Merge pull request #77 from Remi-Gau/method

[ENH] add methods section

giga_connectome/data/methods/template.jinja

@@ -0,0 +1,32 @@
+These results were generated with
+giga_connectome (version {{ data.version }}, https://giga-connectome.readthedocs.io/en/latest/)
+using Nilearn (version {{ data.nilearn_version }}, RRID:SCR_001362)
+and TemplateFlow (version {{ data.templateflow_version }}).
+
+The following steps were followed.
+
+1. Created subject specific grey matter mask in MNI space ({{ data.mni_space }}).
+
+1. Sampled the {{ data.atlas }} atlas to the space of subject specific grey matter mask in MNI space.
+
+1. Calculated the conjunction of the customised grey matter mask and resampled atlas to find valid parcels.
+
+1. Used the new input specific grey matter mask and atlas to extract time series and connectomes for each subject.
+   The time series data were denoised as follow:
+
+    - Time series extractions through label or map maskers are performed on the denoised nifti file.
+    - Denoising steps were performed on the voxel level:
+        - spatial smoothing (FWHM: {{ data.smoothing_fwhm }} mm)
+        - detrending, only if high pass filter was not applied through confounds
+        - regressing out confounds (using a {{ data.strategy }} strategy)
+        - standardized (using {{ data.standardize }})
+    - Extracted time series from atlas
+    - Computed correlation matrix (Pearson's correlation with LedoitWolf covariance estimator)
+
+{% if data.average_correlation %}
+1. Calculate intranetwork correlation of each parcel. The values replace the diagonal of the connectomes.
+{% endif %}
+
+{% if data.analysis_level %}
+1. Create a group average connectome.
+{% endif %}

giga_connectome/methods.py

@@ -0,0 +1,52 @@
+"""Module responsible for generating method section."""
+
+from pathlib import Path
+
+from jinja2 import Environment, FileSystemLoader, select_autoescape
+
+from nilearn import __version__ as nilearn_version
+from templateflow import __version__ as templateflow_version
+
+from giga_connectome import __version__
+
+
+def generate_method_section(
+    output_dir: Path,
+    atlas: str,
+    smoothing_fwhm: float,
+    strategy: str,
+    standardize: str,
+    mni_space: str,
+    average_correlation: bool,
+    analysis_level: bool,
+) -> None:
+
+    env = Environment(
+        loader=FileSystemLoader(Path(__file__).parent),
+        autoescape=select_autoescape(),
+        lstrip_blocks=True,
+        trim_blocks=True,
+    )
+
+    template = env.get_template("data/methods/template.jinja")
+
+    output_file = output_dir / "logs" / "CITATION.md"
+    output_file.parent.mkdir(parents=True, exist_ok=True)
+
+    data = {
+        "version": __version__,
+        "nilearn_version": nilearn_version,
+        "templateflow_version": templateflow_version,
+        "atlas": atlas,
+        "smoothing_fwhm": smoothing_fwhm,
+        "strategy": strategy,
+        "standardize": "percent signal change"
+        if standardize == "psc"
+        else standardize,
+        "mni_space": mni_space,
+        "average_correlation": average_correlation,
+        "analysis_level": analysis_level,
+    }
+
+    with open(output_file, "w") as f:
+        print(template.render(data=data), file=f)

giga_connectome/tests/test_methods.py

@@ -0,0 +1,14 @@
+from giga_connectome import methods
+
+
+def test_generate_method_section(tmp_path):
+    methods.generate_method_section(
+        output_dir=tmp_path,
+        atlas="DiFuMo",
+        smoothing_fwhm=5,
+        standardize="psc",
+        strategy="simple",
+        mni_space="MNI152NLin6Asym",
+        average_correlation=True,
+        analysis_level=True,
+    )

giga_connectome/workflow.py

@@ -11,7 +11,7 @@
 )
 
 from giga_connectome.denoise import is_ica_aroma
-from giga_connectome import utils
+from giga_connectome import utils, methods
 from giga_connectome.logger import gc_logger
 
 
@@ -63,6 +63,17 @@ def workflow(args):
 
     gc_log.info(f"Indexing BIDS directory:\n\t{bids_dir}")
 
+    methods.generate_method_section(
+        output_dir=output_dir,
+        atlas=atlas["name"],
+        smoothing_fwhm=smoothing_fwhm,
+        standardize=args.standardize,
+        strategy=args.denoise_strategy,
+        mni_space=template,
+        average_correlation=calculate_average_correlation,
+        analysis_level=analysis_level == "group",
+    )
+
     # create subject ts and connectomes
     # refactor the two cases into one
 

pyproject.toml

@@ -26,6 +26,7 @@ dependencies = [
   "templateflow < 23.0.0",
   "tqdm",
   "setuptools",
+  "jinja2 >= 2.0",
   "rich",
 ]
 dynamic = ["version"]