This repository contains all scripts to reproduce the associated publication.
- normancontains all analyses related to "Learning and exploring perturbation representations with pertpy".
- mcfarlandcontains all analyses related to "Pertpy streamlines discovery for complex perturbation experiments"
- zhangcontains all analyses related to "Pertpy enables deciphering effects of perturbations on cellular systems".
The benchmark folder has individual tool specific subfolders that each contain a Conda *_environment.yml file together with a *_comparison notebook where the original implementation is compared against the implementation in pertpy with a suitable metric.
The benchmark folder further contains all scripts that were run using a Snakemake pipeline to evaluate the runtime requirements of the tools.
In particular, the visualize_benchmark_results.ipynb notebook aggregates all results from the benchmarking_table.csv which is stored in the tables folder to create the figure.