I downloaded fsl 5.0.9 again, to make sure I had the latest fsl_sub
version. But the version that came with it is still missing the -R flag. I
don't know why this is the case, since at this point I've done both a
manual download and the automatic.
Dr. Orr, I wonder if you would mind sending me the code for the fsl_sub you
have? Not completely sure that will work embedded in the other commands,
but it seems like it's worth a shot.

Sarah


On Fri, May 6, 2016 at 1:18 PM, Joseph Orr <[email protected]> wrote:

> That's from the code package with fsl 5.0.9 (on a CentOS system at least).
> Perhaps your sysadmin copied fsl_sub from an old version to carry over
> modifications, e.g., modification for a non-SGE job manager. My fsl_sub is
> heavily modified for slurm, but the section that contains the fsl_sub
> options is from the original.
>
> --
> Joseph M. Orr, Ph.D.
> Assistant Professor
> Department of Psychology
> Texas A&M University
> College Station, TX
>
> On Fri, May 6, 2016 at 12:08 PM, Sarah Heilbronner <
> [email protected]> wrote:
>
>> Sounds right, since my fsl_sub is missing that part! I have fsl version
>> 5.0.9 running, with the fsl_sub that came with it. Where is your version of
>> fsl_sub from?
>>
>> On Fri, May 6, 2016 at 1:01 PM, Joseph Orr <[email protected]> wrote:
>>
>>> Sounds like you might be using an old version of fsl_sub. You should
>>> have a section in fsl_sub that starts with "$command V1.0beta" that has the
>>> following options for fsl-sub:
>>>
>>>   -T <minutes>          Estimated job length in minutes, used to
>>> auto-set queue name
>>>             This is helpful if you don't set a FSL default queue.
>>>   -q <queuename>        Name of FSL defualt queue possible values for
>>> <queuename>
>>>             are "verylong.q", "long.q" and "short.q". See below for
>>> details.
>>>   -Q <queuename>    Name of Wash U Queue to use, one of (dque, dque_HT,
>>> dque_smp)
>>>   -a <arch-name>        Architecture [e.g., darwin or lx24-amd64]
>>>   -p <job-priority>     Lower priority [0:-1024] default =
>>> 0
>>>   -M <email-address>    Who to email, default = none
>>>   -j <jid>              Place a hold on this task until job jid has
>>> completed
>>>   -t <filename>         Specify a task file of commands to execute in
>>> parallel
>>>   -N <jobname>          Specify jobname as it will appear on queue
>>>   -n <nCPUs>            Number of CPUs per node that job will use
>>> (default = 1)
>>>   -R <RAM>              Max total RAM to use for job (integer in MB)
>>>   -l <logdirname>       Where to output logfiles
>>>   -m <mailoptions>      Change the notification mail options, see qsub
>>> for details
>>>   -F                    Use flags embedded in scripts to set SGE queuing
>>> options
>>>   -v                    Verbose mode.
>>>
>>> -R is the MB of RAM you are allocating for the job. You may run in to
>>> problems with other FSL tools with large HCP datasets, basically for
>>> fslmerge and fslmaths commands you'll need to modify to allocate more RAM.
>>>
>>> --
>>> Joseph M. Orr, Ph.D.
>>> Assistant Professor
>>> Department of Psychology
>>> Texas A&M University
>>> College Station, TX
>>>
>>> On Fri, May 6, 2016 at 9:45 AM, Sarah Heilbronner <
>>> [email protected]> wrote:
>>>
>>>> Hi Stam,
>>>>
>>>> Thanks for thinking this through with me! I added what you suggested,
>>>> but I'm getting an error saying that -R is in invalid option. And when I
>>>> look inside fsl_sub, I don't see any -R flags. Any ideas?
>>>>
>>>> for reference, that line is now:
>>>> preprocid=`${FSLDIR}/bin/fsl_sub -T 60 -R 10000 -m as -N bpx_preproc -l
>>>> ${subjdir}.bedpostX/logs ${FSLDIR}/bin/bedpostx_preproc.sh ${subjdir}
>>>> ${gflag}`
>>>>
>>>> Thanks,
>>>> Sarah
>>>>
>>>>
>>>> On Fri, May 6, 2016 at 10:53 AM, Stamatios Sotiropoulos <
>>>> [email protected]> wrote:
>>>>
>>>>> Hi
>>>>>
>>>>> If the problem is in fslslice, this is likely a memory problem.
>>>>> Depending on how strict your queueing software is and what are the memory
>>>>> limits for each queue, it may kill or not the preproc job if it thinks it
>>>>> asks too much memory. Notice that the HCP data are an order of magnitude
>>>>> larger than conventional data and this may cause such problems. All you
>>>>> need to do is change the amount of memory requested when queuing
>>>>> bpx_preproc.
>>>>>
>>>>> In the $FSLDIR/bin/bedpostx script, the line to modify (e.g. by adding
>>>>> fsl_sub -T 60 -R 10000…) would be the following:
>>>>>
>>>>> preprocid=`${FSLDIR}/bin/fsl_sub -T 60 -m as -N bpx_preproc -l
>>>>> ${subjdir}.bedpostX/logs ${FSLDIR}/bin/bedpostx_preproc.sh ${subjdir}
>>>>> ${gflag}`
>>>>>
>>>>>
>>>>> By the way the “ -rician “ flag should be “ --rician"
>>>>>
>>>>> Hope this helps
>>>>> Stam
>>>>>
>>>>>
>>>>> On 6 May 2016, at 14:56, Sarah Heilbronner <
>>>>> [email protected]> wrote:
>>>>>
>>>>> Hello,
>>>>>
>>>>> I've been trying to run bedpostx on a single subject's HCP
>>>>> preprocessed diffusion data (I haven't made any changes from the initial
>>>>> download) on our cluster. I've run it a number of different ways, and
>>>>> always seem to get the same error:
>>>>> /software/fsl/5.0.7/fsl/bin/bedpostx_preproc.sh: line 77: 97699
>>>>> Killed
>>>>>  ${FSLDIR}/bin/fslslice ${subjdir}/data
>>>>>
>>>>> Seems to be encountering a problem with the fslslice line inside
>>>>> bedpostx_preproc.
>>>>>
>>>>> For reference, I'm using the latest version of FSL. I'm entering this
>>>>> command:
>>>>> bedpostx subject#/T1w/Diffusion -n 3 -model 2 -g -rician
>>>>>
>>>>> I've also tested some variations, like without grad_dev/g, etc, always
>>>>> getting the same error message. Any help?
>>>>>
>>>>> Thanks,
>>>>> Sarah Heilbronner
>>>>> University of Rochester
>>>>>
>>>>> _______________________________________________
>>>>> HCP-Users mailing list
>>>>> [email protected]
>>>>> http://lists.humanconnectome.org/mailman/listinfo/hcp-users
>>>>>
>>>>>
>>>>>
>>>> _______________________________________________
>>>> HCP-Users mailing list
>>>> [email protected]
>>>> http://lists.humanconnectome.org/mailman/listinfo/hcp-users
>>>>
>>>
>>>
>>
>

_______________________________________________
HCP-Users mailing list
[email protected]
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

Reply via email to