Re: radio or variance component for random effect?

From: luansheng <luansheng_at_GMAIL.COM>
Date: Mon, 29 Oct 2007 10:50:42 +0800

Thank you, doctor gilmour. I have understand this question from you mail.
Yours sincerely,

luan sheng
----------------------------------------
Yellow Sea Fisheries Research Institute
Qingdao 266071
China
luansheng_at_gmail.com
luansheng_at_ysfri.ac.cn
----------------------------------------

From: arthur.gilmour_at_DPI.NSW.GOV.AU
Sent: Monday, October 29, 2007 7:07 AM
To: ASREML-L_at_AGRIC.NSW.GOV.AU
Subject: Re: radio or variance component for random effect?

Dear luan sheng,

Whether the analysis is performed on variance ratios or components depends basically on the complexity.

The components are treated as ratios if it is a univariate analysis of a single experiment unless the user explicitly
defines a residual variance. When there are multiple residuals (as in multiple trait or multiple experiments),
the analysis is performed in terms of variances (not ratios).

The advantage of using ratios, is that the scale od the data does not upset the analysis.
When using the variance scale, the data should generally be scaled so that the variance is between 0.1 and 1000.
Sometimes convergence problems occur when the variances are extremely small or large.

You example would be analysed on the variance scale if you appended

1 1 0
0 0 IDV 10 !S2==1.0

where 10 becomes the explicit residual variance and 1.0 is the overal scaling.

This is discussed in the user guide.

NB There is some confusion in the syntax because in mult-site analyses
the !S2 qualifier has a different meaning, in that it sets the residual variance for the site.

May Jesus Christ be gracious to you,

Arthur Gilmour, His servant .

Mixed model regression mapping for QTL detection in experimental crosses. Computational Statistics and Data Analysis 51:3749-3764 now available at http://dx.doi.org/10.1016/j.csda.2006.12.031

Personal website: http://www.cargovale.com.au/

mailto:Arthur.Gilmour_at_dpi.nsw.gov.au, arthur_at_cargovale.com.au
Principal Research Scientist (Biometrics)
NSW Department of Primary Industries
Orange Agricultural Institute, Forest Rd, ORANGE, 2800, AUSTRALIA

fax: 02 6391 3899; 02 6391 3922 Australia +61
telephone work: 02 6391 3815; home: 02 6364 3288; mobile: 0438 251 426

ASREML 2 is now available from http://www.VSNi.co.uk/products/asreml
The ASReml discussion group is at ASREML-L_at_dpi.nsw.gov.au
To join it, mailto:arthur.gilmour_at_dpi.nsw.gov.au
Archives are at
https://gatekeeper.dpi.nsw.gov.au/Listserv/archives/asreml-l.html
Cookbook: http://uncronopio.org/ASReml

Proposed travel:
UK 28 Jan
USA 11-15 Feb
><><><><><><><><><><><><><><><><><><><><><><><>

luansheng <luansheng_at_GMAIL.COM>
Sent by: ASReml users discussion group <ASREML-L_at_AGRIC.NSW.GOV.AU>
28/10/2007 12:07 AM Please respond to
ASReml users discussion group <ASREML-L_at_AGRIC.NSW.GOV.AU>

To ASREML-L_at_AGRIC.NSW.GOV.AU
cc
Subject radio or variance component for random effect?

Dear friends:

I know from the userguide that the initial value (variance component or ratio) can be specified after the model term. I use the the following asreml file to estimate the predictions of random additive genetic effects (breeding value) . The number of 0.2628 and 0.0116 are the ratios (relative to the residual variance), respectively.
My question is that Asreml how to know these numbers are the ratios, but not the variance components?

Growth traits for chinese shrimp
animal !P
sire !P
dam !P
familynum 70 !I
tank 3
sex 2 !I
days
lg
wt
meanwt
wt2007.ped !MAKE
wt200712fullsib.txt !SKIP 1 !MAXIT 1

wt ~ mu tank sex days !r animal 0.2628 ide(dam) 0.0116

Yours sincerely,

luan sheng
----------------------------------------
Yellow Sea Fisheries Research Institute