|Hanford, Kathryn -|
|Kachman, Stephen -|
|Quaas, Richard -|
|Tempelman, Richard -|
|Fernando, Rohan -|
Submitted to: World Congress of Genetics Applied in Livestock Production
Publication Type: Proceedings
Publication Acceptance Date: May 9, 2010
Publication Date: August 1, 2010
Citation: Thallman, R.M., Hanford, K.J., Kachman, S.D., Kuehn, L.A., Quaas, R.L., Tempelman, R.J., Fernando, R., Pollak, E.J. 2010. Estimation Of The Proportion Of Variation Accounted For By DNA Tests. II: Phenotypic Variance. Proceedings of the 9th World Congress on Genetics Applied to Livestock Production, Leipzig, Germany. August 1-6, 2010. CD-ROM Communication No. 0918. Interpretive Summary: A relevant question in evaluating commercial DNA tests intended for application in marker assisted management (MAM) is "What proportion of the phenotypic variation in the target trait is accounted for by the test?" Therefore, two estimators of this quantity were evaluated by simulation of a population of 1000 animals with 100 sires, each with 10 progeny. One estimator was derived from a model including the target trait and the molecular genetic value (MGV) as a second trait. The MGV is a numerical value derived from the DNA test that predicts the total genetic merit of an animal. The second estimator was derived from single trait models for the target trait, with or without, the MGV in the model. Both estimators performed well and are recommended for evaluation of DNA tests for use in MAM.
Technical Abstract: The proportion of phenotypic variation accounted for (Rp2) is an important characteristic of a DNA test. Therefore, several estimators of this quantity were evaluated by simulation of 500 replicates of a population of 1000 progeny of 100 sires (3 levels of narrow sense heritability and 4 levels of Rp2). For each progeny, the observed trait and a molecular genetic value (MGV: prediction of total genetic merit from the DNA test) were simulated. The square of the phenotypic correlation from a two-trait animal model with the MGV and observed phenotype included as correlated traits can be used to estimate Rp2. The reduction in phenotypic variance between single trait sire models with and without the MGV covariate is an alternative estimator of Rp2. Both estimators worked well and are recommended for use.