example illustrating data preprocessing of fMRIprep outputs for use i… #64
Conversation
There was a problem hiding this comment.
Quick question:
is this supposed to be
A) an example on a specific frmriprep dataset and its raw counterpart?
B) something more generic that could apply to any BIDS derivatives?
If A then this will work better if the dataset is openly accessible (openneuro?).
If B then the code needs to be made more generic.
Option A sounds way easier unless you feel like embarking on a journey to create a BIDS app.
| @@ -0,0 +1,187 @@ | |||
| #!/group/tuominen/anaconda3/bin/python3 | |||
There was a problem hiding this comment.
probably need to change this to make the code more portable
| import numpy as np | ||
| import nibabel as nib | ||
| import pandas as pd | ||
| from bids import BIDSLayout |
There was a problem hiding this comment.
I suspect you will need to add a requirements.txt to mention pybids need to be installed
| new_name['suffix'] = 'events_glmsingle' | ||
| pattern = 'sub-{}_ses-{}_task-{}_run-{}_{}.csv' | ||
| # Save the event file in derivatives, exclude TR column | ||
| new_design.to_csv(os.path.join(root, 'derivatives', 'iter4', | ||
| 'sub-'+new_name['subject'], 'ses-'+new_name['session'], 'func', | ||
| pattern.format(new_name['subject'], | ||
| new_name['session'], new_name['task'], new_name['run'], | ||
| new_name['suffix'])), columns=['CSplusshock','CSplus','CSminus'], | ||
| index=False, header=False) |
There was a problem hiding this comment.
You probably :
A) want to use pybids build_path methods (https://bids-standard.github.io/pybids/generated/bids.layout.writing.html#module-bids.layout.writing) that should happily accept the dict of entities you have just created.
B) want to give a BIDS friendly name to your output file: keep events as suffix but add a desc-glmsingle entity
| def smooth_bold(processed, smooth_num, subjects): | ||
| ''' Smooth data prior to first-level analysis using FSL in nipype. | ||
| Output will have suffix _smooth appended.''' | ||
| from nipype.interfaces.fsl import Smooth |
There was a problem hiding this comment.
maybe I missed it but you did not import nipype, right?
and I don't think it is a dependency of glmsingle so this should probably go into the requirements
also not a python expert, but do we have to go through nipype FSL interface to smooth the data? Doesn't that pre supposes that people have FSL installed on their system?
| gs = GLM_single() | ||
| # Name output folders | ||
| metadata = processed.parse_file_entities(bold_runs[0]) | ||
| outdir = os.path.join(os.path.dirname(bold_runs[0]), metadata['task'] + '_concat_glmsingle') |
There was a problem hiding this comment.
same comment as above to have BIDS friendly filenames
Grabs data using pybids
Makes glmsingle compatible event files from .tsv event files in BIDS format
Resamples bold timeseries
Smooths bold timeseries
Batches runs by subject and executes