Hi Ana,

The rank-deficient messages have to do with the option "-demean". If you remove that option, the warnings will disappear for these designs.

The options you are using are fine. Consider replacing "-save1-p" for "-logp". Consider also "-nouncorrected -approx tail", as that will extend the p-values further into the tail of the permutation distribution, something helpful if the number of permutations is small (and you cal keep it in case you further increase to 5000 or 10000).

All the best,

Anderson

On Wed, 17 Oct 2018 at 11:13, Ana Coelho <[log in to unmask]> wrote:

Hi Anderson,

Thanks for the design matrices you sent. I checked it and expand it to 51 subjects that I have.
Then I ran palm with the following options:
-i FA_3t_skeletonised.nii
-d design_within.mat
-t design_within.con
-within
-d design_between.mat
-t design_between.con
-whole
-eb design_eb.grp
-n 500
-save1-p
-o test
-T
-tfce2d
-m mean_FA_skeleton_mask.nii
-corrcon
-demean

It gives me warnings that matrix is rank deficient for design1 (within).
This is the first time I am using palm for tbss data, so I am not sure of the options I used. Particularly, I don't know if I need the demean option and if there are other options missing. Also, I used 500 permutations just to test if the command worked.

I want to know what is causing the matrix being rank deficient and also if these options for palm are fine.
In attachment I send the design matrices and contrasts that I used.

Thanks in advance!
Best regards,
Ana