hyemin-han / bayesfactorfmri Goto Github PK

Hi Hyemin,

First, thanks for making this software available.
I have a question specifically about your implementation of the Westfall/Jeffrey's multiple-comparison correction procedure, described in your 2020 Cognitive Neuroscience paper (as implemented in correct_scale.R).
You've chosen to correct for the number of "conditions" (m) that would result in a number of pairwise comparisons (k) greater or equal to the number of voxelwise tests. In Westfall et al. (1997), as described in de Jong et al. (2019), this is justified by the fact that they're actually correcting across pairwise comparisons of levels within a factor. But in the case of mass-univariate tests, how do we justify this? In other words, how do we know that the dependence structure of the data across voxels is analogous to that of all pairwise comparisons of levels within a factor? Shouldn't you be using the number of voxels instead of m, and do the rescaling using 1-alpha^(1/n_voxels), as in Jeffreys (1938), instead of (1 - alpha^(2/m)) as in Westfall et al. (1997)?
I guess this would make the correction too conservative?

Thanks!
Julien

Hi,
I was wondering if this program can be used to run 2nd level analyses on fmri data with unbalanced designs? In the tutorial, it seems that you make the contrast within subject, and then test if the value in each voxel is e.g. >0 at the group level.

However, I am in a situation in which each subject has not been exposed to all the conditions. I have 34 Ss for Condition 1, and 50 for Condition 2. So, I cannot make a within subject contrast and follow the same procedure described in the tutorial with my own data, it seems.

So, I think that what I would need is being able to feed the data for Condition 1 and for Condition 2, and contrast them at the 2nd level between subject. Do you think that it would be possible to conduct such an analysis with BayesFactorFMRI?

Hello! I'm very interested in using this package. However, I have troubles getting it to run. I installed the python dependencies in a conda environment and tried to run bayes_select_ui.py from this environment in the anaconda prompt. The gui appears and i'm asked to choose the task to do next. I select bayesian correction. Then, I get this error message:

Python was not found; run without arguments to install from the Microsoft Store, or disable this shortcut from Settings > Manage App Execution Aliases.

Then I tried to remove python from the "app execution aliases". The gui opens as before, but I get a new error message:

Traceback (most recent call last):
File "C:\Users\allst\BayesFactorFMRI-master\V1.0.0\bayes_select_ui.py", line 89, in
bayes_select_ui()
File "C:\Users\allst\BayesFactorFMRI-master\V1.0.0\bayes_select_ui.py", line 37, in bayes_select_ui
at_the_end(nexts)
File "C:\Users\allst\BayesFactorFMRI-master\V1.0.0\bayes_select_ui.py", line 50, in at_the_end
subprocess.call(["python3", "bayes_correction_ui.py"], cwd="Correction_UI")
File "C:\Users\allst\AppData\Local\Programs\Python\Python310\lib\subprocess.py", line 345, in call
with Popen(*popenargs, **kwargs) as p:
File "C:\Users\allst\AppData\Local\Programs\Python\Python310\lib\subprocess.py", line 971, in init
self._execute_child(args, executable, preexec_fn, close_fds,
File "C:\Users\allst\AppData\Local\Programs\Python\Python310\lib\subprocess.py", line 1440, in _execute_child
hp, ht, pid, tid = _winapi.CreateProcess(executable, args,
FileNotFoundError: [WinError 2] The system cannot find the file specified

Any thought on how to solve this?