[general] self.output_directory initialisation issue

[manuegrx]

As explain in #22 we need to add a sub-folder in derived_data for each subject for some brick (as in EstimateModel).

The idea was to redefine self.output_directory path in each brick.
For example in the EstmiateModel brick :

          if self.output_directory:
               # Change output_directory for this process in order to
               # use a specific directory for each analysis

               if os.path.dirname(
                   spm_mat_dir
               ) == self.output_directory and "data" in os.path.basename(
                   spm_mat_dir
               ):
                   # if spm_mat already in a subfolder for a analysis
                   out_directory = spm_mat_dir
               else:
                   # if spm_mat file not in a subfolder(for e.g spm_mat
                   # file in download data)
                   sub_name = get_dbFieldValue(
                       self.project, self.spm_mat_file, "PatientName"
                   )
                   if sub_name is None:
                       print(
                           "Please, fill 'PatientName' tag "
                           "in the database for SPM file"
                       )
                       return self.make_initResult()

                   out_directory = os.path.join(
                       self.output_directory, sub_name + "_data"
                   )

                   if not os.path.exists(out_directory):
                       os.mkdir(out_directory)
               self.output_directory = out_directory
           else:
               print("No output_directory was found...!\n")
               return self.make_initResult()

It is working correctly when using the brick alone or in a pipeline.

However, if we are using iteration there is an issue :

• at the begging of the brick for the second subject : self.output_directory = ".../data/derived_data/sub-01" and not self.output_directory = ".../data/derived_data/"
• the outputs are therefore created in ".../data/derived_data/sub-01/sub-02"

[manuegrx]

To quickly reproduce the issue, for example, in a new project:

• add two time the test subject alej, one time with the name 001 and one time with the name 002
• add (using the "add document" button) the regressor physio (regressor_physio_EtC02_ctl.mat)
• for each subject (001 and 002) run level1design brick (only filled the sess_scans parameter with the scan and the sess_mtlti_reg parameter with the regressor)
• For the estimateModel brick, click on "Iterate pipeline" and choose to iterate over spm_mat_file. Select both SPM.mat file created before and run the brick

(It will be better to test also the iteration with a pipeline with level1design, estimateModel and EstimateContrast brick but there is currently an issue to iterate Level1design brick)

[manuegrx]

The issue also appears when you stay on the same pipeline/ brick to do a second subject

To reproduce:

• add two time the test subject alej, one time with the name 001 and one time with the name 002
• add (using the "add document" button) the regressor physio (regressor_physio_EtC02_ctl.mat)
• for each subject (001 and 002) run level1design brick (only filled the sess_scans parameter with the scan and the sess_mtlti_reg parameter with the regressor)
• Run EstimateModel for subject 001
• Stay on the same pipeline and change the smp_mat_file for subject 002. Run the brick

[servoz]

To reproduce:

add two time the test subject alej, one time with the name 001 and one time with the name 002
add (using the "add document" button) the regressor physio (regressor_physio_EtC02_ctl.mat)
for each subject (001 and 002) run level1design brick (only filled the sess_scans parameter with the scan and the sess_mtlti_reg parameter with the regressor)
Run EstimateModel for subject 001
Stay on the same pipeline and change the smp_mat_file for subject 002. Run the brick

OK I reproduce ... I start to look for a fix ..

[servoz]

I made a fix for the last To reproduce case.

@manuegrx, can you please try on your side with different cases.

I also need to do some tests on my end.

[manuegrx]

It works on my side for both issues (it is okay when I stay on the same pipeline/ brick to do a second subject and it is okay when I iterated ) :)

Thanks !

[servoz]

Thank you.
As it's good for you, I'm going to close this ticket after a final survey.
I think we can do better. I need some time for a final fix and testing with bricks that are not an instance of ProcessMIA.