Experience with single-step GEBV validation in CRV Holstein-Friesian breeding program

Abstract

Validation of genomic evaluations is an important step in the development cycle. The usual way to validate is to regress deregressed conventional EBV of young bulls from a full evaluation on predicted GEBV from a reduced evaluation, where some recent years of data have been omitted. Some alternative approaches to validation of (single step) GEBV were tested. The types of validation performed were regression of GEBV or deregressed GEBV on predicted GEBV of either validation bulls or validation cows. Additionally we compared validations on DEBV DGEBV for some traits using either in house validation methods or the newly implemented Interbull validation software. The results indicate that validations on DGEBV produce satisfactory results. Validation using cow GEBV can be used when the number of validation bull is too limited (< 300 bulls). Validation on DEBV seems no longer to be a good approach, since input GEBV are often more reliable than the DEBV they are tested against.