Re: Maternal Genetic effects in ASReml-R

Hi,
Please run your model in ASREML not ASREML-R. If you get the same, then there is something not correct in your model and data. Also, from the information given, you should be able to run it as a 'fixed effects' model without relationships as this will give you some indication of dependencies.

You only have four offspring to estimate a genetic additive effect, an environmental effect, a maternal additive genetic effect and a maternal environmental effect from the same dam plus an estimate of the crossfostering effect. Thus you have very few degrees of freedom and power to estimate all of these terms.

Regards
Bruce

---- Original message ----
>Date: Sat, 2 Feb 2008 13:33:27 +0000
>From: Jarrod Hadfield <J.Hadfield_at_ED.AC.UK>
>Subject: Re: Maternal Genetic effects in ASReml-R
>To: ASREML-L_at_AGRIC.NSW.GOV.AU
>
> Dear Arthur/Bruce,
> Sorry, I should have made it clearer. The design I explained is only
> a single replicate within a larger experiment (each experiment has 100
> of these replicates). I am now certain that the problem lies with
> the model specification since the original model:
> asreml(y~1, random=~nurse+ped(id, var=T)+ped(nurse, var=T),
> ginverse(id=Gped, nurse=Gped))
> can actually be reparameterised as:
> asreml(y~1, random=~nurse+ped(id, var=T)+dnurse, ginverse(id=Gped))
> where dnurse is the dam identifier of the nurse. This gives unbiased
> variance estimates where
> nurse = Maternal environment + 0.5Maternal genetic
> and
> dnurse = 0.5Maternal genetic
> so I am confident the design is adequate. Is it possible that asreml-r
> is not associating the nurse identifiers with the G inverse? The
> estimates suggest that "ped(nurse, var=T)" is specifying the same
> structure as "nurse".
> Also, I cannot seem to get the link function working for this type of
> model - the manual says that the implementation of link is still not
> finished so this may be the problem?
> I realise that these models are straightforward in stand alone ASReml,
> but we are running a course on evolutionary quantitative genetics at
> the end of the month and we had hoped to do everything through the R
> interface, as it will be easier for the uninitiated.
> Thanks for the help,
> Jarrod
> On 1 Feb 2008, at 19:56, arthur.gilmour_at_DPI.NSW.GOV.AU wrote:
>
> Dear Jarrod,
>
> Let me add my thoughts.
>
> It would appear that the design is adequate, otherwise you would
> generate a singularity in the AI matrix.
>
> However, the data set is small so it is highly likely that sampling
> variation could generate the result you have observed.
> If you have generated 100 samples and they all show this feature,
> then maybe there is a problem with the simulation.
>
> Just check that you have the correct number of effects fitted.
>
> One check of coding is to export the data frame and run the job in
> standalone ASReml (which uses the same license).
>
> May Jesus Christ be gracious to you in 2008,
>
> Arthur Gilmour, His servant .
>
> Mixed model regression mapping for QTL detection in experimental
> crosses. Computational Statistics and Data Analysis 51:3749-3764
> at http://dx.doi.org/10.1016/j.csda.2006.12.031
>
> Profile: http://www.dpi.nsw.gov.au/reader/17263
> Personal website: http://www.cargovale.com.au/
> Skype: arthur.gilmour
> 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-8 Feb
> USA 11-15 Feb
> Argentina 16-19 September Biometrics .
> ><><><><><><><><><><><><><><><><><><><><><><><>
>
> Jarrod Hadfield To ASREML-L_at_AGRIC.NSW.GOV.AU
> <J.Hadfield_at_ED.AC.UK> cc
> Sent by: ASReml users Subject Re: Maternal Genetic effects
> discussion group in ASReml-R
> <ASREML-L_at_AGRIC.NSW.GOV.AU>
>
> 01/02/2008 06:52 AM
>
> +-----------------------------+
> | Please respond to |
> | ASReml users discussion |
> | group |
> | <ASREML-L_at_AGRIC.NSW.GOV.AU> |
> +-----------------------------+
>
> Dear Brian,
>
> Thanks for the quick reply. I haven't been able to get correlated
> random effects using "link" yet (I get various error messages), but
> I'm not sure I've specified ASReml-R to fit the simpler model
> correctly.
>
> I am using simulated data in order to test the power of different
> designs for estimating maternal genetic effects, and I'm reasonably
> confident that the design I have allows me to separate additive
> genetic, maternal additive genetic and maternal environment effects.
> However when I fit the model the maternal environment effect is
> constrained at zero. The estimate of the additive genetic variance
> seems to be unbiased but the estimate of the maternal genetic
> variance is equal to the total maternal variance (genetic +
> maternal):
>
> The model I fit is:
>
> asreml(y~1, random=~nurse+ped(id, var=T)+ped(nurse, var=T),
> ginverse(id=Gped, nurse=Gped))
>
> but the estimates suggest that nurse and ped(nurse, var=T) are
> fitting the same thing (allthough they are not).
>
> The replicates consist of 4 pairs of sisters mated to 8 unrelated
> males to form 8 parental pairs. Each of the 8 parental pairs then
> produce 4 offspring. The 8 parental pairs are then organised into 4
> dyads where the 2 pairs making up the dyad are unrelated to each
> other. 1/2 of the offspring for each pair are reared by the
> other parental parents within the dyad. Offspring are therefore
> raised by nurses, which in 50% of cases are also their dams. In
> addition the dyads are formed so that a sister pair are not paired
> with individuals that are also sisters. Hence, I'm reasonably
> confident the effects can be estimated since unrelated offspring are
> reared by sisters (maternal genetic effects), unrelated offspring
> are reared by the same nurse (maternal environmental effects) and
> related offspring are reared by their mothers and unrelated nurses
> (additive genetic effects).
>
> Perhaps I'm mistaken?
>
> Jarrod
>
> On 30 Jan 2008, at 17:20, brian.cullis_at_DPI.NSW.GOV.AU wrote:
>
> Dear jarrod
> I have to admit that I have never fitted these but I would imagine
> that you could as shown in the ASReml-R manual page 58 section 5.4.
> As for correlating additive and maternal I would think this is also
> possible using us(link(~ped(Calf) + ped(Dam))) but I dont think this
> has been tested or requested. Hence I am ccing this to David Butler,
> the author of ASReml-R to check the syntax and the use of ped()
> within link()
> Arthur and I are in the UK at the moment and so DBs response will be
> in Oz time
>
>
> warm regards
>
> Brian Cullis
> Research Leader, Biometrics &
> Senior Principal Research Scientist
> NSW Department of Primary Industries
> Wagga Wagga Agricultural Institute
>
> Professor,
> Faculty of Agriculture, Food & Natural Resources
> The University of Sydney
>
> Phone: 61 2 6938 1855
> Fax: 61 2 6938 1809
> Mobile: 0439 448 591
>
> Jarrod Hadfield To ASREML-L_at_AGRIC.NSW.GOV.AU
> <J.Hadfield_at_ED.AC.UK> cc
> Sent by: ASReml users discussion Subject Maternal Genetic effects
> group <ASREML-L_at_AGRIC.NSW.GOV.AU> in ASReml-R
>
> 31/01/2008 04:02 AM
>
> +--------------------------------+
> | Please respond to |
> | ASReml users discussion group |
> | <ASREML-L_at_AGRIC.NSW.GOV.AU> |
> +--------------------------------+
>
> Hi Everyone,
>
> Is it possible to fit maternal genetic effects in ASReml-R yet? If
> so does any one have any example code they could show me (preferably
> with the covariance between additive and maternal genetic effects
> estimated)?
>
> Thanks,
>
> Jarrod
Received on Tue Feb 02 2008 - 12:30:51 EST