This repository contains the complete benchmarking suite used to evaluate and compare the performance of Python and Rust implementations for various computational biology and data analysis tasks.
We have made this repository publicly available to provide full transparency into our benchmarking methodology and results. This allows users to:
- Understand our methodology: See exactly how we conducted our performance comparisons
- Reproduce our results: Run the same benchmarks on your own systems
- Verify our findings: Independently validate our performance claims
- Extend our work: Use our framework to benchmark additional methods or implementations
py-methods/- Python implementations of various algorithmsrs-methods/- Rust implementations of the same algorithms
runners/- Shell scripts to execute benchmarks for both Python and Rusttimings/- Raw performance and memory usage data from benchmark runsanalysis/- R scripts for statistical analysis and visualizationanalysis-results/- Generated plots and statistical summaries
The repository includes performance comparisons for:
- DEG (Differential Expression Gene) Analysis - Gene expression analysis
- Filtering - Data filtering operations
- HVG (Highly Variable Genes) - Gene variability analysis
- KNN (K-Nearest Neighbors) - Nearest neighbor computations
- Normalization - Data normalization methods
- PCA (Principal Component Analysis) - Dimensionality reduction
- Python 3.x with required scientific computing libraries (Scanpy 1.11.1)
- The complete Python environment specification can be found in
environment-py.yml - To recreate the environment:
conda env create -f environment-py.yml
- The complete Python environment specification can be found in
- Rust toolchain (latest stable)
- R for analysis scripts
Due to file size restrictions, we are currently unable to publish the count files and datasets used in our testing within this repository. However, if you are interested in obtaining these data files to reproduce our exact benchmarks, please feel free to contact us directly and we will be happy to share them with you.
- Navigate to the appropriate runner directory (
runners/python/orrunners/rust/) - Execute the desired benchmark script (e.g.,
./pca_performance.sh) - Results will be saved to the
timings/directory
Run the R scripts in the analysis/ directory to generate statistical summaries and visualizations from the timing data.
This repository represents our commitment to open science and reproducible research. All code, data, and analysis methods are provided as-is, allowing the community to:
- Scrutinize our methods
- Identify potential improvements
- Build upon our work
- Contribute back to the community
We welcome contributions, suggestions, and feedback. Feel free to open issues or submit pull requests to help improve our benchmarking methodology.
This benchmarking suite was created to provide objective, reproducible performance comparisons between Python and Rust implementations of common computational biology algorithms.