starting-a-new-project

Set up a reproducible HPC project directory structure by creating a standardized skeleton under /gpfs/project that separates code, raw data, derived data, and logs to ensure safe collaboration and workflow restartability.

Organize raw and derived data files for Slurm workflows by using dedicated data/raw and data/derived directories within your project path, ensuring clear data governance and preventing accidental modification of source files.

Initialize a Git repository for research code on the Yale HPC cluster by creating a dedicated code directory within your /gpfs/project path, then running standard version control initialization to track changes and enforce best practices.

You can use this project skeleton to onboard new research assistants consistently by applying step-by-step governance guidelines that enforce a uniform directory structure and standardized workflows across the entire team.

You need an environment manager like uv to establish a reproducible project layout, allowing you to initialize the code repository and apply the first setup steps with isolated dependencies on the HPC cluster.

Your HPC project layout needs separate directories for logs and outputs to maintain reproducibility, prevent file collision during Slurm job execution, and ensure that computational results remain organized and traceable.