Dear Arthur,
Yes, under a random model, Genstat reports the "BLUP" of mean performance
(mu+effect) as BLUE(mu) + BLUP(effect). This is how Searle et al. (1992,
Variance Components, p 57, pp 269-272) also define/derive BLUP(mu+effect).
I will need to check how PREDICT works in ASReml.
Thanks and regards,
Subhash :: 26 Apr 2001
-----Original Message-----
From: Arthur Gilmour [mailto:gilmoua@agric.nsw.gov.au]
Sent: Thursday, April 26, 2001 11:59 AM
To: S.CHANDRA@CGIAR.ORG
Cc: asreml@chiswick.anprod.csiro.au
Subject: Re: BLUPs of mean performance of genotypes
Dear Subhash,
I thought Genstat just calculated mu + BLUP
I do not think 'BLUP of mean performance'is strictly defined.
The new PREDICT statement makes this easy to calculate in ASREML.
Arthur
> Date: Mon, 23 Apr 2001 02:25:38 -0700
> From: Chandra S <S.CHANDRA@CGIAR.ORG>
> Subject: BLUPs of mean performance of genotypes
> To: Arthur Gilmour <gilmoua@agric.nsw.gov.au>
> Cc: asreml@chiswick.anprod.csiro.au
>
> Dear Arthur,
>
> ASReml provides BLUPs of genotype effects. In a fixed effects model, a
> genotype's mean performance is simply the sum of the BLUE of its effect
and
> the GM. In random/mixed models, this simple sum is not the BLUP of a
> genotype's mean performance. It would perhaps good if, in your final
> commercial release of ASReml, you could also provide for directly
obtaining
> BLUPs of genotypes' mean performance, as in GenStat. I wonder whether all
> users of ASReml are aware of the way in which the BLUP of mean performance
> are to be obtained from the BLUP of effects!
>
> Best regards,
>
> Subhash :: 23 Apr 2001
--
Asreml mailinglist archive: http://www.chiswick.anprod.csiro.au/lists/asreml