as the time required for sufficient animals to be genotyped with the new chip to achieve adequate imputation accuracy. With this strategy, the benefits from adding new markers to genomic evaluation can be realized sooner.

Keywords: causative variant, sequence data, genomic evaluation