Causality Analysis — Simulated Time Series Dataset

This repository contains the analysis code and tutorial notebooks for a reproducible effective-connectivity workflow applied to simulated multichannel time-series data. The project demonstrates a layered reproducibility framework covering data standardization, containerized execution, workflow automation, and method evaluation. Simulated five-node systems with known directed ground-truth structure are generated and converted into a BIDS-compatible EEG dataset. Effective-connectivity methods, including ADTF, PDC, DTF, conditional Granger causality, phase slope index, and transfer entropy, are evaluated against the known ground truth using AUC-ROC and average precision in the main tutorial. Additional benchmark and statistical validation notebooks are included for extended analyses. The repository includes: source code for simulation, BIDS conversion, connectivity estimation, and evaluation; Jupyter notebooks for the tutorial workflow; a Snakemake workflow for reproducing tutorial outputs; environment files for Conda-based execution; a Dockerfile and matching Docker image for containerized reproduction. This archived release corresponds to GitHub release v1.1.0 and the Docker image: viola1003/causality-analysis:v1.1.0.