Page 298 - Linear Models for the Prediction of Animal Breeding Values

These solutions are the same as those obtained by direct inversion of the coefficient
matrix in Section 3.2 or iterating on the coefficient matrix in Section 17.2. However,
as stated earlier, the advantage of this method is that the MME are not set up and
therefore memory requirement is minimal and can be applied to large data sets.

17.4.2 Animal model with groups

Example 17.4
With unknown parents assigned to phantom groups, the procedure is very similar to
that described in Section 17.4.1, with no groups in the model except in the way the
pedigree file is set up and animal solutions are computed. Using the same data,
parameters and model as in Example 3.4, the methodology is illustrated below.

DATA PREPARATION
The pedigree file is set up as described in Section 17.4.1 with ancestors with unknown
parentage assigned to groups. The assignment of unknown parents for the example
pedigree has been described in Section 3.6. However, there is also an additional
column for each animal indicating the number of unknown parents for each animal.
The pedigree with unknown parents assigned to groups and the additional column
indicating the number of unknown parents is as follows:

Calf Sire Dam Number of unknown parents

1 9 10 2
2 9 10 2
3 9 10 2
4 1 10 1
5 3 2 0
6 1 2 0
7 4 5 0
8 3 6 0

and the ordered pedigree for the analysis is:

Animal Code Sire or progeny Dam or mate Number of unknown parents
1 1 9 10 2
1 2 4 10 1
1 2 6 2 0
2 1 9 10 2
2 2 5 3 0
2 2 6 1 0
3 1 9 10 2
3 2 5 2 0
3 2 8 6 0
Continued

282 Chapter 17