Draft pre-init ingression

.circleci/config.yml

@@ -4,7 +4,7 @@ orbs:
 
 .dockersetup: &dockersetup
   docker:
-    - image: pennlinc/qsirecon_build:24.8.3
+    - image: pennlinc/qsirecon_build:24.9.0
   working_directory: /src/qsirecon
 
 runinstall: &runinstall

Dockerfile

@@ -5,7 +5,7 @@ RUN pip install build
 RUN apt-get update && \
     apt-get install -y --no-install-recommends git
 
-FROM pennlinc/qsirecon_build:24.8.3
+FROM pennlinc/qsirecon_build:24.9.0
 
 # Install qsirecon
 COPY . /src/qsirecon

docs/input_data.rst

@@ -59,7 +59,10 @@ HCP Young Adult Preprocessed Data
 =================================
 
 To use minimally preprocessed dMRI data from HCP-YA specify ``--input-type hcpya``.
-The included FNIRT transforms are usable directly. NOTE: this does not work yet.
+Note that the transforms to/from MNI space are not able to be used at this time. Please note that if you have the
+HCPYA dataset from datalad (https://github.com/datalad-datasets/human-connectome-project-openaccess)
+then you should ``datalad get`` relevant subject data before running QSIRecon,
+and be mindful about how you mount the directory in Docker/Apptainer.
 
 .. _anat_reqs:
 

pyproject.toml

@@ -23,11 +23,12 @@ dependencies = [
     "filelock",
     "fury",
     "indexed_gzip <= 1.8.7",
+    "Ingress2QSIRecon == 0.2.1",
     "jinja2 < 3.1",
     "matplotlib",
     "networkx ~= 2.8.8",
-    "nibabel <= 5.2.0",
-    "nilearn == 0.10.1",
+    "nibabel <= 6.0.0",
+    "nilearn",
     "nipype == 1.8.6",
     "nireports ~= 24.0.2",
     "niworkflows >=1.9,<= 1.10",

qsirecon/cli/parser.py

@@ -116,11 +116,16 @@ def _bids_filter(value, parser):
     # required, positional arguments
     # IMPORTANT: they must go directly with the parser object
     parser.add_argument(
-        "bids_dir",
+        "input_dir",
         action="store",
+        metavar="input_dir",
         type=PathExists,
-        help="The root folder of a BIDS valid dataset (sub-XXXXX folders should "
-        "be found at the top level in this folder).",
+        help=(
+            "The root folder of the input dataset  "
+            "(subject-level folders should be found at the top level in this folder). "
+            "If the dataset is not BIDS-valid, "
+            "then a BIDS-compliant version will be created based on the --input-type value."
+        ),
     )
     parser.add_argument(
         "output_dir",
@@ -294,7 +299,7 @@ def _bids_filter(value, parser):
         default="qsiprep",
         choices=["qsiprep", "ukb", "hcpya"],
         help=(
-            "Specify which pipeline was used to create the data specified as the bids_dir."
+            "Specify which pipeline was used to create the data specified as the input_dir."
             "Not necessary to specify if the data was processed by QSIPrep. "
             "Other options include "
             '"ukb" for data processed with the UK BioBank minimal preprocessing pipeline and '
@@ -442,39 +447,84 @@ def parse_args(args=None, namespace=None):
             f"total threads (--nthreads/--n_cpus={config.nipype.nprocs})"
         )
 
-    bids_dir = config.execution.bids_dir
+    input_dir = config.execution.input_dir
     output_dir = config.execution.output_dir
     work_dir = config.execution.work_dir
     version = config.environment.version
 
-    # Update the config with an empty dict to trigger initialization of all config
-    # sections (we used `init=False` above).
-    # This must be done after cleaning the work directory, or we could delete an
-    # open SQLite database
-    config.from_dict({})
-
     # Ensure input and output folders are not the same
-    if output_dir == bids_dir:
+    if output_dir == input_dir:
         parser.error(
             "The selected output folder is the same as the input BIDS folder. "
             "Please modify the output path (suggestion: %s)."
-            % bids_dir
+            % input_dir
             / "derivatives"
             / ("qsirecon-%s" % version.split("+")[0])
         )
 
-    if bids_dir in work_dir.parents:
+    if input_dir in work_dir.parents:
         parser.error(
             "The selected working directory is a subdirectory of the input BIDS folder. "
             "Please modify the output path."
         )
 
     # Setup directories
-    config.execution.log_dir = config.execution.output_dir / "logs"
+    log_dir = output_dir / "logs"
     # Check and create output and working directories
-    config.execution.log_dir.mkdir(exist_ok=True, parents=True)
+    log_dir.mkdir(exist_ok=True, parents=True)
     work_dir.mkdir(exist_ok=True, parents=True)
 
+    # Run ingression if necessary
+    if config.workflow.input_type in ("hcpya", "ukb"):
+        import os.path as op
+        import shutil
+
+        from ingress2qsirecon.data import load_resource
+        from ingress2qsirecon.utils.functions import create_layout
+        from ingress2qsirecon.utils.workflows import create_ingress2qsirecon_wf
+
+        # Fake BIDS directory to be created
+        config.execution.bids_dir = work_dir / "bids"
+
+        # Make fake BIDS files
+        bids_scaffold = str(load_resource("bids_scaffold"))
+        if not op.exists(op.join(config.execution.bids_dir, "dataset_description.json")):
+            shutil.copytree(
+                bids_scaffold,
+                config.execution.bids_dir,
+                dirs_exist_ok=True,
+            )
+
+        if config.execution.participant_label is None:
+            participants_ingression = []
+        else:
+            participants_ingression = list(config.execution.participant_label)
+        layouts = create_layout(
+            config.execution.input_dir,
+            config.execution.bids_dir,
+            config.workflow.input_type,
+            participants_ingression,
+        )
+
+        # Create the ingression workflow
+        wf = create_ingress2qsirecon_wf(
+            layouts,
+            base_dir=work_dir,
+        )
+
+        # Configure the nipype workflow
+        wf.config["execution"]["crashdump_dir"] = str(log_dir)
+        wf.run()
+    else:
+        config.execution.bids_dir = config.execution.input_dir
+
+    # Update the config with an empty dict to trigger initialization of all config
+    # sections (we used `init=False` above).
+    # This must be done after cleaning the work directory, or we could delete an
+    # open SQLite database
+    config.from_dict({})
+    config.execution.log_dir = log_dir
+
     # Force initialization of the BIDSLayout
     config.execution.init()
     all_subjects = config.execution.layout.get_subjects()

qsirecon/config.py

@@ -383,6 +383,9 @@ class execution(_Config):
 
     bids_dir = None
     """An existing path to the dataset, which must be BIDS-compliant."""
+    input_dir = None
+    """An existing path to the input data, which may not be BIDS-compliant
+    (in which case a BIDS-compliant version will be created and stored as bids_dir)."""
     derivatives = {}
     """Path(s) to search for pre-computed derivatives"""
     bids_database_dir = None

qsirecon/interfaces/anatomical.py

@@ -11,7 +11,6 @@
 
 import os.path as op
 from glob import glob
-from pathlib import Path
 
 import nibabel as nb
 import numpy as np
@@ -26,9 +25,6 @@
 )
 from nipype.utils.filemanip import fname_presuffix
 
-from ..utils.ingress import ukb_dirname_to_bids
-from .images import to_lps
-
 LOGGER = logging.getLogger("nipype.interface")
 
 
@@ -151,35 +147,6 @@ def _get_if_exists(self, name, pattern, excludes=None):
             self._results[name] = files[0]
 
 
-class UKBAnatomicalIngressInputSpec(QSIPrepAnatomicalIngressInputSpec):
-    recon_input_dir = traits.Directory(
-        exists=True, mandatory=True, help="directory containing a single subject's results"
-    )
-
-
-class UKBAnatomicalIngress(QSIPrepAnatomicalIngress):
-    input_spec = UKBAnatomicalIngressInputSpec
-
-    def _run_interface(self, runtime):
-        # Load the Bias-corrected brain and brain mask
-        input_path = Path(self.inputs.recon_input_dir)
-        bids_name = ukb_dirname_to_bids(self.inputs.recon_input_dir)
-
-        ukb_brain = input_path / "T1" / "T1_unbiased_brain.nii.gz"
-        ukb_brain_mask = input_path / "T1" / "T1_brain_mask.nii.gz"
-
-        conformed_t1w_file = str(Path(runtime.cwd) / (bids_name + "_desc-preproc_T1w.nii.gz"))
-        conformed_mask_file = str(Path(runtime.cwd) / (bids_name + "_desc-brain_mask.nii.gz"))
-
-        to_lps(nb.load(ukb_brain)).to_filename(conformed_t1w_file)
-        to_lps(nb.load(ukb_brain_mask)).to_filename(conformed_mask_file)
-
-        self._results["t1_preproc"] = conformed_t1w_file
-        self._results["t1_brain_mask"] = conformed_mask_file
-
-        return runtime
-
-
 """
 
 The spherical harmonic coefficients are stored as follows. First, since the

qsirecon/interfaces/ingress.py

@@ -20,11 +20,8 @@
 """
 
 import os.path as op
-import shutil
 from glob import glob
-from pathlib import Path
 
-import nibabel as nb
 from nipype import logging
 from nipype.interfaces.base import (
     BaseInterfaceInputSpec,
@@ -36,9 +33,6 @@
 from nipype.utils.filemanip import split_filename
 
 from .bids import get_bids_params
-from .dsi_studio import btable_from_bvals_bvecs
-from .images import ConformDwi, to_lps
-from .mrtrix import _convert_fsl_to_mrtrix
 
 LOGGER = logging.getLogger("nipype.interface")
 
@@ -119,61 +113,3 @@ def _get_qc_filename(self, out_root, params, desc, suffix):
         used_keys = ["subject_id", "session_id", "acq_id", "dir_id", "run_id"]
         fname = "_".join([params[key] for key in used_keys if params[key]])
         return out_root + "/" + fname + "_desc-%s_dwi.%s" % (desc, suffix)
-
-
-class _UKBioBankDWIIngressInputSpec(QSIPrepDWIIngressInputSpec):
-    dwi_file = File(exists=False, help="The name of what a BIDS dwi file may have been")
-    data_dir = traits.Directory(
-        exists=True, help="The UKB data directory for a subject. Must contain DTI/ and T1/"
-    )
-
-
-class UKBioBankDWIIngress(SimpleInterface):
-    input_spec = _UKBioBankDWIIngressInputSpec
-    output_spec = QSIPrepDWIIngressOutputSpec
-
-    def _run_interface(self, runtime):
-        runpath = Path(runtime.cwd)
-
-        # The UKB input files
-        in_dir = Path(self.inputs.data_dir)
-        dwi_dir = in_dir / "DTI" / "dMRI" / "dMRI"
-        ukb_bval_file = dwi_dir / "bvals"
-        ukb_bvec_file = dwi_dir / "bvecs"  # These are the same as eddy rotated
-        ukb_dwi_file = dwi_dir / "data_ud.nii.gz"
-        ukb_dwiref_file = dwi_dir / "dti_FA.nii.gz"
-
-        # The bids_name is what the images will be renamed to
-        bids_name = Path(self.inputs.dwi_file).name.replace(".nii.gz", "")
-        dwi_file = str(runpath / (bids_name + ".nii.gz"))
-        bval_file = str(runpath / (bids_name + ".bval"))
-        bvec_file = str(runpath / (bids_name + ".bvec"))
-        b_file = str(runpath / (bids_name + ".b"))
-        btable_file = str(runpath / (bids_name + "btable.txt"))
-        dwiref_file = str(runpath / (bids_name.replace("_dwi", "_dwiref") + ".nii.gz"))
-
-        dwi_conform = ConformDwi(
-            dwi_file=str(ukb_dwi_file), bval_file=str(ukb_bval_file), bvec_file=str(ukb_bvec_file)
-        )
-
-        result = dwi_conform.run()
-        Path(result.outputs.dwi_file).rename(dwi_file)
-        Path(result.outputs.bvec_file).rename(bvec_file)
-        shutil.copyfile(result.outputs.bval_file, bval_file)
-        # Reorient the dwi file to LPS+
-        self._results["dwi_file"] = dwi_file
-        self._results["bvec_file"] = bvec_file
-        self._results["bval_file"] = bval_file
-
-        # Create a btable_txt file for DSI Studio
-        btable_from_bvals_bvecs(bval_file, bvec_file, btable_file)
-        self._results["btable_file"] = btable_file
-
-        # Create a mrtrix .b file
-        _convert_fsl_to_mrtrix(bval_file, bvec_file, b_file)
-        self._results["b_file"] = b_file
-
-        # Create a dwi ref file
-        to_lps(nb.load(ukb_dwiref_file)).to_filename(dwiref_file)
-        self._results["dwi_ref"] = dwiref_file
-        return runtime