Suppose I have a very simple GS model and data from an augmented design where checks were used in every block but not every entry is represented in every block. I now want to regard the genotypes at the level of the checks as fixed and the genotypes at the level of the entries as random. For this I prepare a column in a dataset that seperates checks and entries called "check" with a simple two level factor and prepare my relationship matrix (GRM) to have only as many levels as the number of entries.
I know that you can fit effects only for certain factors with the at() keyword, however when I try to do it in the following way with a genomic relationship matrix I get a warning message:
trait ~ mu at(check,y).genotype,
!r at(check,n).grm1(genotype) Block
Warning message:
"The GRM matrix specified in grm1(genotype) is smaller (100) than genotype (110) and is extended with an Identity to cover the extra levels."
Should I be worried about this warning note. Is there a better way to seperate checks and entries?
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication