Hi Gabriela,

Since your goal is to test for the main effects that are already in the contrasts.csv, yes, you can perform only the t-tests as  palm -i input.csv -d design.csv -t contrasts.csv -corrcon -n 100

As you are using a small number of permutations, you can also add '-accel tail' to get smaller p-values (there is some explanation about it on PALM User guide, but you can also check this post), because with 100 permutations, the smallest p-value you will get is 1/100 = 0.01. If you use the tail acceleration, you'll probably have to add the option '-nouncorrected', otherwise it will consume too much memory. When using the tail acceleration, the p-values can be very small and you might lose precision when saving them in the .csv file. Thus I recommend use the option '-logp' to save the values as -log10(p). All together the PALM command would be: palm -i input.csv -d design.csv -t contrasts.csv -corrcon -n 100 -logp -accel tail -nouncorrected

All the best,

Bianca

 

Em qui, 31 de out de 2019 às 13:01, Gabriela Dzięgiel <[log in to unmask]> escreveu:

Hi Bianca!

Thanks for your answer, my design and contrasts have the right dimensions. However PALM still does accept the input, giving me the same error message. When I don't add the -corrcon clause then the analysis is running, without any interesting output files though (only plam_elapsed.csv).

I have 120 subjects, so the data file has 120 rows. My design matrix has 120 rows and 8 columns, for 8 regressors. My contrast file looks like that:

 -1     1       -1      1       0       0       0       0
-1      -1      1       1       0       0       0       0
1       -1      -1      1       0       0       0       0

I have three contrasts: two for main effects and third for an interaction. The ftests file looks as follows:

1 0 0
0 1 0
0 0 1

Again, testing for main effects and interaction. If I don't add the ftests.csv and -fonly option than I do get some results. The t-test and f-test for testing for the main effect should be equivalent here, right? So maybe I don't need the f-test here and I shouldn't be adding them to the analysis?

All the best,
Gabriela

Od: Bianca Alberton
Wysłano: czwartek, 31 października 2019 15:16
Do: [log in to unmask]
Temat: Re: [FSL] PALM for several sMRI modalities and ANOVA

Hi Gabriela,

PALM is expecting files with the following dimensions:
input =>    nS x nV
design => nS x nR
contrast => nC x nR

where
nS = number of subjects
nV = number of image points
nR  = number of regressors in the design matrix

So, if you transpose your design matrix and contrast matrix, it should work. About the ftests.csv, it is also a contrast matrix, thus it has also to be a nR x nC matrix.

If you're following the GLM example in the FSl webpage, you can use the matrix
1 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0
0 0 1 0 0 0 0 0

If you're testing for differences between groups This older thread might help you setup your F-tests: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;3846fd1a.1910

All the best,

Bianca

Em qua, 30 de out de 2019 às 09:51, Gabriela Dzięgiel <[log in to unmask]> escreveu:
Hello!

Unfortunately, I am not an FSL user and so despite reading through PALM and FSL GLM Guide I don't know where I am going wrong. I want to condunct PALM analysis, correcting for multiple contrasts (the aim is also to control for several modalities but for now I'm just trying to make the toolbox work). I use the following command:

palm -i input.csv -d design.csv -t contrasts.csv -f ftests.csv -fonly -corrcon -n 100

and I get the following error message:

Reading input 1/1: input.csv
Reading design matrix and contrasts.
Index exceeds matrix dimensions.

Error in palm_takeargs (line 2308)
    rC1 = plm.rC{1}(1);

Error in palm_core (line 33)
[opts,plm] = palm_takeargs(varargin{:});

Error in palm (line 81)
palm_core(varargin{:});

I am really confused because my csv files seem to have the right dimensions (input: 120x163842; design: 120x8; contrasts: 3x8; ftests: 3x3). They are csv files separated with spaces.

Can anybody give me any clue on what is wrong with my command or with design, contrasts, etc. files? I would be very grateful!

Best regards,
Gabriela Dzięgiel-Fivet

########################################################################

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1


To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1

########################################################################

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1

---

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1