Characterization of the genome-wide linkage disequilibrium in 2 divergent selection lines of dairy cows

The objective of this study was to describe results of a genome-wide map of single nucleotide polymorphisms (SNP) and assess the linkage disequilibrium (LD) level in 2 divergent selection lines of dairy cows. DNA extracted from 299 Holstein cows was used to determine genotypes in 54,001 SNP loci using the BovineSNP50 array (Illumina Inc., San Diego, CA). Animals were from 2 genetic lines (166 genetically selected for fat and protein yield vs. 133 controls) raised on an experimental farm. Data edits removed loci with a major allele frequency greater than 0.95, genotypes in fewer than 100 cows, and missing valid chromosomal assignment or position. After edits, 41,859 loci (77.5% of the original total) were kept for further analysis. Linkage disequilibrium (LD) values were calculated for all possible syntenic SNP locus pairs located within intervals of 1 million base pairs, as the squared correlation between alleles. Pairwise haplotypes were determined using parsimony. Linkage disequilibrium was calculated for all animals and then for each genetic line separately. The average LD calculated across all chromosomes was 0.069, 0.071, and 0.075 for all, control, and select line cows, respectively. Genetic line had a statistically significant effect on LD. Of all locus pairs studied, 53,487 to 95,279 (depending on the data set) were in LD >0.30, which may be considered the minimum useful for mapping purposes and genomic selection. Useful LD was mostly found between adjacent pairs located within 30,000 to 50,000 bases. A few locus pairs (844–1,070 in the 3 data sets) were found in almost perfect (>0.99) LD. The overall product-moment correlation of LD values between the control and select lines was 0.79 (significantly different from 1), ranging from 0.71 to 0.84 for different chromosomes. Looking at this correlation by SNP pair distance revealed that persistence of LD phase across the 2 lines extended chiefly for 200,000 bases. Selection is likely to have affected LD in the studied cow population. These results may be useful to gene detection and genome-wide association studies.