Skip to content

PHOTONAI Projects Documentation

Welcome to the PHOTONAI Projects toolbox — a lightweight framework for organizing, running, and statistically comparing PHOTONAI analyses.

This package helps you:

  • create and manage multiple analyses inside a structured project folder
  • run PHOTONAI Hyperpipes on stored datasets
  • perform permutation tests (locally or on SLURM clusters)
  • compare multiple analyses statistically (using Nadeau-Bengio or permutations)

Project Management

Project Structure

Each PHOTONAI analysis is stored in a clean folder structure:

project/
├── analysis_1/
│   ├── data/
│   │   ├── X.npy
│   │   └── y.npy
│   ├── permutations/
│   │   ├── 0/
│   │   ├── 1/
│   │   └── ...
│   ├── slurm_job.cmd
│   ├── hyperpipe_constructor.py
│   └── hyperpipe_meta.json
├── analysis_2/
└── ...

Create Project

At the beginning of your project, create a new PHOTONAI project.

project = PhotonaiProject(project_folder='project')
project.add(name="analysis_name", 
            X=X, 
            y=y, 
            hyperpipe_script="path/to/hyperpipe_constructor.py",
            name_hyperpipe_constructor="create_hyperpipe")

Add Analysis

Then add PHOTONAI analyses using .add(). 

project.add(name="analysis_name", 
            X=X, 
            y=y, 
            hyperpipe_script="path/to/hyperpipe_constructor.py",
            name_hyperpipe_constructor="create_hyperpipe")

Run Analysis

Simply use the .run() method to run a specific analysis that has already been added:

project.run("analysis_name")

Statistics

Permutation Test

Generate permutation-based null distributions for a specific analysis:

project.run_permutation_test("analysis_name", n_perms=1000)

Compare Analyses

Compare two analyses using:

  • method="nadeau-bengio"
  • method="permutation"
project.compare_analyses("analysis_1", "analysis_2", method="permutation")