Comparison between SNP array and imputed data to estimate population structure and ROH hotspots in horse breeds

Area of Science:

• Animal Genomics
• Population Genetics
• Bioinformatics

Background:

• Single nucleotide polymorphism (SNP) arrays are common for livestock genomic studies, but whole-genome sequencing (WGS) offers higher resolution.
• Genotype imputation is standard for increasing genomic resolution in association studies.
• This study extends imputation to biodiversity analyses, comparing SNP array data before and after imputation in horses.

Purpose of the Study:

• To evaluate the utility of imputed sequence-level genotypes for horse breed biodiversity analyses.
• To compare genetic variability, population structure, and runs of homozygosity (ROH) using SNP array data versus imputed WGS data.
• To assess the impact of imputation on the detection and annotation of ROH islands.

Main Methods:

• Imputed a 40 k SNP dataset from 281 horses (12 breeds) to sequence-level using a 327-sequenced individual reference panel.
• Generated approximately 9 million markers post-filtering for the imputed dataset (DS_IMP).
• Analyzed genetic variability, population structure (MDS, admixture), and ROH using both SNP array (DS_SNP) and imputed datasets.

Main Results:

• Genetic indices and relationships showed high correlations (>0.8) between SNP array and imputed data, confirming imputation reliability.
• Imputation amplified genetic proximity between breeds, especially those with limited representation in the WGS reference panel.
• ROH investigations revealed overlapping regions, with 79 out of 141 ROH islands from DS_SNP perfectly matching imputed data, aiding gene annotation.

Conclusions:

• Imputed genotypes are a reliable alternative to SNP array data for assessing horse breed population structure and genetic diversity.
• Imputation improves ROH detection and gene annotation within ROH islands.
• The accuracy of imputation-based results depends on the reference panel's quality and breed representation.