VCOV structure for GxE

Dear all,

This is related to an earlier post on weighted analysis of METs.

In addition to the field trial data, marker information (SNP) on barley lines is also available. Therefore, it is possible to construct the kinship matrix and use it as VCOV matrix for the genotypic effect:

yield !wt weight ~ mu !r giv(gen,1) env gen.env
1 1 0
0 !S2==1.0

kinship matrix turned out to be not positive definite, but this problem is solved when !NSD qualifier is used in ASREML 3

however, when I want to build a VCOV structure for GxE effect, model failed to converge due to singularities, using the following code:

yield !wt weight ~ mu !r giv(gen,1) env gen.env
1 1 1
0 !S2==1.0
gen.env 2
gen 0 GIV1 1
env

instead of identity matrix for env, I have also tried to use CORUH, but it failed to converge as well

are there other suggestions for env part of the matrix, or better strategy for selecting the starting values?

Thanks,
Jerko

-------------------- m2f --------------------

Sent using Mail2Forum (http://www.mail2forum.com).

Read this topic online here:
http://www.vsni.co.uk/forum/viewtopic.php?p=776#776

-------------------- m2f --------------------
Received on Wed Aug 25 2009 - 12:56:28 EST