How to set --filter-preproc_bold in denoise-fmri (CBS server)

, , ,
#1

I’m currently using the bids app denoise-fmri on the CBS server. My data has multiple tasks in the func folder which is throwing the error:

ConfigError in file /cifs/open_data_hbn/denoise-fmri/denoise_fmri/workflow/Snakefile, line 18:
More than one snakemake filename for preproc_bold, taking the first. To correct this, use the --filter_preproc_bold option to narrow the search.

How can I set this option “filter_preproc_bold” to specify using only one of the tasks/spaces? My code looks something like the following:

/directory/denoise-fmri/denoise_fmri/run.py /directory/bids_data /directory/denoised_data participant -np

With the filter:

/directory/denoise-fmri/denoise_fmri/run.py /directory/bids_data /directory/denoised_data --filter_preproc_bold task-taskname_space-MNI_res-01_desc-preproc_bold.nii.gz participant -np

Which throws a value error. I’m not sure exactly how or where to invoke the argument and can’t seem to find much information online. Thanks in advance!

#2

I’m not too familiar with the bids app itself, but I believe you should be able to filter by using the flag similar to: --filter_preproc_bold task=<task_name>, replacing <task_name> with your task.

#3

Thanks for the reply! It still doesn’t seem to be working. It’s odd because it seems to recognize the filters but can’t find the data with those labels:

Config file config/snakebids.yml is extended by additional config specified via the command line.
No input files found for snakebids component preproc_bold:
filters:
suffix: bold
desc: preproc
extension: .nii.gz
datatype: func
task: movieTP
space: MNI152NLin2009cAsym_res-02

I have a somewhat inefficient workaround for now but just wondering if there’s a solution.

#4

Would you be able to provide an example of the movie task filename and the directory it is in (e.g. is it in the /directory/bids_data/ folder)?

From the filters you showed above, I am guessing it looks something like
/directory/bids_data/sub-{subject}/func/sub-{subject}_space-MNI152NLin2009cAsym_res-02_task-movieTP_desc-preproc_bold.nii.gz?

#5

It looks like:

/cifs/open_data_hbn/data/student_data/student_bids/sub-NRADA709231/sub-NRADA709231_task-movieTP_space-MNI152NLin2009cAsym_res-02_desc-preproc_bold.nii.gz

And the denoise-fmri program is in cifs/open_data_hbn

#6

Would you also be able to provide a portion (see below) of the config file for the bids app - it should be the /cifs/open_data_hbn/denoise_fmri/denoise_fmri/config/snakebids.yml file.

There should be a section in that file similar to:

pybids_inputs:
  preproc_bold:
    filters:
      suffix: 'bold'
      desc: 'preproc'
      extension: '.nii.gz'
      scope: 'derivatives'
      invalid_filters: 'allow'
#7

This is what I have:

pybids_inputs:
  preproc_bold:
    filters:
      suffix: bold
      desc: preproc
      extension: .nii.gz
      # scope: derivatives    # raw, derivatives (or can be pipeline name too!)
      # invalid_filters: allow
      datatype: func
      # run: true
    wildcards:
      - subject
      - session
      - task
      - acquisition
      - run
      - space
      - res
#8

Ahh okay, I am not sure if this was a typo, but this may be the cause of the issue. The filters are looking for a file in a sub-directory named func (as suggested by the datatype: func in the filters). You can try to either comment out the datatype filter by adding a # in front to that line (similar to scope, invalid_filters, etc.), though I am not sure if this will work.

Alternatively, you can move the fMRI scans to a folder named “func” within the subject folder (e.g.
/cifs/open_data_hbn/data/student_data/student_bids/sub-NRADA709231/func/sub-NRADA709231_task-movieTP_space-MNI152NLin2009cAsym_res-02_desc-preproc_bold.nii.gz).

#9

That was a typo, the data is in a func folder.

I think I figured it out though. I also had to specify the task and space in --filter_preproc_mask and --filter_confounds (task only) in addition to --filter_preproc_bold. Makes sense!

Thanks for your time @tkai

1 Like
#10

Hey there. Thanks for your input. I have tried that, but still facing some issues. Would you mind giving me a step-by-step guide on how to set the filter preproc_bold in the denoise fMRI CBS server? I am a bit stuck.

Thanks in advance.

#11

Hi @Domvat, this is the whole command I used to denoise all subject files in a folder:

for i in $(cat ./data/student_data/test.txt); do /cifs/open_data_hbn/denoise-fmri/denoise_fmri/run.py /cifs/open_data_hbn/data/student_data/student_bids/$i /cifs/open_data_hbn/data/student_data/test_results participant --fwhm 0 --filter_preproc_bold task=movieTP space=MNI152NLin2009cAsym --filter_preproc_mask task=movieTP space=MNI152NLin2009cAsym --filter_confounds task=movieTP -np; done

I found that in order to use filter_preproc_bold, I had to set the task AND the space AND also set filter_preproc_mask and filter_confounds (task only). Your data should be in BIDS format.

1 Like
#12

May be this is in BIDS format.
How can I change it?

#13

BIDS is a standardized way of organizing neuroimaging data (bids.neuroimaging.io). My data was already in BIDS so I do not have previous experience converting it. Maybe autobids can help you (https://github.com/khanlab/autobids)? There appears to be tutorials on the BIDS site as well.