Hi all,
So this is a question about running FAST through matlab parallel desktop: I know it is very funny (and inefficient) because I open up a bash in the matlab to run a csh script that generates and runs a matlab code, but just I’m familiar with matlab in doing many things including string manipulation, so that’s how I’m doing now. If there are matlab scripts that does a similar thing as csh scripts, it would be nice.
Anyway, I realized that I cannot run all the fitting with different parameters and with the identical name of model (the GLM directory name) at the same time because the code appears to refer to the directory at $SUBJECTS_DIR.
But I tried to run multiple subjects (with all different GLM names in their own native spaces) through MATLAB on a server, it returns (seems just one of random subject):
ERROR: could not read /tmp/tmp354512.nii rm: cannot remove ‘/tmp/tmp354512.nii’: No such file or directory ERROR: loading ${filename}.nii.gz as analyze
So I traced back to the functions that return those errors (i.e., load_nifti.m run in MRIread.m) and it seems to be something with the temporary directory managed by matlab workers. The temporary file seems to be only necessary for nii.gz file.
If so, I would rather just use .nii file format (or maybe .mgz?) format instead of dealing with the void complexity. Then can it be done by just putting:
setenv(‘FSF_OUTPUTFORMAT’,’.mgz’)
before I run "preproc-sess" and "selxavg3-sess” in my scripts?
Best regards,
+Self-update: I figured out “.nii” works well. I thought “.mgz" should work well, but "preproc-sess” tried to read “template.mgz.nii” while it creates “template.nii”. So just the simple format “.nii” works well even in a harsh virtual environment (e.g.,matlab>parallel-workers>bash>csh>matlab).
That should work (the env var is FSF_OUTPUT_FORMAT). You will also need to set it in your linux shell before running any of the scripts
On 02/03/2017 09:51 AM, Seung-Goo KIM wrote:
Hi all,
So this is a question about running FAST through matlab parallel desktop: I know it is very funny (and inefficient) because I open up a bash in the matlab to run a csh script that generates and runs a matlab code, but just I’m familiar with matlab in doing many things including string manipulation, so that’s how I’m doing now. If there are matlab scripts that does a similar thing as csh scripts, it would be nice.
Anyway, I realized that I cannot run all the fitting with different parameters and with the identical name of model (the GLM directory name) at the same time because the code appears to refer to the directory at $SUBJECTS_DIR.
But I tried to run multiple subjects (with all different GLM names in their own native spaces) through MATLAB on a server, it returns (seems just one of random subject):
ERROR: could not read /tmp/tmp354512.nii rm: cannot remove ‘/tmp/tmp354512.nii’: No such file or directory ERROR: loading ${filename}.nii.gz as analyze
So I traced back to the functions that return those errors (i.e., load_nifti.m run in MRIread.m) and it seems to be something with the temporary directory managed by matlab workers. The temporary file seems to be only necessary for nii.gz file.
If so, I would rather just use .nii file format (or maybe .mgz?) format instead of dealing with the void complexity. Then can it be done by just putting:
setenv(‘FSF_OUTPUTFORMAT’,’.mgz’)
before I run "preproc-sess" and "selxavg3-sess” in my scripts?
Best regards,
Seung-Goo KIM
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Douglas N Greve -
Seung-Goo KIM