Dear FSL Gurus,
I'm a FSL novice and am attempting to use Randomise_parallel to run a TBSS model, but it seems that the Seed defragmentation is not working and the process will run for as long as I let it (weeks, even).
I'm running the latest version of FSL on a Mac Pro with 64GB RAM and dual 12-core processors. However, instead of SGE, I am using Mac OS XGRID with a modified FSL_SUB bash file <https://rcbi.rochester.edu/users/vanooste/weblog/f657b/fsl_sub_implemented_for_local_multiprocessor_systems.html> which could be a contributing factor. For example, that modified script seems to break bedpostx from running correctly.
So, I'm not sure if the problem is that the model is not set up correctly, or if it's due to some other issue with the parallel processing component.
As far as the model is concerned, I would greatly appreciate some assistance with reality checking. There are 118 subjects in 2 groups. I'm trying to get Randomise to model group differences as well as age as a regressor (already demeaned) and the interactions thereof. In addition, I'm unclear what the /PPheights command does in randomize for DW data and how to determine those values for the design.con and design.mat files. The documentation that I have found has only added to my confusion.
Randomise_parallel runs and generates 13 SEEDS rather quickly but doesn't seem to run the defragment. I read through some previous posts to the email list (link here: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;b9d28ec9.1003) that suggested making a couple of changes to the randomise_parallel script (e.g., changing #!/bin/sh to #!/bin/bash), and changing the fsl_sub command to veryshort.q instead of the default. However, because neither solution seems to address the problem, I'm not sure if it's a problem with the randomize model or due to the modified fsl_sub_bash file.
Below is the command syntax and design.mat and design.con files I've used:
Command: fsl_sub -q veryshort.q randomise_parallel -i /MY_FOLDER/mytbss/stats/all_FA_skeletonised -o tbss -m mean_FA_skeleton_mask -d design.mat -t design.con -n 500 -c 3 -V --T2
design.mat:
/NumWaves 5
/NumPoints 118
/PPheights 1 1
/Matrix
0.00 1.00 -29.12 -29.12 0.00
0.00 1.00 -12.12 -12.12 0.00
0.00 1.00 5.88 5.88 0.00
0.00 1.00 -0.12 -0.12 0.00
0.00 1.00 2.88 2.88 0.00
0.00 1.00 -2.12 -2.12 0.00
0.00 1.00 31.88 31.88 0.00
0.00 1.00 -2.12 -2.12 0.00
0.00 1.00 1.88 1.88 0.00
0.00 1.00 14.88 14.88 0.00
0.00 1.00 10.88 10.88 0.00
0.00 1.00 4.88 4.88 0.00
0.00 1.00 9.88 9.88 0.00
.
.
.
1.00 0.00 26.88 0.00 26.88
1.00 0.00 -13.12 0.00 -13.12
1.00 0.00 -13.12 0.00 -13.12
1.00 0.00 -8.12 0.00 -8.12
1.00 0.00 7.88 0.00 7.88
1.00 0.00 -19.12 0.00 -19.12
1.00 0.00 -32.12 0.00 -32.12
1.00 0.00 -26.12 0.00 -26.12
1.00 0.00 -28.12 0.00 -28.12
design.con:
/NumWaves 5
/NumContrasts 5
/PPheights 1 1 1 1 1
/Matrix
1 0 0 0 0
0 1 0 0 0
0 0 1 0 0
0 0 -1 0 0
0 0 0 1 -1
0 0 0 -1 1
Thank you for your expertise, patience, and assistance!
Andrew
********
Andrew R. Bender, M.A.
Graduate Research Assistant
Cognitive Neuroscience of Aging Lab
Department of Psychology
Institute of Gerontology
Wayne State University
Detroit, Michigan 48202
lab 313.664.2649
fax 313.664.2667
|