I solved it myself. I was specifying the F-contrasts wrong (for anyone
wondering: you need to specify the *.con file with -t and the *.fts file
for -f)
The only thing which remains is the warning about the "-cmcx" option. Can
this warning be ignored?
On Tue, Apr 16, 2019 at 5:42 PM Paul Zhutovsky <[log in to unmask]>
wrote:
> Dear Anderson Winkler,
>
> I am trying to run a 2x2 mixed-effects ANOVA (group x time (40 subjects))
> in PALM. Using the "design_ekaterina.ods" example I set up 2 corresponding
> design matrices (attached) with 2 t-contrast and 2 F-contrast files.
> Design and contrast files can be found in this dropbox folder:
> https://www.dropbox.com/sh/te4humtaiklijq5/AABcg2Jj0gfDSablekV0h8baa?dl=0
> I run PALM (a115) under Matlab2018b with FSL 6.0.1 by using the following
> command:
>
> palm -i subjects_data.nii ...
> -m mask.nii ...
> -d design_temporal.mat ...
> -t design_temporal_tcon.con ...
> -f design_temporal.con ...
> -d design_group.mat ...
> -t design_group_tcon.con ...
> -f design_group.con ...
> -eb EB.grp ...
> -n 10000 ...
> -T ...
> -whole ...
> -within ...
> -corrcon ...
> -logp ...
> -o palm_anova
>
> The first warning I received immediately is:
>
> Warning: You chose to correct over contrasts, or run NPC
> between contrasts, but with the design(s) and,
> contrasts given it is not possible to run
> synchronised permutations without ignoring repeated
> elements in the design matrix (or matrices). To
> solve this, adding the option "-cmcx" automatically.
>
> After running the first contrasts for the first design matrix I get the
> following errors:
> 0.00111% [Shuffling 1/10000, Design 1/2, Contrast 1/6, Modality 1/1]
> Saving file: palm_anova_vox_tstat_d1_c1
> 0.00222% [Shuffling 1/10000, Design 1/2, Contrast 2/6, Modality 1/1]
> Saving file: palm_anova_vox_tstat_d1_c2
> 0.00333% [Shuffling 1/10000, Design 1/2, Contrast 3/6, Modality 1/1]
> Saving file: palm_anova_vox_tstat_d1_c3
> 0.00444% [Shuffling 1/10000, Design 1/2, Contrast 4/6, Modality 1/1]
> Saving file: palm_anova_vox_tstat_d1_c4
> 0.00556% [Shuffling 1/10000, Design 1/2, Contrast 5/6, Modality 1/1]
>
> Warning: Rank deficient, rank = 42, tol = 1.962050e-13.
> > In palm_core (line 1417)
> In palm (line 81)
> In palm_run (line 9)
>
> Warning: Matrix is close to singular or badly scaled. Results may be
> inaccurate. RCOND =
> 1.945966e-18.
> > In palm_core>fastt (line 2749)
> In palm_core (line 1469)
> In palm (line 81)
> In palm_run (line 9)
>
> Error using *
> Incorrect dimensions for matrix multiplication. Check that the number of
> columns in the
> first matrix matches the number of rows in the second matrix. To perform
> elementwise
> multiplication, use '.*'.
>
> Error in palm_core>fastt (line 2749)
> den =
> sqrt(plm.mrdiv(plm.eC{y}{m}{c}{o}',(M'*M))*plm.eC{y}{m}{c}{o}*sum(res.^2)./df2);
>
> Error in palm_core (line 1469)
> [G{y}{m}{c},df2{y}{m}{c}] =
> fastpiv{m}{c}(M,psi,res,y,m,c,1,plm);
>
> Error in palm (line 81)
> palm_core(varargin{:});
>
> Error in palm_run (line 9)
> palm -i
> subjects_data.nii
> ...
>
> It seems to me that I specified the F-contrasts incorrectly. But I don't
> know how else I would specify them?
> I would be very glad for any help on the correct setup of this analysis!
>
> Cheers,
> Paul
>
>
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1
