[pymvpa] Loading multiple ROIs with fmri_dataset

Lyam Bailey Lyam.Bailey at Dal.Ca
Wed Aug 14 16:39:11 BST 2019

Dear Yaroslav

Woops, that was embarrassing! Thanks for clearing that up. However, I've just spotted another potential problem in my code. I assumed that the experimental design was the same for all runs, because only task001_run001 (in the sub001/BOLD directory) has an attributes.txt file indicating the order of stimulus presentations (unless I'm very much mistaken?) None of the other runs contain this file, so I used...

targets = SampleAttributes(pjoin('haxby2001, 'sub001', 'BOLD','task001_run001', 'attributes.txt'))['targets']

for every chunk in ds. However, when examining the data loaded in by load_tutorial_data...

print ds.targets[ds.chunks==0]
> ['scissors' 'scissors' 'scissors'...
print ds.targets[ds.chunks==1]
> ['face' 'face' 'face'....

SO I was clearly wrong on that front. Is there a way to access information about the experimental design from the other runs? As I mentioned above, only task001_run001 contains the relevant attributes.txt file, so I'm rather at a loss as to where the necessary info might be located.


Lyam Bailey, B.Sc., M.Sc.

Doctoral Student

Department of Psychology & Neuroscience
Dalhousie University

From: Pkg-ExpPsy-PyMVPA <pkg-exppsy-pymvpa-bounces+lyam.bailey=dal.ca at alioth-lists.debian.net> on behalf of Yaroslav O Halchenko <yoh at onerussian.com>
Sent: Wednesday, August 14, 2019 12:25:19 PM
To: pkg-exppsy-pymvpa at alioth-lists.debian.net <pkg-exppsy-pymvpa at alioth-lists.debian.net>
Subject: Re: [pymvpa] Loading multiple ROIs with fmri_dataset

On Wed, 14 Aug 2019, Lyam Bailey wrote:

>    Dear Yaroslav,

>    Thanks so much for the quick reply. I tried implementing this with:

>    fa_fname = {a: "haxby2001/sub001/masks/25mm/%s.nii.gz" % a for a in ["vt,
>    hoc"]}
>    temp_ds = fmri_dataset(bold_fname, mask=None, targets = targets,
>    add_fa=fa_fname, chunks = int(i)-1)

>    The code ran without errors, however when I continued through the example
>    RSA script (http://www.pymvpa.org/examples/rsa_fmri.html) the
>    dissimilarity matrices appeared very different to those generated when
>    using load_tutorial_data. I experimented with
>    fa_fname = {a: "haxby2001/sub001/masks/25mm/%s.nii.gz" % a for a in
>    ["white"]}

>    ...and this turned out to generate exactly the same results. I also get
>    the same results with add_fa = None. So I get the feeling that add_fa is
>    not actually being implemented. Can you see a problem in the way I've
>    called fmri_dataset?

add_fa just adds those feature attributes.  You would still need to
subselect those features if you decide to work with specific ROIs.

for se3lection could do smth like

    roi_ds = temp_ds[:, temp_ds.fa.vt != 0]

or am I missing what you are missing? ;)

Yaroslav O. Halchenko
Center for Open Neuroscience     http://centerforopenneuroscience.org
Dartmouth College, 419 Moore Hall, Hinman Box 6207, Hanover, NH 03755
Phone: +1 (603) 646-9834                       Fax: +1 (603) 646-1419
WWW:   http://www.linkedin.com/in/yarik

Pkg-ExpPsy-PyMVPA mailing list
Pkg-ExpPsy-PyMVPA at alioth-lists.debian.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/pkg-exppsy-pymvpa/attachments/20190814/7e7318d6/attachment-0001.html>

More information about the Pkg-ExpPsy-PyMVPA mailing list