Skip to content

CLAUDE.md

Latest commit

 

History

History
92 lines (65 loc) · 3.26 KB

CLAUDE.md

File metadata and controls

92 lines (65 loc) · 3.26 KB

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

sMRIPrep is a structural MRI preprocessing pipeline in the NiPreps ecosystem (alongside fMRIPrep, dMRIPrep). It preprocesses T1w/T2w/FLAIR anatomical MRI data following BIDS conventions. It uses Nipype workflow engine, FreeSurfer, ANTs, FSL, and Connectome Workbench.

Common Commands

Testing

# Run full test suite (uses pytest-xdist for parallelism)
tox

# Run tests for a specific Python version
tox -e py312-latest

# Run a single test file
pytest src/smriprep/interfaces/tests/test_surf.py

# Run a single test
pytest src/smriprep/interfaces/tests/test_surf.py::test_function_name

# Note: default pytest addopts include -svx, --doctest-modules, --cov
# To skip doctests when running a subset:
pytest --override-ini="addopts=" src/smriprep/interfaces/tests/test_surf.py

Linting and Formatting

# Check style (ruff lint + format check)
tox -e style

# Auto-fix style issues
tox -e style-fix

# Spell check
tox -e spellcheck

# Direct ruff usage
ruff check --diff
ruff format --diff

Building

tox -e build

Code Style

  • Formatter/Linter: Ruff (line length 99, single quotes)
  • Quote style: Single quotes everywhere
  • Black is explicitly disabled

Architecture

Source Layout

Uses src/ layout: all source code lives under src/smriprep/.

Workflow Layer (workflows/)

The core of the pipeline. All workflows are Nipype pe.Workflow objects built by init_*_wf() factory functions.

  • base.py — Top-level orchestrator init_smriprep_wf() → creates per-subject workflows
  • anatomical.py — Core anatomical pipeline init_anat_preproc_wf(): validation → conformance → bias correction (N4) → skull stripping (ANTs) → segmentation (FAST) → spatial normalization → optional FreeSurfer surface reconstruction
  • surfaces.py — Surface processing: FreeSurfer recon, GIFTI conversion, fsLR registration (MSM), grayordinates (CIFTI), HCP morphometrics
  • outputs.py — BIDS-Derivatives output writing via DerivativesDataSink
  • fit/registration.py — Spatial normalization (registration to standard templates)

Interfaces Layer (interfaces/)

Custom Nipype interfaces wrapping external tools (FreeSurfer, FSL, Workbench, MSM) and internal operations (GIFTI/CIFTI manipulation, TemplateFlow queries, surface operations).

CLI (cli/run.py)

BIDS-App compliant command-line interface. Entry point: smriprep.cli.run:main.

Key Ecosystem Dependencies

  • niworkflows — Shared NiPreps building blocks (brain extraction, image utilities, BIDS I/O)
  • templateflow — Standardized neuroimaging templates
  • nireports — Visual QC reports
  • nitransforms — Spatial transforms

Naming Conventions (from CONTRIBUTING.md)

  • Workflow factory functions: init_<name>_wf()
  • Workflow variables end in _wf
  • Node names must match their variable names (aids debugging working directories)
  • PR titles use prefixes: ENH, FIX, TST, DOC, STY, REF, CI, MAINT

Environment Variables

Key variables passed through tox: TEMPLATEFLOW_HOME, FREESURFER_HOME, SUBJECTS_DIR, FS_LICENSE, TEST_DATA_HOME, TEST_OUTPUT_DIR, TEST_WORK_DIR.