Re: MA Blups

From: Arthur <asremlforum_at_VSNI.CO.UK>
Date: Sat, 11 Jul 2009 02:13:58 +0100

Dear Henry

Indeed.

I assume you just want one variance component (your model has two)

I am not familiar with the paper.

I reinverted the giv matrix to see its structure.
Its structure is 7 sets of 2, with the values based on the pedigree and marker states.

For the A matrix, the design matrix is an identity (7x7)

For the giv matrix, I think you want a design matrix

1 1 0 0 0 0 0 0 0 0 0 0
0 0 1 1 0 0 0 0 0 0 0 0
0 0 0 0 1 1 0 0 0 0 0 0
0 0 0 0 0 0 1 1 0 0 0 0
0 0 0 0 0 0 0 0 1 1 0 0
0 0 0 0 0 0 0 0 0 0 1 1

Given the animal coding,
A2 14 != animal !*2
A1 14 != A1 !-1

trait ~ mu fixed !r animal giv(A1) and(A2)

should do the trick

Title: exampledat.
animal !P
fixed *
trait
pmarker !V101=V4 !-1
mmarker !V102=V5 !-1 !+V101 !*0.98
A2 14 != animal !*2
A1 14 != A2 !-1
YY !=trait !+V102

exampleped.txt
example.GIV
exampledat.txt !SKIP 1 !EXTRA 5

YY ~ mu fixed !r animal .1 !GF giv(A1,1) and(A2)

In this code, I have increased the marker effect in the data,
and fixed the animal component so that some variation is picked up
by the giv term

The giv() function associated the giv matrix with the term A1

The and() functions overlays the A2 design matrix with the A2 matrix
to create the expandel design matrix required.

You could formally associate the models by appending

0 0 2
animal 0 AINV 0.1 !GF
giv(A1,1) 0 GIV1 0.1

Note the .giv file needs file extension .GIV

------------------------
Arthur Gilmour

Retired Principal Research Scientist (Biometrics)

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

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