Equivalent mixed model for joint genetic evaluation considering molecular and phenotypic information

Currently efforts are underway to introduce molecular information into
genetic evaluation systems. A particular situation is genomic selection
however simpler cases exists where major genes are known and used
by breeders. A new alternative strategy for the prediction of gene effects
and especially their smooth integration into genetic evaluations based on
an equivalent method was developed from existing theory. Underlying
hypothesis were based on the idea that knowledge of genotypes will not
affect overall additive genetic variance but only change expected values
of genetic effects for animals with known genotypes. The developed
equations were modified to allow that not all animals were genotyped.
As the underlying mixed model is open a very large range of models
can be used in situations including random regression models, multipletrait,
maternal effects and multiple-across-country-evaluation models.
Computations involved successive solving of two mixed models, with
the use of an linear extrapolation to speed up convergence of gene effects.
The method was tested for several known major genes and QTL, e.g. for
the mh gene in the dual-purpose Belgian Blue population in Belgium.
Modifications of the method could also be developed to be useful in the
context of genomic selection.