Treat anatomical data differently in data collection (#75)

src/fmripost_aroma/data/io_spec.json

@@ -63,7 +63,7 @@
                     ".nii"
                 ]
             },
-            "confounds": {
+            "bold_confounds": {
                 "datatype": "func",
                 "echo": null,
                 "part": [
@@ -79,12 +79,27 @@
                     ".tsv"
                 ]
             },
+            "anat_mni152nlin6asym": {
+                "datatype": "anat",
+                "echo": null,
+                "part": [
+                    "mag",
+                    null
+                ],
+                "res": "2",
+                "space": "MNI152NLin6Asym",
+                "desc": "preproc",
+                "suffix": [
+                    "T1w",
+                    "T2w"
+                ],
+                "extension": [
+                    ".nii.gz",
+                    ".nii"
+                ]
+            },
             "anat_dseg": {
                 "datatype": "anat",
-                "task": null,
-                "acquisition": null,
-                "run": null,
-                "part": null,
                 "space": null,
                 "res": null,
                 "den": null,
@@ -97,9 +112,8 @@
             }
         },
         "transforms": {
-            "hmc": {
+            "bold_hmc": {
                 "datatype": "func",
-                "part": null,
                 "from": "orig",
                 "to": "boldref",
                 "mode": "image",
@@ -108,7 +122,6 @@
             },
             "boldref2anat": {
                 "datatype": "func",
-                "part": null,
                 "from": "boldref",
                 "to": ["anat", "T1w", "T2w"],
                 "mode": "image",
@@ -117,18 +130,14 @@
             },
             "boldref2fmap": {
                 "datatype": "func",
-                "part": null,
-                "from": "boldref",
+                "from": "orig",
                 "mode": "image",
                 "suffix": "xfm",
                 "extension": ".txt"
             },
             "anat2mni152nlin6asym": {
                 "datatype": "anat",
-                "task": null,
-                "acquisition": null,
                 "run": null,
-                "part": null,
                 "from": ["anat", "T1w", "T2w"],
                 "to": "MNI152NLin6Asym",
                 "space": null,

src/fmripost_aroma/tests/test_utils_bids.py

@@ -80,8 +80,8 @@ def test_collect_derivatives_minimal(minimal_ignore_list):
         # TODO: Add bold_mask_native to the dataset
         # 'bold_mask_native': 'sub-01_task-mixedgamblestask_run-01_desc-brain_mask.nii.gz',
         'bold_mask_native': None,
-        'confounds': None,
-        'hmc': (
+        'bold_confounds': None,
+        'bold_hmc': (
             'sub-01_task-mixedgamblestask_run-01_from-orig_to-boldref_mode-image_desc-hmc_xfm.txt'
         ),
         'boldref2anat': (
@@ -123,8 +123,8 @@ def test_collect_derivatives_full(full_ignore_list):
             'mask.nii.gz'
         ),
         'bold_mask_native': None,
-        'confounds': 'sub-01_task-mixedgamblestask_run-01_desc-confounds_timeseries.tsv',
-        'hmc': (
+        'bold_confounds': 'sub-01_task-mixedgamblestask_run-01_desc-confounds_timeseries.tsv',
+        'bold_hmc': (
             'sub-01_task-mixedgamblestask_run-01_from-orig_to-boldref_mode-image_desc-hmc_xfm.txt'
         ),
         'boldref2anat': (

src/fmripost_aroma/utils/bids.py

@@ -119,26 +119,43 @@ def collect_derivatives(
             )
 
         for k, q in spec['derivatives'].items():
-            # Combine entities with query. Query values override file entities.
-            query = {**entities, **q}
+            if k.startswith('anat'):
+                # Allow anatomical derivatives at session level or subject level
+                query = {**{'session': [entities.get('session'), None]}, **q}
+            else:
+                # Combine entities with query. Query values override file entities.
+                query = {**entities, **q}
+
             item = layout.get(return_type='filename', **query)
             if not item:
                 derivs_cache[k] = None
+            elif not allow_multiple and len(item) > 1 and k.startswith('anat'):
+                # Anatomical derivatives are allowed to have multiple files (e.g., T1w and T2w)
+                # but we just grab the first one
+                derivs_cache[k] = item[0]
             elif not allow_multiple and len(item) > 1:
                 raise ValueError(f'Multiple files found for {k}: {item}')
             else:
                 derivs_cache[k] = item[0] if len(item) == 1 else item
 
         for k, q in spec['transforms'].items():
-            # Combine entities with query. Query values override file entities.
-            # TODO: Drop functional entities (task, run, etc.) from anat transforms.
-            query = {**entities, **q}
+            if k.startswith('anat'):
+                # Allow anatomical derivatives at session level or subject level
+                query = {**{'session': [entities.get('session'), None]}, **q}
+            else:
+                # Combine entities with query. Query values override file entities.
+                query = {**entities, **q}
+
             if k == 'boldref2fmap':
                 query['to'] = fieldmap_id
 
             item = layout.get(return_type='filename', **query)
             if not item:
                 derivs_cache[k] = None
+            elif not allow_multiple and len(item) > 1 and k.startswith('anat'):
+                # Anatomical derivatives are allowed to have multiple files (e.g., T1w and T2w)
+                # but we just grab the first one
+                derivs_cache[k] = item[0]
             elif not allow_multiple and len(item) > 1:
                 raise ValueError(f'Multiple files found for {k}: {item}')
             else:
@@ -178,8 +195,11 @@ def collect_derivatives(
 
         spaces_found, anat2outputspaces_xfm = [], []
         for space in spaces.references:
-            # First try to find processed BOLD+mask files in the requested space
-            anat2space_query = {**entities, **spec['transforms']['anat2mni152nlin6asym']}
+            # Now try to find transform to the requested space
+            anat2space_query = {
+                **{'session': [entities.get('session'), None]},
+                **spec['transforms']['anat2mni152nlin6asym'],
+            }
             anat2space_query['to'] = space.space
             item = layout.get(return_type='filename', **anat2space_query)
             anat2outputspaces_xfm.append(item[0] if item else None)

src/fmripost_aroma/workflows/base.py

@@ -322,7 +322,7 @@ def init_single_run_wf(bold_file):
                 ),
             )
 
-        if not functional_cache['confounds']:
+        if not functional_cache['bold_confounds']:
             if config.workflow.dummy_scans is None:
                 raise ValueError(
                     'No confounds detected. '
@@ -357,17 +357,17 @@ def init_single_run_wf(bold_file):
     if config.workflow.dummy_scans is not None:
         skip_vols = config.workflow.dummy_scans
     else:
-        if not functional_cache['confounds']:
+        if not functional_cache['bold_confounds']:
             raise ValueError(
                 'No confounds detected. '
                 'Automatical dummy scan detection cannot be performed. '
                 'Please set the `--dummy-scans` flag explicitly.'
             )
-        skip_vols = get_nss(functional_cache['confounds'])
+        skip_vols = get_nss(functional_cache['bold_confounds'])
 
     # Run ICA-AROMA
     ica_aroma_wf = init_ica_aroma_wf(bold_file=bold_file, metadata=bold_metadata, mem_gb=mem_gb)
-    ica_aroma_wf.inputs.inputnode.confounds = functional_cache['confounds']
+    ica_aroma_wf.inputs.inputnode.confounds = functional_cache['bold_confounds']
     ica_aroma_wf.inputs.inputnode.skip_vols = skip_vols
 
     mni6_buffer = pe.Node(niu.IdentityInterface(fields=['bold', 'bold_mask']), name='mni6_buffer')
@@ -427,7 +427,7 @@ def init_single_run_wf(bold_file):
             jacobian='fmap-jacobian' not in config.workflow.ignore,
             name='bold_MNI6_wf',
         )
-        bold_MNI6_wf.inputs.inputnode.motion_xfm = functional_cache['hmc']
+        bold_MNI6_wf.inputs.inputnode.motion_xfm = functional_cache['bold_hmc']
         bold_MNI6_wf.inputs.inputnode.boldref2fmap_xfm = functional_cache['boldref2fmap']
         bold_MNI6_wf.inputs.inputnode.boldref2anat_xfm = functional_cache['boldref2anat']
         bold_MNI6_wf.inputs.inputnode.anat2std_xfm = functional_cache['anat2mni152nlin6asym']