From: Fikret Isik <fisik_at_UNITY.NCSU.EDU>

Date: Tue, 18 Mar 2008 23:10:16 -0400

Received on Thu Mar 18 2008 - 23:10:16 EST

Date: Tue, 18 Mar 2008 23:10:16 -0400

Dear ASReml users,

I am working on a Multi Environment Test (MET) data.

DATA:

Diallel meting design was used to produce crosses.

Thus, some of the parents were used both as female and male.

There are also 4 different checklots tested at each site.

Checklots are unimproved seedlots tested with parents to measure genetic

improvement.

There are 3 test series. Each series has 4 sites.

A randomized complete block design was used at each site.

Data were standardized by each site's phenotypic standard deviation to

reduce the scale effect.

PEDIGREE FILE:

I created a pedigree file so the design matrices are coded conformably

for factors in the model.

With pedigree, I can get predictions for 3 genetic groups.

OBJECTIVE:

To obtain BLUP breeding values of parents

COMMAND FILE (.AS):

!PART1

Fits a heterogeneous error variance structure (separate error for each

site) and no specific G structure.

In order to estimate a separate variance for checklots, I included

UNI(GENOTYPE,1) as suggested by Greg Dutkowski.

Because the relationships among progeny of checklots is not known.

This part looks running fine (LogL converges).

The only message indicating a problem is “Notice: 185 singularities

detected in design matrix” ??

The Problem: I get odd BLUP values. They are all positive instead of a

distribution around zero?

!PART2

Again, Error variance structure is heterogeneous. G structure is

homogeneous (each site has the same additive genetic variance). However,

we assume that parents are correlated across 4 sites within a series

(but they are independent across 3 series).

Thus, I would like to add a correlation to the G matrix to account for

correlations between parental GCA values across sites (within a series).

This part is not correct I belive.

The actual data have 190 sites. The code here is to develop model and

fit all the data (about 200 000 observations).

How can I focus on SERIES instead of writing a complex G structure for

190 SITES?

!PART3

Has a homogeneous additive genetic variance (0.26) with a fixed

correlation between sites (0.78)

Except the following message in the output file (.ASR) I did not see any

error message:

"Notice: 196 singularities detected in design matrix."

The problem withs part is that it treats the same traits (HEIGHT) as

different variables at each site and

the BLUP for parents are odd (range from 3 to 18).

The predictions for the UNI(genotype,1) are distribibuted around zero.

I will appreciate receiving any comments on the code

as well as overall suggestions on writing a code for a large job with

complex structures.

Best regards,

Fikret Isik

AS file:

-------------------------------

!WORK 1 !ARG 1

Title: Region4 subset data BLUP analysis March 08

treeID 12681 !A

parent1 77 !P !LL 24 # needs 20 characters to hold level names

parent2 77 !P !LL 24

genotype 2 !A # checklots =1, normal parents=2

series 3 !A #

site 12 !A # Each test Series has 4 Sites

rep 6 !I

rust

strt

fork

height_z

vol_z

C:\BLUP\Data\R4_ped.txt !SKIP 1 !ALPHA !MAKE !GROUP 3 !GIV !DIAG

C:\BLUP\Data\R4_dat.csv !CSV !SKIP 1 !DOPART $A

!PART 1

# Model has US structure for G, (GCA, GxE variances) are estimated.

# Heterogeneous Error, Homogeneous G structure

! CS model with Heterogeneous Error variance structure

!CONTINUE !EXTRA 6

height_z ~ mu series site.series rep.site.series !r ,

parent1 and(parent2) -parent2 and(parent2) uni(genotype,1) ,

site.series.parent1 -site.series.parent2 and(site.series.parent2)

12 1 0

952 0 IDEN !S2=3.9

1279 0 IDEN !S2=5.5

926 0 IDEN !S2=3.1

762 0 IDEN !S2=2.6

1080 0 IDEN !S2=4.0

1165 0 IDEN !S2=6.7

1115 0 IDEN !S2=6.1

1094 0 IDEN !S2=6.0

1170 0 IDEN !S2=5.1

1026 0 IDEN !S2=3.0

1001 0 IDEN !S2=3.8

1111 0 IDEN !S2=3.8

!PART 2

# Homogeneous G structure,

# Parental GCA values are correlated across sites within series

! CS model with Heterogeneous Error variance structure.

!CONTINUE !EXTRA 6

height_z ~ mu series site.series rep.site.series !r ,

site.parent1 -site.parent2 and(site.parent2) uni(genotype,1)

12 1 1

952 0 IDEN !S2=3.9

1279 0 IDEN !S2=5.5

926 0 IDEN !S2=3.1

762 0 IDEN !S2=2.6

1080 0 IDEN !S2=4.0

1165 0 IDEN !S2=6.7

1115 0 IDEN !S2=6.1

1094 0 IDEN !S2=6.0

1170 0 IDEN !S2=5.1

1026 0 IDEN !S2=3.0

1001 0 IDEN !S2=3.8

1111 0 IDEN !S2=3.8

site.parent1 2

site 0 US !+78 !GPUPUUP3UP4FP4FUP4FUUP4F3UP8FUP8F2UP8F3UP

.3

.09 .3

.09 .09 .3

.09 .09 .09 .3

0 0 0 0 .3

0 0 0 0 .05 .3

0 0 0 0 .05 .05 .3

0 0 0 0 .05 .05 .05 .3

0 0 0 0 0 0 0 0 .3

0 0 0 0 0 0 0 0 .06 .3

0 0 0 0 0 0 0 0 .06 .06 .3

0 0 0 0 0 0 0 0 .06 .06 .06 .3

parent1 0 IDEN

!PART 3

# Model has CORR structure for G

# Homogeneous G structure

! CS model with Heterogeneous Error variance structure

!CONTINUE !EXTRA 6

height_z ~ mu series site.series rep.site.series !r ,

site.parent1 -site.parent2 and(site.parent2) uni(genotype,1)

12 1 1

952 0 IDEN !S2=3.9

1279 0 IDEN !S2=5.5

926 0 IDEN !S2=3.1

762 0 IDEN !S2=2.6

1080 0 IDEN !S2=4.0

1165 0 IDEN !S2=6.7

1115 0 IDEN !S2=6.1

1094 0 IDEN !S2=6.0

1170 0 IDEN !S2=5.1

1026 0 IDEN !S2=3.0

1001 0 IDEN !S2=3.8

1111 0 IDEN !S2=3.8

site.parent1 2

site 0 CORUV 0.85 0.26

parent1

- text/x-vcard attachment: fisik.vcf

This webpage is part of the ASReml-l discussion list archives 2004-2010. More information on ASReml can be found at the VSN website. This discussion list is now deprecated - please use the VSN forum for discussion on ASReml. (These online archives were generated using the hypermail package.)