Interbull validation test for genomic evaluations

Authors: MANTYSAARI, E - Mtt Agrifood Research Finland; LIU, Z - Mtt Agrifood Research Finland; Vanraden, Paul

Submitted to: Interbull Annual Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 5/1/2010
Publication Date: 6/1/2010
Citation: Mantysaari, E., Liu, Z., Van Raden, P.M. 2010. Interbull validation test for genomic evaluations. Interbull Annual Meeting Proceedings. Interbull Bulletin 41, 5 pp.

Interpretive Summary: Genomic evaluations across countries can be validated by comparing genomic estimated breeding values of bulls in the youngest age classes to the evaluations that the same bulls received later. Participating countries submit two data sets of breeding values, one with reduced data where the last four years of phenotypic data are removed which is compared to a complete data set. By using validation tests, young bulls are less likely to be over or under evaluated which will lead to more accurate international genomic evaluations.

Technical Abstract: Genomic evaluations (GEBV) can be validated by comparing GEBVs of bulls in the youngest age classes to the daughter yield deviations (DYD) that the bulls will receive later. The GEBV are calculated from truncated data where the last four years of phenotypic data are removed. The test consists of weighted linear regression of DYD on GEBV. The intercept, regression coefficient and the coefficient of determination R2 are interpreted to represent average bias, prediction bias, and the validation accuracy, respectively.