Dear FS experts,

I am trying to run the cluster-wise correction by using permutation based multiple comparison. For this, as you would know, we use: mri_glmfit-sim

In attempt to this, I tried running different options

mri_glmfit --y lh.thickness.AT.10B.mgh --fsgd CT91_AT6.fsgd doss --C Contrast91.txt --surf fsaverage lh --cortex --glmdir glmdirlh_lat3.glmdir

mri_glmfit-sim --glmdir glmdir${i}_lat3.glmdir --cache 1.3 pos --cwp 0.05 --3spaces

Since, there was recommendations in this forum to use --perm for 1.3, I did so

mri_glmfit --y lh.thickness.AT.10B.mgh --X design_91subjs.txt --C Contrast_CT91.txt --surf fsaverage lh --cortex --glmdir glmdirlh_design.glmdir

mri_glmfit-sim --glmdir glmdir_design.glmdir --perm 1000 1.3 pos --cwp 0.05 --3spaces

However, it is displayed an error message 'design matrix is not orthogonal', though I used CT91_AT.fsgd and not the design matrix --X option.

Despite that, I tried again starting from design matrix with the mri_glmfit, it was ok. For the same I used, --perm for the mri-glmfit-sim. But, it was a same orthogonalisation error message.

I am also attaching my Group Desscriptor file: .fsgd, design matrix: GLM_all_Scan_Mim_91subjs.txt and the contrast.txt files for your reference.

It would be great, if anyone could please help sort out this issue. Thanks !

kind regards,
Varun

On Fri, Mar 13, 2020 at 10:40 PM Douglas N. Greve <dgreve@mgh.harvard.edu> wrote:

It looks like you are using gender as a continuous variable. You should set up two classes, one for male one for female. You can then test for the AT slope with a contrast of
0 0 0 0 0.5 0.5
you don't need separate pos and neg contrasts

For mri_glmfit-sim, a threshold of 1.3 (p<.05) is way too liberal for the MCZ simulation. I would use permutation. See https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/MultipleComparisonsV6.0Perm

On 3/12/2020 6:30 AM, Varun Chandran wrote:
        External Email - Use Caution

Dear FS experts,

I am looking at the relationship between the cortical thickness and a behaviour measure, while controlling for the effects of age and gender. In this analysis, I am using CT_AT.fsgd, contrast_pos.txt and a contrast_neg.txt.

GroupDescriptorFile 1
Title Autistictraits
Class AT
Variables Age Gender AT

Input Con_P001 AT -4.142 -0.428 21
Input Con_P002 AT -8.142 -0.428 31

My separate two contrast files:

Contrast_pos.txt 0 0 0 1



Contrast_neg.txt 0 0 0 -1

mri_glmfit --y lh.thickness.AT.10B.mgh --fsgd CT91_4.fsgd doss --C Contrast91_pos.txt --surf fsaverage lh --cortex --glmdir glmdirlhposlat.glmdir

mri_glmfit --y lh.thickness.AT.10B.mgh --fsgd CT91_4.fsgd doss --C Contrast91_neg.txt --surf fsaverage lh --cortex --glmdir glmdirlhneglat.glmdir

mri_glmfit-sim --glmdir glmdirlhposlat.glmdir --cache 1.3 pos --cwp 0.05

mri_glmfit-sim --glmdir glmdirlhneglat.glmdir --cache 1.3 neg --cwp 0.05

By changing the commands appropriately, I am looking at the right hemisphere too.

Can anyone please help me to clarify, if I am applying the command lines correctly? Any sort of help would be very much appreciated. Thanks !

best,

Varun Arunachalam Chandran
PhD scholar (Neurosciences)
School of Psychology and Clinical Language Sciences,
University of Reading, Harry Pitt Building,

Whiteknights campus, Reading-RG6 7BE,

Berkshire, United Kingdom.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer