info/scripts/configs/rules/rules.yaml

# To organize raw MRI data
source: ""
destination: ""
pattern: "sub-[0-9]+"
tag_rules:
  - name: subject
    pattern: "sub-\\d+(?:\\d{3})(\\d{6})"
    replace:
      col: subjectID
      with: D_number
      from: ""
  - name: session
    pattern: "sub-\\d+(\\d{3})(?:\\d{6})"
  - name: run
    value: "01"
  - name: suffix
    value: mri
  - name: sourcetype
    value: mri
  - name: extension
    value: dcm
---
# To organize raw EEG data
source: ""
destination: ""
pattern: "EEG[0-9_]+.mff"
tag_rules:
  - name: subject
    pattern: "EEG(?:.+)_(?:\\d{3})?(\\d{6})_"
    replace:
      col: subjectID
      with: D_number
      from: ""
  - name: session
    pattern: "EEG(?:.+)_(\\d{3})?(?:\\d{6})_"
    default: 101
  - name: task
    value: rest
  - name: run
    value: "01"
  - name: suffix
    value: eeg
  - name: sourcetype
    value: eeg
  - name: extension
    value: mff
---
# To organize raw Eye tracking data
source: ""
destination: ""
pattern: "[0-9]+.edf"
copy_fellows: True
tag_rules:
  - name: subject
    pattern: "(?:\\d{0,3})(\\d{6}).edf"
    replace:
      col: subjectID
      with: D_number
      from: ""    
  - name: session
    pattern: "(\\d{0,3})(?:\\d{6}).edf"
    length: 3
    iffy_prepend: 1
    default: 101
  - name: task
    pattern: "\\d+[-_]([\\w]+)"
    case: lower
  - name: run
    value: "01"
  - name: suffix
    value: eyetrack
  - name: sourcetype
    value: eyetrack
  - name: extension
    value: edf
---
# To organize raw fNIRS data
source: ""
destination: ""
pattern: "(\\d{6,9}(_\\d)?_[\\w ]+|\\d+-\\d+-\\d+_\\d+|EO|Resting_EO|Resting EO)"
overwrite: Yes
add:
  - path: "probeInfo.mat"
    position: content
  - path: "digpts.txt"
    position: content
tag_rules:
  - name: subject
    pattern: "(?:\\d{3})?(\\d{6})"
    replace:
      col: subjectID
      with: D_number
      from: ""
  - name: session
    pattern: "(\\d{3})?(?:\\d{6})"
    default: 101
  - name: run
    value: "01"
  - name: suffix
    value: nirs
  - name: task
    value: rest
  - name: sourcetype
    value: nirs
  - name: extension
    value: nirx
---
# To organize genome sequence aligned reads
source: ""
destination: ""
pattern: ".*bam"
tag_rules:
  - name: subject
    pattern: "Sample_([a-zA-Z0-9]+)_"
    pad:
      length: 5
      character: 0    
    replace:
      col: sampleID
      with: D_number
      from: ""
  - name: session
    pattern: "Sample_([a-zA-Z0-9]+)_"
    pad:
      length: 5
      character: 0    
    replace:
      col: sampleID
      with: sessionID
      from: ""
  - name: sample
    pattern: "Sample_([a-zA-Z0-9]+)_"
    pad:
      length: 5
      character: 0    
  - name: acquisition
    value: seq
  - name: suffix
    value: exome
  - name: extension
    value: bam