Hi Shannon - Which version of FSL do you have? Older versions didn't use to save the mean_dsamples.nii.gz, but that was a while ago.
How quickly it runs depends on the size of the brain mask, so I'd check that that's ok too.
a.y
On Wed, 9 Nov 2011, Shannon Buckley wrote:
Hi Anastasia,
It turns out that I was NOT able to run the original bedpostx on our cluster, so thank you for pointing me in the right direction! In case other users experience a similar problem, this is what we did to get the ball rolling on our grid:
We found the "fsl_sub_seychelles" and "fsl_sub" scripts where the default q-names were identified. We modified the files (changing long and short.q to reflect the naming convention used here) and ran the original bedpostx. The preproc and datacheck functions worked perfectly. The bedpostx function SEEMED to have launched fine, but we hit another snag...
The program finished far too quickly for it to have worked properly (an hour, tops), and when I try to run the next "-prep" step I get the following error, which confirms that suspicion:
_+_+_+_+_+_+_+_+_+_+_ "Loading BEDPOST parameter samples from /cind/07/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dmri.bedpostX niiRead(): error opening file /cind/07/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dmri.bedpostX/mean_dsamples.nii.gz ERROR: Could not read /cind/07/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dmri.bedpostX/mean_dsamples.nii.gz Linux paris 2.6.18-194.11.1.el5 #1 SMP Tue Aug 10 19:05:06 EDT 2010 x86_64 x86_64 x86_64 GNU/Linux
trac-paths exited with ERRORS at Wed Nov 9 16:18:53 PST 2011
Word too long." +_+_+_+_+_+_+_+_+_+_+_
I have to confess I'm not familiar with the "word too long" error. In any case, bedpostx seems to fail to produce the requisite files for -prep (mean_dsamples.nii.gz). Since I'm not sure what the inputs would be for creating such a file, I'm including my "ls" output for our dmri.bedpostX path.
_+_+_+_+_+_+_+_+_+_+_+
"[vhasfcbuckls@paris dmri.bedpostX]$ ls bvals dyads2.nii.gz mean_ph2samples.nii.gz merged_ph2samples.nii.gz bvecs error.log mean_th1samples.nii.gz merged_th1samples.nii.gz cancel logs mean_th2samples.nii.gz merged_th2samples.nii.gz commands.txt mean_f1samples.nii.gz merged_f1samples.nii.gz monitor dmri.bedpostX.nii.gz mean_f2samples.nii.gz merged_f2samples.nii.gz nodif_brain_mask.nii.gz dyads1.nii.gz mean_ph1samples.nii.gz merged_ph1samples.nii.gz xfms [vhasfcbuckls@paris dmri.bedpostX]$" +_+_+_+_+_+_+_+_+_+_
Do you have any advice for getting passed this step? We are fairly in the dark with this function over here, so any help / hints you can provide would be wonderful. Please let me know if you need more information.
Thanks again for your help and patience,
~Shannon B. -----Original Message----- From: Anastasia Yendiki [mailto:ayendiki@nmr.mgh.harvard.edu] Sent: Thursday, November 03, 2011 7:39 AM To: Shannon.Buckley@radiology.ucsf.edu Cc: freesurfer@nmr.mgh.harvard.edu; Mensch, Henry Subject: Re: [Freesurfer] Tracula Question, step 2
Hi Shannon - Does the original bedpostx run fine on your cluster? You'd probably have to edit that to get it to work, it's not made to work on every possible cluster. Bedpostx_seychelles is the modified version that tracula calls that I've modified from the original FSL version so that it runs on our clusters.
In either case, trac-all is meant to always be run from the command line, whether you're on a personal computer (where is runs jobs serially) or a cluster (where it submits them as jobs using qsub). But probably some editing will be needed as I said.
Hope this helps, a.y
On Wed, 2 Nov 2011, Shannon Buckley wrote:
Hi Anastasia,
I am trying to run Tracula for the first time in our lab andI’m getting errors during step 2 that seem to involve how the program wants to be launched via grid. I’ve attached the configuration file I’m using and the error message is below...
$trac-all -bedp -c /cind/01/users/vhasfcbuckls/PPMI/Tracula.sh
Using Freesurfer version 5.1.
Subjects Directory is /cind/00/PPMI/Freesurfer/Tracula/
Sessions Directory is /home/vhasfcbuckls/sessions
INFO: SUBJECTS_DIR is /cind/07/PPMI/Freesurfer/subjects
INFO: Diffusion root is /cind/00/PPMI/Freesurfer/Tracula
Actual FREESURFER_HOME /cind/01/apps/freesurfer/64bit/freesurfer5.1
ln -sf/cind/00/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dlabel/diff/anat_brain_ mask.bbr.nii.gz/cind/00/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dmri/nodif_brain_mask.n ii.gz
ln -sf /cind/00/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dmri/dwi.nii.gz /cind/00/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dmri/data.nii.gz
WARN: Running FSL's bedbost locally - this might take a while
WARN: It is recommended to run this step on a cluster
bedpostx_seychelles /cind/00/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dmri
subjectdir is /cind/00/PPMI/Freesurfer/Tracula/PPMI_088_S_3551m00/dmri
Making bedpostx directory structure
Queuing preprocessing stages
Unable to run job: Job was rejected because job requests unknown queue "short.q".
Exiting.
Queuing parallel processing stage
Unable to run job: Job was rejected because job requests unknown queue "long.q".
Exiting.
Queuing post processing stage
Unable to run job: denied: "60" is not a valid object name (cannot start with a digit)
Job was rejected because job requests unknown queue "long.q".
Exiting.
Although we DO have a grid, I am trying to test the program for feasibility purposes, just on my own system’s terminal, since we aren’t sure how results will look with our specific data set (3.0 Tesla Siemens, 64 directional DWI @ b=1000, with 1 b0 image). We are very interested in performing tractography in T1-space… I am open to running this over the grid, if that’s really the best way to do it, but the problem is that I don’t see anywhere to specify the name of the queue that should be used for launching the program.
Can you please tell me how to launch this without using the so-called long.q and short.q? We typically launch jobs to our grid using our “bash.linux.q”, but I haven’t been able to locate a place in the configuration file, or any switches to use in the cmd line, to specify that this queue should be used. I tried specifically launching Tracula’s step 2 onto our bash.linux.q via the cmd line using “qsub”, but received the same old error I had been getting. I checked the data independently using “bedpostx_preproc.sh”, and it appears that all the required data are set up properly. As far as I can tell, the only issue we have is this inability to use the long.q and short.q…
What do you think?
I’ve cc’d our IT specialist, Henry Mensch, who oversees our grid system in case you need someone to chime-in with any specific information about how our Linux systems are set-up.
Thanks very much for your time,
Shannon T. Buckley SRA III - Supervisor CIND, UCSF 4150 Clement St. (114M) San Francisco, CA 94121 Phone: (415) 221-4810 x4251 Fax: (415) 386-3954
The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.