[WIP] refactor narrowing, add test, run narrow before copying

.pre-commit-config.yaml

@@ -4,7 +4,7 @@
 
 repos:
 - repo: https://github.com/pre-commit/pre-commit-hooks
-  rev: v4.4.0
+  rev: v6.0.0
   hooks:
   - id: trailing-whitespace
   - id: end-of-file-fixer
@@ -23,7 +23,7 @@ repos:
 
 
 - repo: https://github.com/hadolint/hadolint
-  rev: v2.12.1-beta
+  rev: v2.14.0
   hooks:
   - id: hadolint-docker
     name: Lint Dockerfiles

Dockerfile

@@ -38,7 +38,7 @@ RUN wget --progress=dot:giga -P /opt https://www.fil.ion.ucl.ac.uk/spm/download/
     ${SPM_EXEC} function exit
 
 # Configure SPM BIDS App entry point
-COPY run.sh spm_BIDS_App.m pipeline_participant.m pipeline_group.m /opt/spm${SPM_VERSION}/
+COPY run.sh spm_BIDS_App.m narrow_participants.m pipeline_participant.m pipeline_group.m /opt/spm${SPM_VERSION}/
 RUN chmod +x /opt/spm${SPM_VERSION}/run.sh && \
     chmod +x /opt/spm${SPM_VERSION}/spm${SPM_VERSION} && \
     chmod +x /opt/spm${SPM_VERSION}/run_spm12.sh

narrow_participants.m

@@ -0,0 +1,32 @@
+function BIDS = narrow_participants(BIDS, labels)
+% NARROW_PARTICIPANTS - restrict BIDS struct to only requested labels
+%
+
+% nothing to do if no labels to change?
+if isempty(labels)
+   return
+end
+
+idx = ismember({BIDS.subjects.name},labels);
+BIDS.subjects = BIDS.subjects(idx);
+
+if isempty(BIDS.participants)
+   warning('no participants information. missing participants.csv?')
+else
+   if ~ismember(fieldnames(BIDS.participants), 'participant_id')
+      error('participants.tsv does not contain a participant_id column')
+   end
+   idx = ismember(BIDS.participants.participant_id, labels);
+   for fn=fieldnames(BIDS.participants)'
+     replace = BIDS.participants.(char(fn));
+     % BIDS.participants.meta is 1x1. other fields are 1xN
+     if(length(replace) < length(idx))
+        warning('BIDS participants field "%s" too short: ID idx len %d > number of field values %d; ignored', ...
+           char(fn), length(idx), length(replace))
+        continue
+     end
+     BIDS.participants.(char(fn)) = replace(idx);
+   end
+end
+
+end

pipeline_participant.m

@@ -3,7 +3,7 @@
 %==========================================================================
 
 % Available variables: BIDS and BIDS_App
-
+global BIDS;
 %==========================================================================
 %-fMRI Preprocessing
 %==========================================================================

spm_BIDS_App.m

@@ -201,6 +201,9 @@
 spm('defaults','fmri');
 spm_jobman('initcfg');
 
+% if we used  --participant_label PARTICIPANT_LABEL [PARTICIPANT_LABEL ...]
+BIDS = narrow_participants(BIDS, BIDS_App.participants);
+
 %==========================================================================
 %-(Temporary) Copy of input data and uncompress image files
 %==========================================================================
@@ -253,17 +256,6 @@
     BIDS = spm_changepath(BIDS,'.nii.gz','.nii',false);
 end
 
-%-Simplify BIDS structure to only contain participants under study
-%--------------------------------------------------------------------------
-idx = ismember({BIDS.subjects.name},BIDS_App.participants);
-BIDS.subjects = BIDS.subjects(idx);
-if ~isempty(BIDS.participants)
-    idx = ismember(BIDS.participants.participant_id,BIDS_App.participants);
-    for fn=fieldnames(BIDS.participants)'
-        BIDS.participants.(char(fn)) = BIDS.participants.(char(fn))(idx);
-    end
-end
-
 %==========================================================================
 %-Analysis level: participant*
 %==========================================================================

test_narrow_participants.m

@@ -0,0 +1,30 @@
+%
+% script based testing for narrow_participants
+% participants.json adds a 'meta' field that can be avoided when selecting only some participants
+%
+
+bids_dir = [tempname '_bids_test'];
+try
+  % build minimal BIDS filesystem: participants.json and 1 anat file pair
+  system(['mkdir -p ' bids_dir '/sub-{a,b}/ses-{1,2}/anat/']);
+  system([   'touch ' bids_dir '/sub-{a,b}/ses-{1,2}/anat/T1.{json,nii.gz}']);
+  system(['echo -e ''participant_id\nsub-a\nsub-a\nsub-b\nsub-b'' > '  bids_dir '/participants.tsv']);
+  system(['echo ''{ "Name": "test", "BIDSVersion": "1.4.1"}'' > '  bids_dir '/dataset_description.json']);
+  system(['echo ''{ "age": { "Description": "xyz" } }'' > '  bids_dir '/participants.json']);
+
+  % get what we built -- matches expectations
+  BIDS = spm_BIDS(bids_dir);
+  assert(length(BIDS.participants.participant_id) == 4)
+
+  % narrowing works
+  BIDS_narrow = narrow_participants(BIDS, {'sub-a'})
+  assert(length(BIDS_narrow.participants.participant_id) == 2)
+
+  % no change on empty
+  BIDS_nochange = narrow_participants(BIDS, {}); % TODO: {{}}?
+  assert(length(BIDS_nochange.participants.participant_id) == 4)
+
+catch me
+     rmdir(bids_dir,'s')
+     rethrow(me)
+end