Configuring XCP_D processors for DAX/XNAT

Resource requirements

requirements:
  walltime: 1-0
  memory: 8000

One CPU for one day with 8 GB of memory probably suffices. These might need to be increased for high resolution data (< 2.5mm fmri voxel size) or large number of connectivity map outputs.

Inputs

inputs:
  xnat:  
    scans:
      - name: scan_fmri
        types: REST
    assessors:
      - name: assr_fmriprep
        proctypes: fmriprep-ABIDE_v23

Verify that all fmri scan types are listed, and the appropriate fmriprep proctype is set.

XCP_D command line flags in main command

Verify that the XCP_D options suit the needs of the project. Reference is here: https://xcp-d.readthedocs.io/en/latest/usage.html

Some commonly changed options are

--atlases Glasser
--nuisance-regressors acompcor
--fd-thresh 0
--lower-bpf 0.01
--upper-bpf 0.10
--min-coverage 0.5

Post processing: Standard

stats_csvs.py reformats QC stats to CSV format suitable for REDCap sync.

fisher_z.py computes Fisher Z score matrices from the Pearson correlation matrices.

Post processing: Non-standard atlas

custom_parcellation.py is used if the desired atlas is not one of the XCP_D standards. Check the correct --space is set (must be present in fmriprep, typically MNI152NLin6Asym or MNI152NLin2009cAsym). The atlas must be present and correctly configured in this repo.

custom_parcellation.py
  --fmriprep_dir /INPUTS/fmriprepBIDS/fmriprepBIDS
  --xcpd_dir /OUTPUTS/xcpdBIDS
  --space MNI152NLin6Asym
  --atlas BNST
  --min_coverage 0.5
  --out_dir /OUTPUTS

Post processing: Connectivity maps

connectivity_maps.py is used to generate connectivity maps. Do not generate maps that are not genuinely needed - these use considerable disk space.

connectivity_maps.py
  --xcpd_dir /OUTPUTS/xcpdBIDS
  --space MNI152NLin6Asym
  --atlas BNST
  --saveR
  --saveZ
  --fwhm 0 4

Options are

--saveR         creates Pearson correlation maps
--saveA         creates Fisher Z maps
--fwhm          list of smoothing kernels to apply to maps
--seeds         list of seed regions to create maps for (all, if not given)