Re: Boundary problems

From: Josť Manuel Cano Arias <jose.canoarias_at_HELSINKI.FI>
Date: Thu, 3 Apr 2008 09:28:11 +0300

Dear all,

continuing with the problem of boundary presented by Tuomas and quoting A.
Gilmour's reply:

>it might be interesting
> to fit the equivalent Sire Dam model.
>
> pcsize ~ mu block ucsize !r sire dam
>
> This should give two components but with the dam component smaller than
> the sire component. The id variance from the animal model should be
> about 4 times the average of the sire and dam components from this model.

Souldn't be the sire component(genetic only)smaller than the dam effect
(genetic + common environment)?

More important, I have experienced similar problems when analysing data
obtained from the same breeding design: each father(sire) crossed to
several mothers(dams). The problem is that in many instances the Sire
variance seems to be higher than the Dam variance, which it shouldn't be
possible. I guess that this situation may affect the performance of the
animal model to estimate the Dam effect.

I understand that with this breeding design (each male crossed with several
females) the Dam component includes variance due to genetic dominance in
addition to maternal environment. The animal model as presented by Tuomas
is:

trait ~ mu fixedfactors !r id dam

As I understood, the ID estimates the additive genetic variance and the DAM
estimates common maternal environment and genetic dominance (since dams are
nested within sires).

The use of such model for this "sire(dam within sire)" design raises TWO
QUESTIONS to me:
1) Should the residual variance be modeled further? my concern is whether
it might be some structure within the residual variance (e.g. due to
genetic dominance) and, if not modelled, could cause some bias in the
estimation of the variance of the random effects (e.g. additive genetic
variance).
2) If the animal model (trait ~ mu fixedfactors !r id dam) is correct and
the DAM effect is in the boundary, can be interpreted that both common
maternal environment and genetic dominance are negligible for this trait?

thanks for your time,

Cano

--
J. M. Cano Arias, PhD.
Ecological Genetics Research Unit
Department of Bio- and Environmental Sciences
University of Helsinki
PO Box 65, FIN-00014, Finland
E-mail: jose.canoarias_at_helsinki.fi
http://www.helsinki.fi/people/jose.canoarias
Received on Sat Apr 03 2008 - 09:28:11 EST

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.)