Hi Anderson,
It worked!! I have a second question please:
I wanted to do pearson correlation between X and Y to check the strength
and the direction of the relationship between the following two list of
variables:
* X Y*
0.3766 0.418 0.6362 0.698 0.64 0.575 0.6559 0.596 0.6293 0.471 0.6766
0.462 0.2606 0.329 0.3875 0.658 0.7614 0.568 0.6844 0.515 0.5657 0.652
0.7142 0.561 0.7682 0.647 0.8661 0.816 0.7751 0.696
I included the previous variables in the following csv files : *x.csv* and
*y.csv* then I used the following PALM command line to correlate x and y : palm
-i x.csv -d y.csv
The out put of this command line are : palm_dat_tstat_cx.csv;
palm_dat_tstat_cy.csv; palm_dat_tstat_fwep_cx.csv;
palm_dat_tstat_fwep_cy.csv; palm_dat_tstat_uncp_cx.csv;
palm_dat_tstat_uncp_cy.csv
In general when I use pearson correlation for x and y I will have r and the
P value for r. My question is why in palm the output is for x and y in
other words why I have two P values one for x>y and one for y>x.
I am sorry but I was unable to figure out this point.
Thank you very much for your advices.
Antoine
On Thu, Mar 5, 2015 at 4:26 AM, Anderson M. Winkler <[log in to unmask]>
wrote:
> Hi Antoine,
>
> There are two problems here:
>
> - The option "-pearson" is just that, so no "-pearson demean". The option
> "-demean" is fine, though.
> - The input file must contain the response variable (like the Y in the
> GLM), and you must supply the explanatory variable(s) in another file with
> the option "-d design.csv" (or "-d design.mat"). This is the design matrix
> (or the X of the GLM).
>
> All the best,
>
> Anderson
>
>
> On 5 March 2015 at 00:35, Alshikho, Mohamad J., M.D. <[log in to unmask]>
> wrote:
>
>> Dear Anderson,
>> Thank you very much for your suggestion. PALM is great!!. Kindly I have
>> the following question:
>> If I want to correlate (pearson) to columns of continuous variables
>> included in the file "file.csv" I used the following command line : palm -i
>> file.csv -pearson -demean Then I got the following error message
>>
>> Running PALM using MATLAB with the following options:
>> -i filecsv
>> -demean
>> -pearson demean
>> Error using palm_takeargs (line 961)
>> Unknown option: "demean"
>>
>> Error in palm_backend (line 31)
>> [opts,plm] = palm_takeargs(varargin{:});
>>
>> Error in palm (line 80)
>> palm_backend(varargin{:});
>>
>> What is / are the reasons for this message
>>
>> Thank you
>> Antoine
>>
>>
>>
>> On Mon, Mar 2, 2015 at 2:58 AM, Anderson M. Winkler <
>> [log in to unmask]> wrote:
>>
>>> Hi Antoine,
>>>
>>> You can use fslmeants as indicated. It will produce one value per
>>> subject per mask, which can be analysed using other software (e.g. SPSS)
>>> or, if you want, use permutation methods as available in PALM (available
>>> here <https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/PALM>), as that will allow
>>> correction for multiple testing just like randomise, but for non-imaging
>>> data as this.
>>>
>>> All the best,
>>>
>>> Anderson
>>>
>>>
>>> On 28 February 2015 at 15:50, Antoine Nasimian <
>>> [log in to unmask]> wrote:
>>>
>>>> Dear FSL experts.
>>>> I did TBSS analysis (just I ran the 4 scripts, without Randomise) to
>>>> generate all_FA_skeletonized.
>>>> I am interested in calculating the mean FA in the cortico spinal tract.
>>>>
>>>> - Is is correct technically if I create a 4D mask for the corticospinal
>>>> tract(from all_FA_skeletonized depending on JUH atlas) then I include this
>>>> mask in the following command line:
>>>>
>>>> fslmeants -i all_FA_skeletonized -m my 4D mask
>>>>
>>>> - Can I include the output of the previous command line (FA values) in
>>>> a Bayesian analysis or I correct it for multiple comparisons using FDR or
>>>> Bonferroni methods?
>>>>
>>>> Thanks,
>>>> Antoine
>>>>
>>>>
>>>>
>>>
>>
>>
>> --
>> Please ignore typos. Sent from my phone
>>
>
>
--
Please ignore typos. Sent from my phone
