|
Post by wolfemd on Mar 18, 2015 14:53:33 GMT
Hi again,
I would like to solve a mixed model with GCTA that includes a random effect (for replication). Such a random effect has an identity matrix for a covariance structure (replicates are assumed IID). If I could supply the identity matrix to GCTA this should work (see my other question, just posted), but the size of the identity matrix would be Nreps by Nreps. Would GCTA accept this, or would it expect each matrix to be associated with the number of individuals in the study rather than the number of reps?
If I can do this stuff in GCTA, it wins my vote as an awesome mixed model solver!!!
Cheers and thanks for the great software,
Marnin
|
|
|
Post by Zhihong Zhu on Mar 20, 2015 2:47:48 GMT
Hi Marnin,
I'm afraid the number of rows or columns of GRM is the same as the number of individuals.
I think you may derive covariance matrix of the random effect, fitting into the regression model (as an additional GRM).
Cheers, Zhihong
|
|