[VoxBo] correlation between regressors

[Ranjani Prabhakaran]

Hi Dan,

Thanks for your reply.  I tried using vbi, and I got the following error
message:

[E] vbi: file s1_collapsed.G doesn't match any known filetypes

I copied vbi into the following directory (and marked it executable):
/xjet/prabhakr/michigan_celeb_cfn/s1_cfn/models/s1_collapsed

If you could look into this, that would be great.  Thank you!

Ranjani



Quoting Daniel Y Kimberg <kimberg at mail.med.upenn.edu>:

> Ranjani Prabhakaran wrote:
> > Is there a way to check the correlation between two covariates in
> > VoxBo?
>
> There wasn't a simple way to do this, but now there is.  I've just put
> up a new version of vbi that puts up a quick and dirty correlation
> matrix for your G matrix.  Download the following file:
>
> ftp://voxbo.org/pub/voxbo/alacarte/linux-bleeding/vbi
>
> Mark it executable, and then do something like the following:
>
> vbi -l foo.G
>
> Let me know if it gives you any trouble.
>
> > And how does the information provided by checking "collinearity with
> > interest" differ from that provided in the .info file of a GLM?
>
> In the info file you're supposed to get the equivalent of
> "collinearity with all," not just vars of interest.  Eyeballing a few,
> the numbers are still a little off.  This might be because gdw does
> the calculation with the covariates before downsampling them to the
> resolution of your data.  Dongbo, can you verify that?
>
> dan
>