The CausalBench challenge: A machine learning contest for gene network inference from single-cell perturbation data

Mathieu Chevalley, Jacob Sackett-Sanders, Yusuf H Roohani, Pascal Notin, Artemy Bakulin, Dariusz Brzezinski, Kaiwen Deng, Yuanfang Guan, Justin Hong, Michael Ibrahim, Wojciech Kotlowski, Marcin Kowiel, Panagiotis Misiakos, Achille Nazaret, Markus Püschel, Chris Wendler, Arash Mehrjou, Patrick Schwab
Proceedings of the Fourth Conference on Causal Learning and Reasoning, PMLR 275:533-551, 2025.

Abstract

In drug discovery, mapping interactions between genes within cellular systems is a crucial early step. Such maps are not only foundational for understanding the molecular mechanisms underlying disease biology but also pivotal for formulating hypotheses about potential targets for new medicines. Recognizing the need to elevate the construction of these gene-gene interaction networks, especially from large-scale, real-world datasets of perturbed single cells, the CausalBench Challenge was initiated. This challenge aimed to inspire the machine learning community to enhance state-of-the-art methods, emphasizing better utilization of expansive genetic perturbation data. Using the framework provided by the CausalBench benchmark, participants were tasked with refining the current methodologies or proposing new ones. This report provides an analysis and summary of the methods submitted during the challenge to give a partial image of the state of the art at the time of the challenge. Notably, the winning solutions significantly improved performance compared to previous baselines, establishing a new state of the art for this critical task in biology and medicine.

Cite this Paper

BibTeX
@InProceedings{pmlr-v275-chevalley25a, title = {The CausalBench challenge: A machine learning contest for gene network inference from single-cell perturbation data}, author = {Chevalley, Mathieu and Sackett-Sanders, Jacob and Roohani, Yusuf H and Notin, Pascal and Bakulin, Artemy and Brzezinski, Dariusz and Deng, Kaiwen and Guan, Yuanfang and Hong, Justin and Ibrahim, Michael and Kotlowski, Wojciech and Kowiel, Marcin and Misiakos, Panagiotis and Nazaret, Achille and P\"{u}schel, Markus and Wendler, Chris and Mehrjou, Arash and Schwab, Patrick}, booktitle = {Proceedings of the Fourth Conference on Causal Learning and Reasoning}, pages = {533--551}, year = {2025}, editor = {Huang, Biwei and Drton, Mathias}, volume = {275}, series = {Proceedings of Machine Learning Research}, month = {07--09 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v275/main/assets/chevalley25a/chevalley25a.pdf}, url = {https://proceedings.mlr.press/v275/chevalley25a.html}, abstract = {In drug discovery, mapping interactions between genes within cellular systems is a crucial early step. Such maps are not only foundational for understanding the molecular mechanisms underlying disease biology but also pivotal for formulating hypotheses about potential targets for new medicines. Recognizing the need to elevate the construction of these gene-gene interaction networks, especially from large-scale, real-world datasets of perturbed single cells, the CausalBench Challenge was initiated. This challenge aimed to inspire the machine learning community to enhance state-of-the-art methods, emphasizing better utilization of expansive genetic perturbation data. Using the framework provided by the CausalBench benchmark, participants were tasked with refining the current methodologies or proposing new ones. This report provides an analysis and summary of the methods submitted during the challenge to give a partial image of the state of the art at the time of the challenge. Notably, the winning solutions significantly improved performance compared to previous baselines, establishing a new state of the art for this critical task in biology and medicine.} }
Endnote
%0 Conference Paper %T The CausalBench challenge: A machine learning contest for gene network inference from single-cell perturbation data %A Mathieu Chevalley %A Jacob Sackett-Sanders %A Yusuf H Roohani %A Pascal Notin %A Artemy Bakulin %A Dariusz Brzezinski %A Kaiwen Deng %A Yuanfang Guan %A Justin Hong %A Michael Ibrahim %A Wojciech Kotlowski %A Marcin Kowiel %A Panagiotis Misiakos %A Achille Nazaret %A Markus Püschel %A Chris Wendler %A Arash Mehrjou %A Patrick Schwab %B Proceedings of the Fourth Conference on Causal Learning and Reasoning %C Proceedings of Machine Learning Research %D 2025 %E Biwei Huang %E Mathias Drton %F pmlr-v275-chevalley25a %I PMLR %P 533--551 %U https://proceedings.mlr.press/v275/chevalley25a.html %V 275 %X In drug discovery, mapping interactions between genes within cellular systems is a crucial early step. Such maps are not only foundational for understanding the molecular mechanisms underlying disease biology but also pivotal for formulating hypotheses about potential targets for new medicines. Recognizing the need to elevate the construction of these gene-gene interaction networks, especially from large-scale, real-world datasets of perturbed single cells, the CausalBench Challenge was initiated. This challenge aimed to inspire the machine learning community to enhance state-of-the-art methods, emphasizing better utilization of expansive genetic perturbation data. Using the framework provided by the CausalBench benchmark, participants were tasked with refining the current methodologies or proposing new ones. This report provides an analysis and summary of the methods submitted during the challenge to give a partial image of the state of the art at the time of the challenge. Notably, the winning solutions significantly improved performance compared to previous baselines, establishing a new state of the art for this critical task in biology and medicine.
APA
Chevalley, M., Sackett-Sanders, J., Roohani, Y.H., Notin, P., Bakulin, A., Brzezinski, D., Deng, K., Guan, Y., Hong, J., Ibrahim, M., Kotlowski, W., Kowiel, M., Misiakos, P., Nazaret, A., Püschel, M., Wendler, C., Mehrjou, A. & Schwab, P.. (2025). The CausalBench challenge: A machine learning contest for gene network inference from single-cell perturbation data. Proceedings of the Fourth Conference on Causal Learning and Reasoning, in Proceedings of Machine Learning Research 275:533-551 Available from https://proceedings.mlr.press/v275/chevalley25a.html.

Related Material