In article <46hg2m$e4h at net.bio.net>, Jerry Walton,
</S=G.WALTON/OU1=S24L07A at mhs-fswa.attmail.com> wrote:
>>I would like advice on how to obtain the most (and no more)
>genetic information from a designed experiment. A description of the
>experiment and proposed formulation are attached.
[detailed description snipped out]
> Using C for the
> population mean, the model for the measurement taken of on one of M offspring
> from each mating can be written
>> y(i,j,k,l,m)=C+A(i)+B(j)+G(k)+S(k,l)+e(i,j,k,l,m)+
> +AB(ij)+AG(ik)+ABG(ijk)+AS(ikl)+BS(jkl)+ABS(ijkl)
>> where the first line contains the main effects and error term, the second all
> possible interactions, and only A and B are fixed effects.
...
> Since I will be asked to estimate heritabilty, and since the error terms are
> not likely to be "normally" distributed, there is the question of standard
> errors or confidence intervals for the variance and heritability estimates.
This is not an answer to your many detailed questions, but you ought to get
ahold of Ruth and Frank Shaw's maximum likelihood program for quantitative
genetic data, QUERCUS. It can cope with many situations. However, it does
assume multivariate normality. You can find it by anonymous ftp at
ftp.bio.indiana.edu in directory biology/quantgen/quercus.
-----
Joe Felsenstein joe at genetics.washington.edu (IP No. 128.95.12.41)
Dept. of Genetics, Univ. of Washington, Box 357360, Seattle, WA 98195-7360 USA
