NeuroDAGs

An Extensible and Declarative DAG Framework for Reproducible Neuroscience Workflows

M/EEG studies generate many interdependent intermediate derivatives. Recomputing full pipelines is wasteful; reusing valid intermediates is non-trivial. NeuroDAGs solves this with a declarative, graph-based framework for scalable, reproducible derivative computation.

Core idea: Pipelines are defined as a directed acyclic graph (DAG) of computation nodes. Each node outputs a reusable derivative. The DAG executes for each input file, skipping already-computed derivatives automatically.

Poster

View the NeuroDAGs conference poster — overview of motivation, design, and format.

Get Started

• Installation — install neurodags

• Terminal User Interface (TUI) — manage and run pipelines from the terminal

• Quickstart: Synthetic EEG Pipeline — minimal working pipeline in minutes

• Core Concepts — understand DAGs, derivatives, nodes, and artifacts

• pipeline.yml Reference — full pipeline.yml reference

• datasets.yml Reference — full datasets.yml reference

• Custom Nodes — write and register your own nodes

• Dataframe Assembly — build ML-ready dataframes from derivatives

• Inspection and Visualization — dry-run, error markers, DAG visualization, file explorer

• Parallelism and Execution Control — local n_jobs, subset execution, HPC pointer