Challenges

While validation of national genetic evaluation is a practice very well developed and applied within the International Bull evaluations for dairy cattle (MACE), the situation is very different for the international bull evaluation of beef cattle (Interbeef). One major different between MACE and Interbeef evaluations is that Interbeef evaluations are carried out using raw performance data while MACE evaluations are based on national EBVs. Moreover, the beef industry does have its own peculiarities compared to the dairy industry:

• Lower usage of artificial insemination: fewer “well-proven” bulls available within or across countries;
• Smaller size of contemporary groups;
• Wider use of multi-traits and multi-breeds evaluations;
• Other traits more commonly used as predictors for the trait of evaluation;
• Many traits are maternally influenced.

It is thus relevant to identify a suitable validation method that could work well for the beef evaluations. We will exploring the possibility to apply a modified version of some of the methodology currently applied in dairy, especially given the high interest of applying genomic evaluations in the beef evaluations as well.

A detailed report on beef validation can be found here.

In the context of international genetic evaluations, the definition of the trait as recorded by the different countries does play a very important role. In the MACE evaluations, an important parameter to assess the similarities of trait definitions across countries is represented by the across country correlations.

From this point of view, correlations can be negatively affected if the definition of the traits evaluated differ too much from country to country. In this regard, trying to harmonise traits across countries is an important exercise that can result in stronger correlations and, thus, more accurate evaluations.

Following the work done in 2022 in creating recommendations for harmonizations of calving traits, during 2023/2024 a study on the harmonisation of fertility traits was carried out at Interbull Centre and resulted in a set of recommendations. These have been made available to countries.

The first step in achieving harmonization in performance recording and genetic evaluation methods is comparing information across countries. Until now, information provided by these countries as description of their national systems, was received and stored as free text and, for this reason, difficult to compare. 

As we move from the current GE/GENO/BEEF text forms to a databasestructure (PREPdb), such comparison is required in order to better standardise the expressions and/or descriptions of definitions, procedures and statistical models. At the end of 2022, four trait groups had been standardized and the information of their performance recording and genetic evaluations was moved from text documents to the PREPdb electronic forms. During 2023-2024 the remaining four trait groups have been included in the PREPdb. 

Having all information gathered and in a standard form, creates a better overview of the practices applied by different countries. Such overviews are presented in our resources  here.

Genomic reliabilities (GREL) are less comparable across countries than conventional reliabilities. It is important that the GREL are consistent with the conventional REL. For this reason, we aim to develop standard procedures for GREL calculations for national genomic evaluations. The methodology developed is now also applicable to Single Step model for test day traits. The genomic reliabilities from Single Step models are now consistent between genotyped and not-genotyped groups when compared to a multiple step model. The GREL software has also been improved in its efficiency in relation to computational time which makes it possible to use all SNPs.

As the polygenic effect is not properly accounted for in the calculation process, the scaling effect of genomic reliabilities is still being investigated. It will be important to assess if the final adjustment procedure could still be considerate accurate enough. The effect of percentage of SNPs used for calculating genomic reliabilities should also be investigated further.

With the spreading of genomic evaluations introduced by an increasing number of countries, national genetic evaluations’ results, and therefore also international results, have started accumulating a new source of bias due to genomic pre-selection. The amount of this bias varies from country to country, depending on the level of pre-selection applied. Interbull aims to identify the best way to account for the different level of genomic pre-selection from countries and, consecutively, on developing a new MACE software able to properly model it.

With the onset of genomic validations and the accumulation of genomic pre-selection in the national genetic evaluations’ results, it was necessary to review the validation methodologies in use, so to make them able to better detect traces of genomic pre-selection. This is expected to result in the creation of new GEBVtest software in order to overcome the known concerns related to the current available software which are:

• Arbitrary of de-regressed method applied by countries
• The base of expression in reduced and full data can be different
• Both EBV and dEBV can carry genomic pre-selection’s bias.

The new software will be enhanced with the following features:

• Adjustment of base of expression between reduced and full data
• Use of GEBV and dGEBV instead of EBV and dEBV as they can be biased