2013 Annual Report
1a.Objectives (from AD-416):
The objective of this agreement is to develop statistical methods for the analysis of the eigenstructure (and eigenvalues resulting from decomposition of a matrix) of genetic (co)variance matrices constructed using high-density DNA marker data. At present, the effect of different (co)variance structures on matrix decompositions is not well understood, which hinders interpretation of those values. Knowledge of relationships among marker effects estimates, specific genetic (co)variance structures, and the eigenstructures of those matrices will improve the understanding of the biology of economically important traits in dairy cattle. Such tools may also facilitate the automated discovery of interesting features in high-dimensional data sets.
1b.Approach (from AD-416):
Single nucleotide polymorphism (SNP) effects estimated independently in the Italian and U.S. Brown Swiss cattle populations by Cooperator and ARS, respectively, will be used to develop methods for comparing genetic (co)variance matrices for various traits of economic importance in those breeds. Methods developed will be reported in the scientific literature. ARS and Cooperator will jointly develop the statistical methods using a shared set of simulated data. The methods will be applied by ARS to U.S. data and by the Cooperator to Italian data. Results will be used to compare the two populations.
The project is related to in-house objective 2 (develop a more accurate genomic evaluation system with advanced, efficient methods to combine pedigrees, genotypes, and phenotypes for all animals). Effects of genetic markers are used to predict genetic merit for both animals as a whole and for individual chromosomes. The comparison of effects at the genome-wide and chromosome-specific levels may indicate differences associated with the genetic control of groups of traits. A method for comparing genomic correlation matrices based on multivariate factor analysis was developed and applied to 3,096 U.S. Holstein with genetic evaluations for 31 productive and functional traits and 460 Italian Simmental bulls with evaluations for 12 traits. For Holsteins, between 7 and 9 factors were able to explain between 70 and 80% of the original variance, whereas 3 to 4 latent variables on average explained about 80% of the variance for Simmentals. For U.S. Holsteins, latent factors associated with milk yield, milk composition, udder morphology, strength, productive life, udder health, and daughter pregnancy rate were obtained at the genomewide level, with differences observed at the chromosome level. A single factor on Bos taurus chromosome 14 was associated with both milk yield and composition traits. On chromosome 18, sire reproduction and some type traits were associated with a common factor. For Italian Simmentals, the first latent factor was associated positively with milk yield and milking traits and negatively with muscularity. A second factor was associated with size and daily gain, a third with feet/legs and udder health, and a fourth with milk composition traits. A factor on chromosome 7 was associated with both daily gain and milk composition. A manuscript describing these results is being prepared for submission to a peer-reviewed journal.