relationship-structure-identification 0.0.1

Implementation of our method for identifying the relationship structure among multiple datasets

Simulations to identify the relationship structure among multiple datasets using Independent Vector Analysis

This package contains the code for reproducing the simulations of our paper:

Isabell Lehmann, Tanuj Hasija, Ben Gabrielson, M. A. B. S. Akhonda, Vince D. Calhoun, Tülay Adali, Identifying the Relationship Structure among Multiple Datasets Using Independent Vector Analysis: Application to Multi-task fMRI Data, submitted in 2023

Installing this Package

The only pre-requisite is to have Python 3 (version >=3.11) installed. This package can be installed (optionally in a virtual environment) with:

git clone https://github.com/SSTGroup/relationship_structure_identification
cd relationship_structure_identification
pip install -e .

Required third-party packages will automatically be installed.

Generating Simulations and Results

The simulated data is generated with:

cd relationship_structure_identification
python simulations.py @../simulations/simulation_parameters.txt

After running the code, the folder relationship_structure_identification/simulations will contain the generated .npy files, consisting of the true data and the estimated results for IVA-L-SOS, the bootstrap and the clustering, for each Monte-Carlo run.

Then, the performance metrics are calcuted with (from the relationship_structure_identification folder):

python performance_metrics.py @../simulations/performance_parameters.txt

The .npy files containing the performace metrics will also be saved in the simulations folder.

Visualizing Results

After having calculated the performance metrics, the boxplots can be generated by running the notebook. The metrics_rhoxx.npy files must be in the simulations folder.

Changing Parameters

By changing the scenario in parameters.txt, the simulations for different values of 'rho' are generated. The other parameters are set to the values according to the simulations in our paper. The default values can be changed by adding the parameters to the simulation_parameters.txt or performance_parameters.txt files.

Contact

In case of questions, suggestions, problems etc. please send an email to [e-mail], or open an issue here on Github.

Citing

If you use this code in an academic paper, please cite [1]

@article{Lehmann2023,
  title   = {Identifying the Relationship Structure among Multiple Datasets Using Independent Vector Analysis: Application to Multi-task fMRI Data},
  author  = {Lehmann, Isabell and Hasija, Tanuj and Gabrielson, Ben and Akhonda, M. A. B. S. and Calhoun, Vince D. and Adali, T{\"u}lay},
  booktitle={tba},
  pages={tba},
  year={2023}
  } 

[1] Isabell Lehmann, Tanuj Hasija, et al., Identifying the Relationship Structure among Multiple Datasets Using Independent Vector Analysis: Application to Multi-task fMRI Data, submitted in 2023.