# Random Regression - maternal effects

From: bgudex <asremlforum_at_VSNI.CO.UK>
Date: Mon, 23 Mar 2009 07:34:40 +0000

Hi
I am trying to carry out random regression for live weight on a sheep dataset. but keep on getting one of the following errors when I try to add a polynomial for the maternal effect (Part 4). Simpler models work (eg parts 1 to 3) but I would prefer to use a maternal model if possible.
- convergence failed
or
- singularity in AI Matrix

According to the ASReml manual, the convergence failed error message can alleviated by:
- define US structures as positive definite by using !GP - done
- supply better starting values - tried
- fix parameters that you are confident of - done for the direct effects using values from Part 3
- fit a simpler model - trying to avoid if possible
- reorganise the model to reduce covariance terms (eg, use CORUH instead of US). - What is CORUH [Question] not found in either ASReml manual or in the help file! Tried CORR but it gave the singularity as well.

Have tried editing the data to remove dams with only one progeny and progeny from maternal grandsires with few great offspring with no effect.

What else can I try [Question]

This is the as file
Random Regression Analysis
ID !P
Sire !P
Dam !P
sex 2 !A
wgt
bt * !A
eweage * !A
age
cg * !A
byr
records
grpped !ALPHA !GROUPS 42
dat !DOPART 4 #!CONTINUE !maxit 50
!PART 1 # no random regression, just polynomial fitted to age
wgt ~ mu leg(age,1) bt eweage sex !r ID !f cg

!PART 2 # random regresion on direct effects only
wgt ~ mu leg(age,1) bt eweage sex !r leg(age,1).ID !f cg
1 1 1
31838 # number of individuals in dataset
leg(age,1).ID 2
2 0 US !GP
10 # Vintercept
0 10 # COVintercept&slope Vslope
ID

!PART 3 # random regression on direct effects, dam fitted normally
wgt ~ mu leg(age,1) bt eweage sex !r !{leg(age,1).ID Dam !} !f cg
1 1 1
31838 # number of individuals in dataset
leg(age,1).ID 2
3 0 US 10 0 10 0 0 10 !GPPPFFP
ID

!PART 4 # random regression on both direct and dam effects
wgt ~ mu leg(age,1) bt eweage sex !r !{ leg(age,1).ID leg(age,1).Dam !} !f cg
1 1 1
26160 # number of individuals in dataset

leg(age,1).ID 2
4 0 US 105 64 45 0 0 4 0 0 0 0.2 !GFFFFFPFFPP
#0 0 0 10 # COV COV COV Vslope(maternal)
ID

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

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