Evaluation of Three Methods for Genomic Additive and Dominance Relationships using SNP Markers

Genomic relationships from three methods for genomic prediction were compared using 32 fullsib pairs and 100 halfsib pairs of Holstein cattle. Method I uses the total heterozygosity and Method II uses the average of diagonal elements in the denominator of the genomic additive relationship matrix, and we applied these methods to dominance relationships. Method III has diagonal values of ‘1’ and off-diagonal values in the range of -1 and 1 for additive and dominance correlation matrices. The calculation of genomic relationships used 1K, 3K, 7K and 41K SNP markers, which had similar results, and the calculation of coancestry coefficients used the pedigree tracing back about five generations. Results reported below are the 41K results with the removal of three fullsib outliers and nine halfsib outliers. For fullsibs, genomic additive relationships were approximately twice as large as coancestry coefficients. The mean value of genomic additive relationships was 0.471 for Method I, 0.478 for Method II, and 0.488 for Method III, while the mean value of pedigree coancestry coefficients was 0.262. For halfsibs, Methods I-III had mean additive relationship of 0.213-0.221 and the average pedigree coancestry coefficient was 0.141. Genomic and pedigree additive relationships had low but significant correlations for halfsibs (r = 0.18, P < 0.05). The mean dominance correlation for fullsibs was 0.100 for Method I, 0.248 for Method II and 0.254 for Method III, compared to the expected fullsib dominance correlation of 0.25 assuming no inbreeding. For halfsibs, average genomic dominance correlations were 0.0007-0.002.