Code and Software
Overview
Research is increasingly defined not just by the data collected, but by the code used to analyze it. Curating code is distinct from curating data; it requires ensuring reproducibility, readability, and security. This section provides tools to inspect code files, whether they are interactive notebooks or standalone scripts.
Key Objectives:
- Reproducibility: We identify dependencies (libraries, packages) to ensure the environment can be recreated.
- Cleanliness: We check for absolute paths (e.g.,
C:/Users/Daniel/...) that prevent the code from running on other machines. - Security: We scan for potential leaks of sensitive information, such as API tokens or personal identifiers embedded in the code.
Supported Formats
Jupyter Notebooks (.ipynb)
Interactive notebooks found in Python, R, and Julia research pipelines. - Our Approach: We parse the internal JSON structure to extract kernel information and checking for uncleared output cells that may bloat the file or contain sensitive data.
R and Quarto Scripts (.R, .qmd)
The standard for reproducible research in the R ecosystem. - Our Approach: We scan the text content to list all loaded packages and verify that file paths are relative, not absolute.
Common Curation Challenges
- “Works on My Machine”: The most common issue in code curation. By flagging absolute paths and listing dependencies, we help bridge the gap between the researcher’s environment and the archive.
- Hidden State: Notebooks often contain variables in memory that aren’t explicitly saved. We encourage restarting the kernel and running all cells to ensure the “Code” actually produces the “Output”.