A review of the validation of national genomic evaluations

Abstract

A descriptive review of the data and results of GEBV tests in a period of January 2013 to May 2014 was provided in this study. There were 357 GEBV tests, 259 of them for Holstein. 51 of 259 tests were repeated tests for the same traits. Currently, 4 tests are involved in the GEBV test. 283 cases passed the GEBV test, 206 of them passed the statistical test. These numbers were 216 and 164 cases for Holstein, respectively. Considering the 259 cases for Holstein, the minimum and the maximum number of test bulls were 11 and 4,892. Half of the GEBV tests had less than 330 test bulls, and 58 cases had less than 100 test bulls. The average proportion of genotyped candidate bulls (test bulls) was 0.82, with no clear difference between the statistically passed and the failed tests. (Genotyping) selection intensity was generally low, and expected regression slope (E(b₁)) was close to 1 for most of the cases. The range of the regression slope (b₁) was from 0.16 to 1.8, and the range of |b₁–E(b₁)| was from 0.001 to 0.799. The reason for some cases with large |b₁–E(b₁)| passing the statistical test was large SE(b₁), which was ranged from 0.007 to 0.874. 150 of 259 GEBV tests for Holstein had SE(b₁) < 0.1. R² of the GEBV test model were intermediate to low, with only 46 of 259 cases having R² > 50. There might be concern for populations passing the GEBV test with very low R² and very low number of test bulls.