Concepts & Workflow#

Every Coord2Region run follows the same lifecycle, whether you are running a single coordinate in the CLI or a batch of thousands in Python.

1. Gather Inputs#

The pipeline accepts two primary input types:

  • Coordinates: MNI or Talairach coordinates (x, y, z).

  • Region Names: Anatomical labels (e.g., “Left Amygdala”) which are first mapped to centroids using a reference atlas.

2. Configure Providers#

Coord2Region is modular. You must decide which external services to enable:

  • Atlases: Local anatomical lookups (e.g., Harvard-Oxford). See ../atlases.

  • Literature: Meta-analytic databases (Neurosynth/NeuroQuery) for finding studies.

  • AI Providers: LLMs (OpenAI, Gemini) for semantic summaries and images. See ../providers.

Tip: Run ``python scripts/configure_coord2region.py`` to generate a persistent YAML config.

3. Choose an Interface#

Select the tool that matches your technical comfort level:

  • Web Builder: Design pipelines visually and export the config. See Web Interface.

  • CLI: The standard way to run reproducible recipes. See CLI Tools.

  • Python API: For direct integration into scripts or notebooks. See ../pipeline.

4. Inspect Artefacts#

By default, results land in coord2region-output/.

  • Structured Data: JSON and CSV files containing the raw mappings and study lists.

  • Reports: PDF or Markdown summaries of the findings.

  • Media: Generated brain overlays or AI illustrations.

  • Provenance: A YAML copy of the exact configuration used to generate the run.