simChef Case Study

This repository uses the simChef simulation framework to reproduce a portion of the simulation results presented in "A flexible approach for predictive biomarker discovery" by Boileau et al. The method presented in this preprint is implemented in the publicly available uniCATE R package and the original simulation results are available in the repository pub_uniCATE.

The simulation documentation and results generated by simChef are found here.

Navigating the Repository

The list below summarizes this repository's contents:

• R/: The directory containing all of the R code. The data-generating processes are in dgps-functions/, the methods in method-functions/, simChef evaluator functions in evaluator-functions/, and the visualization functionality in visualizer-functons/. The main simChef file is meal.R; the simulation study code is found here.
• results/: The results produced by running R/meal.R are saved here.
• slurm/: This directory contains a bash script for running R/meal.R using SLURM scheduler
• logs/: The log file generated by the SLURM scheduler is saved in this directory.
• renv/: This directory is automatically generated by the renv R package. Do not modify its contents manually.

Reproducing the Simulation Study

You can reproduce the partial simulation study locally by running the contents of the R/meal.R file. Be sure to install the renv package first to ensure that all required libraries are installed. R/meal.R can also be run on an high-performance computing environment. Checkout the slurm/savio-run-experiment.sh example SLURM script. You might also want to modify the future plan based on the computational resources at your disposal.

Citing simChef and uniCATE

If you use the simChef package for your own simulation study, please cite it with the following BibTeX entry:

@manual{simChef,
  title = {simChef: Intensive Computational Experiments Made Easy},
  author = {James Duncan and Tiffany Tang},
  year = {2022},
  note = {R package version 0.0.2},
  url = {https://yu-group.github.io/simChef},
}

uniCATE can be cited using the following entry:

@misc{boileau-2022,
  doi = {10.48550/ARXIV.2205.01285},
  url = {https://arxiv.org/abs/2205.01285},
  author = {Boileau, Philippe and Qi, Nina Ting and van der Laan, Mark J. and Dudoit, Sandrine and Leng, Ning},
  keywords = {Methodology (stat.ME), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {A Flexible Approach for Predictive Biomarker Discovery},
  publisher = {arXiv},
  year = {2022}
}

---

© 2022 Philippe Boileau

The contents of this repository are distributed under the MIT license. See file LICENSE for details.