Asreml users,
Hello all. I’m a fairly new user of asreml and am having some difficulties with fitting the appropriate G matrix and getting relevant predictions.
I have a dataset of 26 genotypes (slightly unbalanced) at three locations and wish to allow genetic variance at each location to be different. I have tried to fit a US G matrix, but am not 100% sure what order to fit the various covariance and variance components.
Assuming I have σ2G11 of 1.26, σ2G22 of 1.06, σ2G33 0.36; covar 1,2 of 0.7, covar 1,3 of 0.5, covar 2,3 of 0.9, is the order I’ve specified them below in the .as file correct?
.as file is as follows:
98-52 Week 2
Row 5 !I
Plot 10
Rep 2
Clone 26 !A
Locn 3
ccs
98wk2.dat !skip 1
ccs ~ mu !r Locn Locn.Clone
3 2 1
5 Row ID 0 !S2=0.77
10 Plot AR -0.2
5 Row AR -0.55 !S2=0.94
10 Plot AR 0.42
5 Row AR -0.48 !S2=2.43
10 Plot AR 0.32
Locn.Clone 2
3 0 US 1.26 .7 1.06 .5 .9 .36
26 0 0
Predict Clone
Also, is there some way of getting asreml to calculate the covar components for me, or must this be done manually before inputting this into .as file?
Then, how do I get predict to work? Obviously this above is not correct, as it gives all clones (in the .pvs file) as having the same prediction. When I looked at the .sln file, the blups are quite different within and across locations.
Finally, how do I interprete the outputted G matrix. For the above analysis (if correct), the outputted G matrix is:
1.12 0.43 -0.09
0.35 0.59 1.77
-0.04 0.66 0.24
How can I calculate a pooled σ2G, σ2GL and the genetic correlations for this analysis. Also, as the upper triangular values as rG coefficients, the value of 1.77 seems very wrong. Or, do these values need to be multiple by some factor, such as final error?
If you could help me out, it would be greatly appreciated.
Thanks, Allan
Allan Rattey
PO Box 86
Indooroopilly 4068
Ph: 33313337
Fax: 38710383
Email: arattey@bses.org.au