Published July 18, 2025 | Version v5

Code and data for the benchmark conducted in "Deep-learning-based gene perturbation effect prediction does not yet outperform simple linear baselines"

Authors/Creators

  • 1. UCL

Description

This is an archive of the code and data accompanying the publication "Deep-learning-based predictions of gene perturbation effects do not yet outperform simple linear baselines."

The zip file contains a copy of the GitHub repository available at https://github.com/const-ae/linear_perturbation_prediction-Paper. The files are the content of the `benchmark/output` folder, as they were too large for git.

Files

const-ae/linear_perturbation_prediction-Paper-publication.zip

Files (7.6 GB)

Name Size
md5:e352192c22e6577edba3dd788677caf2
25.9 MB Preview Download
md5:16a800b188229f8457571a416c35921c
6.5 kB Download
md5:8d37871d0a442f721e2c3c7d22efffbe
1.6 GB Download
md5:6ae6860fb62cc69afa9d2d5517bec09e
1.7 kB Download
md5:7a65b75022967ee4e8e4c0c1c1682a52
42.3 MB Download
md5:2b1bdd900d4ffe219ced2cce2fa64001
66.1 kB Download
md5:b3a58e203a11c4abf1ebd04dd7a9bf68
45.6 kB Download
md5:e6055477fb2c4dfbdde12997d1b6e006
5.9 GB Download

Additional details

References

  • Deep-learning-based predictions of gene perturbation effects do not yet outperform simple linear baselines. Constantin Ahlmann-Eltze, Wolfgang Huber, Simon Anders. Nature Methods 2025; https://doi.org/10.1038/s41592-025-02772-6