Skip to content

Workflows

A workflow is a whole analysis pipeline: the blocks you've placed on the canvas, the connections between their ports, and every parameter you've set. It is the unit you save, load, run, and share.

Anatomy of a workflow

Part What it is
Nodes Block instances on the canvas (built-in, custom, or composite).
Edges Connections from an output port to a compatible input port.
Parameters Per-node settings you edit by double-clicking a block.
Embedded block sources The Python source of any custom/composite block used, saved inside the workflow file so it is self-contained.

Running a workflow

Click Run to execute the whole graph. SynapChart resolves the dependency order automatically (a topological sort of the DAG), runs each block once its inputs are ready, and streams status + any disp() output to the Run Progress panel.

Smart caching. After the first run, each block caches its output keyed by its inputs, parameters, and source. On the next run, unchanged blocks are skipped and their cached outputs reused — only the branches you actually modified re-execute. This keeps iteration fast even on large recordings.

Saving, loading, and sharing

Save writes the current canvas to a .json file. The file is self-contained: it embeds the definitions of any custom or composite blocks the workflow uses, so a collaborator can open it without having those blocks installed.

Opening workflows that contain code

Because a saved workflow can carry custom Python, opening one that contains local or embedded blocks shows you a code manifest first and runs nothing until you review and approve it (built-in templates are trusted). See SECURITY.md for the consent model.

Templates

Templates are ready-made workflows you can open as a starting point (toolbar → Templates). Opening one drops a fully wired pipeline onto a new canvas tab; modify it freely without affecting the original. The example templates ship with the package (backend/workflows/templates/); the tutorial datasets are generated from the repository's scripts/ folder — see Getting Started for the setup.

Worked examples

  • Theta Phase Precession — the reference end-to-end workflow, block by block.
  • Tutorial 1 — build a workflow from scratch, then package it as a composite block.
  • Tutorial 2 — batch-process many sessions with the Dataset Iterator.
  • Tutorial 3 — a real-data theta-sequence workflow on CRCNS HC-11.