Marker imputation with low-density marker panels in Dutch Holstein cattle

Area of Science:

• Animal Genetics
• Quantitative Genetics
• Bioinformatics

Background:

• High-density genotyping arrays enabled genomic selection in dairy cattle.
• Low-density single nucleotide polymorphism (SNP) panels are crucial for broader genomic selection implementation.
• Imputation, the prediction of ungenotyped markers, allows consistent chip usage across traits and breeds.

Purpose of the Study:

• To evaluate the accuracy of imputation using low-density genotyping arrays in the Dutch Holstein cattle population.
• To compare the performance of different SNP densities and imputation programs.

Main Methods:

• Tested five genotyping array densities (384 to 6,000 SNP).
• Utilized DAGPHASE and CHROMIBD programs for imputation, assessing error rates.
• Analyzed imputation efficiency based on animal relationships and marker density.

Main Results:

• Allelic imputation error rates ranged from 11.7% to 2.0% (DAGPHASE) and 10.7% to 3.3% (CHROMIBD) with increasing marker density.
• Imputation accuracy improved significantly when both parents were genotyped, with error rates below 1% using 1,500+ SNP.
• Approximately 3,000 markers (1 SNP/Mb) predicted missing alleles with 3-4% error.

Conclusions:

• Low-density SNP panels can achieve accurate imputation for genomic selection in dairy cattle.
• CHROMIBD outperformed DAGPHASE at lower marker densities or with more genotyped ancestors.
• Further research is needed to assess the impact of imputation errors on genomic selection accuracy.