| T O P I C R E V I E W |
| lh115 |
Posted - May 06 2010 : 11:01:20 AM
Hi,
So I'm trying to run a higher level analysis with 14 subjects. I've set up the level 3 .fsf template in FEAT the same way I have in the past. When I submit the job to the cluster, it does not complete. There are no errors in the cluster log or in the feat report log. The feat log just stops at:
/usr/local/packages/fsl-4.1.5/bin/fslmaths mean_func -Tmean mean_func
in older, complete L3 analysis, feat would continue past this fslmaths step.
The strange thing is that this L3 analysis will complete with a slightly older .fsf template set up the SAME way in the feat GUI. Set up in the GUI, they looked exactly the same, but when I ran the linux 'diff' command there were some small differences (the top is the newly created fsf and the bottom is the old):
248c248
< set fmri(regstandard) "/usr/local/packages/fsl-4.1.5/data/standard/MNI152_T1_2mm_brain"
---
> set fmri(regstandard) "/usr/local/fsl/data/standard/MNI152_T1_2mm_brain"
269c269,272
< set fmri(ncopeinputs) 0
---
> set fmri(ncopeinputs) 1
>
> # Use lower-level cope 1 for higher-level analysis
> set fmri(copeinput.1) 1
480c483
< set fmri(overwrite_yn) 0
---
> set fmri(overwrite_yn) 0
I guess I'm just confused why the two have the same setup in the GUI but only the older works. does anyone know what the ncopeinputs selection is?
|
| 5 L A T E S T R E P L I E S (Newest First) |
| lh115 |
Posted - May 07 2010 : 09:57:38 AM
Syam, the problem was caused when i didn't load an existing feat directory. I guess its just one of those quirky FSL things. Thanks for the help. |
| petty |
Posted - May 06 2010 : 4:25:47 PM
also for it to automagically work all your subjects have to have the same number/order of copes in their second level.
something with your subject (or subjects) is messed up since feat isn't detecting your 2nd level copes.
|
| syam.gadde |
Posted - May 06 2010 : 4:00:33 PM
I believe that parameter (ncopeinputs) is set automatically when you choose the first input directory for higher-level stats. If you didn't explicitly choose an existing input directory in the GUI (even though you may be changing it later), it will stay at 0. Maybe that's what happened? |
| lh115 |
Posted - May 06 2010 : 3:53:41 PM
Chris,
I could see why having that set to 0 would create problems. But why isn't that an option in the GUI? If the default is 0, then wouldn't it not run for anyone?
I've also opened the fsf with a text editor and set this parameter to 1 and it still doesn't run... |
| petty |
Posted - May 06 2010 : 2:55:47 PM
ncopeinputs is "# Number of lower-level copes feeding into higher-level analysis"
basically when you are trying to do your 3rd level, its how many copes from each subject are being fed into the final results.
the fact that yours is 0, would mean that nothing is being sent to the higher level analysis. the fsf with 1 is at least passing one cope per subject.
|
