Hi Philipp,
I think the F contrast that you want is
1 -1 0
0 1 -1
to test for any differences between genotypes. The F contrast you
mention will show you all voxels having significantly positive GM for
any genotype which will be pretty much everywhere unless you also
included a constant in your model. Often, if you assume that your
genetic variation will have an approximately linear effect on GM volume
(i.e. you assume codominance related to this quantitative intermediate
phenotype), you can also use a covariate where you code homozygotes as 1
and 3 and heterozygotes as 2 and then use simple regression (with a -1
or +1 contrast). In this case you definitely need to include a constant
in the model.
Hope this helps
Andreas
Andreas Meyer-Lindenberg, M.D., Ph.D., M.Sc.
Investigator, Genes, Cognition and Psychosis Program
Unit for Systems Neuroscience in Psychiatry and Neuroimaging Core
Facility
10-3C101, 9000 Rockville Pike, Bethesda, MD 20892-1377, USA
phone: 301 4969672, fax: 301 4800169, email: [log in to unmask]
web: snp.nimh.nih.gov
-----Original Message-----
From: Philipp Saemann [mailto:[log in to unmask]]
Sent: Sunday, March 12, 2006 12:50 PM
To: [log in to unmask]
Subject: [SPM] VBM, polymprhism, one factor, three levels
Hello list,
what model would be suitable to investigate if a genetical polymorphism
with three levels is associated with altered grey matter volume?
Usually in VBM t-tests / two-group ANCOVAs are used.
In SPM5 would this be a one-factor, three-level analysis (plus three
covariates)?
What would the F-contrast be that asks if there is a significant
association between the genetical status and GM volume?
(regardless of the direction of the effect (i. e. more or less GM)?)
With the polymorphisms encoded in the first three colums I used an
F-contrast
1 0 0 ...
0 1 0 ...
0 0 1 ...
and find very strong effects. However, I am not sure if this contrast
makes sense.
The pairwise t-contrasts (A<B, B<A, A<C, C<A, B<C, C<B) calculated from
the same model also reveal effects which seem very weak as compared with
the F-contrast above.
thanks a lot for any help here,
Philipp
|