Merge pull request #27 from cpp-lln-lab/meta

[ENH] improve Sources and RawSource metadata creation

Author: Remi-Gau

docs/source/conf.py

@@ -22,7 +22,8 @@
 author = "Rémi Gau"
 
 # The full version, including alpha/beta/rc tags
-release = "v0.2.0dev"
+with open('../../version.txt', encoding='utf-8') as version_file:
+    release = version_file.read()
 
 
 # -- General configuration ---------------------------------------------------

run_tests.m

@@ -4,12 +4,16 @@
 
 if isdir(fullfile(thisDir, 'lib', 'bids-matlab'))
     addpath(fullfile(thisDir, 'lib', 'bids-matlab'));
+end
+if isdir(fullfile(thisDir, 'lib', 'JSONio'))
     addpath(fullfile(thisDir, 'lib', 'JSONio'));
 end
 
 folderToCover = fullfile(thisDir, 'src');
 testFolder = fullfile(thisDir, 'tests');
 
+addpath(fullfile(testFolder, 'utils'));
+
 success = moxunit_runtests(testFolder, ...
                            '-verbose', '-recursive', '-with_coverage', ...
                            '-cover', folderToCover, ...

src/Mapping.m

@@ -28,9 +28,6 @@
     %
     % (C) Copyright 2021 spm_2_bids developers
 
-    % TODO add a print_mapping method to allow easy visualization
-    % of the input --> output relationship
-
     properties
 
         mapping

src/defaults/check_cfg.m

@@ -77,6 +77,7 @@
     % just to keep track of all the BIDS entities in the cfg
     bidsFile = bids.File('sub-01_T1.nii', 'use_schema', false);
     bidsFile = bidsFile.reorder_entities;
+    fields_to_set.schema = bids.Schema;
     fields_to_set.all_entity_order = bidsFile.entity_order;
 
     fields_to_set.entity_order = {'hemi'; ...

src/spm_2_bids.m

@@ -154,7 +154,7 @@
 
     content = json.content;
 
-    content.RawSources = identify_rawsources(file, verbose);
+    content.RawSources = identify_rawsources(file, map, verbose);
 
     Sources = identify_sources(file, map, verbose);
     if ~isempty(Sources)

src/utils/identify_rawsources.m

@@ -1,8 +1,10 @@
-function rawsource = identify_rawsources(derivatives, verbose)
+function rawsource = identify_rawsources(derivatives, map, verbose)
     %
     % find the most likely files in the raw dataset
     % that was used to create this derivatives
     %
+    % Assumption: the raw file was zipped
+    %
     % USAGE::
     %
     %   rawsource = identify_rawsources(derivatives)
@@ -41,10 +43,34 @@
         derivatives = strrep(derivatives, '_uw.mat', '.nii');
     end
 
+    % - only deal with files with official BIDS suffixes
+    % - remove prefix
+    % - remove eventual derivatives entities
+    % - change surface extension to a volume one
+    % - use only .nii.gz
     bf = bids.File(derivatives, 'verbose', verbose, 'use_schema', false);
 
+    if ~ismember(bf.suffix, fieldnames(map.cfg.schema.content.objects.suffixes))
+        rawsource{1} = 'TODO';
+        return
+    end
+
     bf.prefix = '';
 
+    entities = fieldnames(bf.entities);
+    idx = find(ismember(entities, map.cfg.entity_order));
+    for i = 1:numel(idx)
+        bf.entities.(entities{idx(i)}) = '';
+    end
+
+    if strcmp(bf.extension, '.surf.gii')
+        bf.extension = '.nii';
+    end
+
+    if strcmp(bf.extension, '.nii')
+        bf.extension = '.nii.gz';
+    end
+
     rawsource{1} = fullfile(bf.bids_path, bf.filename);
 
 end

src/utils/identify_sources.m

@@ -39,26 +39,6 @@
     add_deformation_field = false;
     deformation_field = 'TODO: add deformation field';
 
-    % TODO? grab all the anat suffixes from BIDS schema?
-    covered_suffixes = {'T1w', ...
-                        'T2w', ...
-                        'PDw', ...
-                        'T2starw', ...
-                        'inplaneT1', ...
-                        'inplaneT2', ...
-                        'PD', ...
-                        'PDT2', ...
-                        'T2star', ...
-                        'FLASH', ...
-                        'T1map', ...
-                        'T2map', ...
-                        'T2starmap', ...
-                        'R1map', ...
-                        'R2map', ...
-                        'R2starmap', ...
-                        'PDmap', ...
-                        'UNIT1'};
-
     args = inputParser;
 
     addOptional(args, 'derivatives', pwd, @ischar);
@@ -75,6 +55,7 @@
         return
     end
 
+    % deal with SPM's funky suffixes
     if endsWith(derivatives, '_seg8.mat')
 
         prefix_based = false;
@@ -91,6 +72,18 @@
 
     bf = bids.File(derivatives, 'verbose', verbose, 'use_schema', false);
 
+    if ~ismember(bf.suffix, fieldnames(map.cfg.schema.content.objects.suffixes))
+        sources{1} = 'TODO';
+        return
+    end
+
+    % deal with surface data
+    if strcmp(bf.extension, '.surf.gii')
+        bf.extension = '.nii';
+        prefix_based = false;
+    end
+
+    % anything prefix based
     if prefix_based
 
         if numel(bf.prefix) < 2
@@ -164,7 +157,7 @@
         % for anatomical data we assume that
         % the deformation field comes from the anatomical file itself
         if (~isempty(bf.modality) && ismember(bf.modality, {'anat'})) || ...
-           (~isempty(bf.suffix) && ismember(bf.suffix, covered_suffixes))
+           (~isempty(bf.suffix) && ~isempty(map.cfg.schema.find_suffix_group('anat', bf.suffix)))
 
             bf.prefix = 'y_';
             bf = bf.update;

tests/test_identify_rawsources.m

@@ -8,16 +8,72 @@
     initTestSuite;
 end
 
+function test_identify_rawsources_skip_unknown_suffix()
+
+    input_output = {'wsub-01_label-brain_mask.nii', ''};
+
+    verbose = false;
+
+    map = default_mapping();
+
+    for i = 1:size(input_output, 1)
+
+        rawsource = identify_rawsources(input_output{i, 1}, map, verbose);
+
+        assertEqual(rawsource, {'TODO'});
+
+    end
+
+end
+
+function test_identify_rawsources_surface()
+
+    input_output = {'c1sub-01_T1w.surf.gii', 'sub-01_T1w.nii.gz'};
+
+    verbose = false;
+
+    map = default_mapping();
+
+    for i = 1:size(input_output, 1)
+
+        rawsource = identify_rawsources(input_output{i, 1}, map, verbose);
+
+        assertEqual(rawsource, {['sub-01/' input_output{i, 2}]});
+
+    end
+
+end
+
+function test_identify_rawsources_when_der_entities()
+
+    input_output = {'sub-01_desc-skullstripped_T1w.nii', 'sub-01_T1w.nii.gz'};
+
+    verbose = false;
+
+    map = default_mapping();
+
+    for i = 1:size(input_output, 1)
+
+        rawsource = identify_rawsources(input_output{i, 1}, map, verbose);
+
+        assertEqual(rawsource, {['sub-01/' input_output{i, 2}]});
+
+    end
+
+end
+
 function test_identify_rawsources_suffix()
 
-    input_output = {'sub-01_T1w_seg8.mat', 'sub-01_T1w.nii'
-                    'sub-01_task-foo_bold_uw.mat', 'sub-01_task-foo_bold.nii'};
+    input_output = {'sub-01_T1w_seg8.mat', 'sub-01_T1w.nii.gz'
+                    'sub-01_task-foo_bold_uw.mat', 'sub-01_task-foo_bold.nii.gz'};
 
     verbose = false;
 
+    map = default_mapping();
+
     for i = 1:size(input_output, 1)
 
-        rawsource = identify_rawsources(input_output{i, 1}, verbose);
+        rawsource = identify_rawsources(input_output{i, 1},  map, verbose);
 
         assertEqual(rawsource, {['sub-01/' input_output{i, 2}]});
 
@@ -44,13 +100,15 @@ function test_identify_rawsources_anat()
 
     verbose = false;
 
+    map = default_mapping();
+
     for i = 1:numel(prefixes)
 
         file = [prefixes{i} anat_file];
 
-        rawsource = identify_rawsources(fullfile(pwd, 'sub-01', file), verbose);
+        rawsource = identify_rawsources(fullfile(pwd, 'sub-01', file), map,  verbose);
 
-        assertEqual(rawsource, {'sub-01/sub-01_T1w.nii'});
+        assertEqual(rawsource, {'sub-01/sub-01_T1w.nii.gz'});
 
     end
 
@@ -83,13 +141,15 @@ function test_identify_rawsources_func()
 
     verbose = false;
 
+    map = default_mapping();
+
     for i = 1:numel(prefixes)
 
         file = [prefixes{i} func_file];
 
-        rawsource = identify_rawsources(file, verbose);
+        rawsource = identify_rawsources(file, map, verbose);
 
-        assertEqual(rawsource, {'sub-01/ses-02/sub-01_ses-02_task-foo_bold.nii'});
+        assertEqual(rawsource, {'sub-01/ses-02/sub-01_ses-02_task-foo_bold.nii.gz'});
 
     end
 

tests/test_identify_sources.m

@@ -8,6 +8,24 @@
     initTestSuite;
 end
 
+function test_identify_sources_surface()
+
+    anat_file = 'sub-01_T1w.surf.gii';
+
+    prefix_output = {'wm',  'sub-01/sub-01_space-IXI549Space_desc-preproc_T1w.nii'};
+
+    map = default_mapping();
+
+    for i  = 1:size(prefix_output, 1)
+
+        sources = identify_sources([prefix_output{i, 1} anat_file], map, false);
+
+        assertEqual(sources{1}, prefix_output{i, 2});
+
+    end
+
+end
+
 function test_identify_sources_anat()
 
     anat_file = 'sub-01_T1w.nii';