[vsni.co.uk] Contact us
Author Message
Post new topic

  ASReml  ~  bivariate model with a weight on one of the traits

bart.ducro
Posted: Sun Apr 22, 2018 5:23 pm Reply with quote
Joined: 22 Feb 2008 Posts: 2
Dear all,
I am trying to fit a bivariate genetic model with a weight on (only) one of the response variables., (e.g. a pseudo_phenotype retrieved from a prior repeatability model and the weight factor is number of obs, the pseudo_phenotype is based upon).

The model would be something like:

pseudo_phen score !WT no_obs ~ Trait + Tr.FIX !r Tr.animal !mv

In which '!WT no_obs' should apply to the trait pseudo_phen.

To my surprise the weight appears as one of the terms in the list of variance componenst estimates, but coded with 'S' indicating that no information was available to estimate.
The weighted analysis runs fine in an univariate model

Any help would be appreciated,
Kind regards, Bart Ducro
View user's profile Send private message
Arthur
Posted: Mon Apr 23, 2018 9:48 am Reply with quote
Joined: 05 Aug 2008 Posts: 471 Location: Orange, NSW
Dear Bart,

To have separate weights for each trait, you need to use
!ASUV and specify a second weight variable.

For example
!PART 3 # Separate weights ==> means 12.85 13.24
!ASUV
Pphen Score !wt=Pwt !wt=One ~ Trait
residual units.us(Trait)

The attached job has 6 different specifications but only 2 of them (parts 1 and 3)
weight the traits differently.



bvd.txt
 Description:
Data file

Download
 Filename:  bvd.txt
 Filesize:  79 Bytes
 Downloaded:  231 Time(s)


bd.as
 Description:
Command file

Download
 Filename:  bd.as
 Filesize:  997 Bytes
 Downloaded:  241 Time(s)


_________________
Arthur Gilmour

Retired Principal Research Scientist (Biometrics)
View user's profile Send private message Send e-mail Visit poster's website
bart.ducro
Posted: Wed May 02, 2018 6:53 am Reply with quote
Joined: 22 Feb 2008 Posts: 2
Dear Arthur,
Thank you very much for your reply. I added a weight=1 on the second trait (i.e. separate weights on the traits) as you suggested.
The analysis failed however, probably due to the data structure. Both traits are recorded on only a part of the animals, which means that there are quite some missing observations on the traits. I checked with your small examples if missing observations are problematic, but that seems not to be the case. Separate univariate analysis runs fine (i.e. without nonrecorded animals), but including nonrecorded animals (i.e. bivariate with diagonal variance structures for both genetic and residual variance components) results in singularities in the AI matrix and even the note that no variance is present. Am I looking for mission impossible?
Kind regards, Bart
View user's profile Send private message
Arthur
Posted: Wed May 02, 2018 10:14 pm Reply with quote
Joined: 05 Aug 2008 Posts: 471 Location: Orange, NSW
DearBart,

I can probably get the job to run, but I would need the data and command file in front of me.
If you like send send them to me (arthur@vsni.co.uk), I will have a look. Else at least send the .as file and the .asr file you get.

However, I am travelling over the next few weeks so I may not be able to respond as quickly, depending on when it
all arrives, and how hard it is.

_________________
Arthur Gilmour

Retired Principal Research Scientist (Biometrics)
View user's profile Send private message Send e-mail Visit poster's website

Display posts from previous:  

All times are GMT
Page 1 of 1
Post new topic

Jump to:  

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You can attach files in this forum
You can download files in this forum