Hi

You should double check but I don`t think you need an exchangeability block file (as you would if the case was repeated measures, twin design etc), otherwise the con and fts files should be fine.

I would recommend looking into the tfce --T2 option, sensitive to widespread signals of cluster-based thresholds that doesn`t require initial thresholding, in which case you do need the -n where the usual number of permutation for the null distribution varies depending on the quality of your data starting from -n 5000 per contrast.

What is the error message you are getting?

Best,

Rali

On 9 April 2013 17:35, Zhiliang Long <[log in to unmask]> wrote:

Hi FSL experts:
    I met problems when I conducted comparison analysis among three groups in TBSS.
    I have three groups, named HC (23 subjects), RP (17 subjects), LP (13 subjects). Now the null hypothesis is that are there any differences of FA skeleton between the three groups. The results were expected to be corrected for multi-comparison using Cluster-based thresholding. Then I called randomise as follow:

randomise -i all_FA_skeleton -o tbss -d design.mat -t design.con -f design.fts -m mean_FA_skeleton_mask -e design.grp -F 5

In this command, design files are set up as follows:

In design.mat file: the last two colums are nuisance variables, which would be regressed out in the analysis.
Ev1     Ev2     Ev3     Ev4(gender)     Ev5(age)
1       0       0       2       26
1       0       0       2       20
1       0       0       1       19
1       0       0       2       27
1       0       0       1       24
1       0       0       2       25
1       0       0       1       23
1       0       0       2       19
1       0       0       2       18
1       0       0       1       29
1       0       0       1       25
1       0       0       2       35
1       0       0       2       49
1       0       0       1       31
1       0       0       1       30
1       0       0       1       19
1       0       0       1       21
1       0       0       1       18
1       0       0       2       27
1       0       0       2       26
1       0       0       2       25
1       0       0       1       24
1       0       0       1       24
0       1       0       1       16
0       1       0       2       23
0       1       0       1       27
0       1       0       2       24
0       1       0       1       28
0       1       0       1       20
0       1       0       2       30
0       1       0       2       15
0       1       0       2       29
0       0       1       1       18
0       0       1       1       17
0       0       1       2       25
0       0       1       2       38
0       0       1       1       15
0       0       1       1       19
0       0       1       1       18
0       0       1       1       22
0       0       1       2       35
0       0       1       1       25
0       0       1       2       22
0       1       0       2       24
0       0       1       1       36
0       0       1       2       18
0       1       0       2       25
0       1       0       2       20
0       1       0       1       23
0       1       0       2       28
0       1       0       1       53
0       1       0       1       26
0       1       0       2       21

In the design.con file, the contrast matrix is :
1       -1      0       0       0
1       0       -1      0       0
0       1       -1      0       0

And the F contrast file (design.fts) is :
1 1 0

The design.grp file is :
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
2
3
3
2
2
2
2
2
2
2

Now, i have some questions about the process.
1. are these design files correct?
2. In the randomise command above, should I use -n option to set the numbers of permutation or does the command do it automatically?