Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Neural optimal transport predicts perturbation responses at the single-cell level

We developed CellOT, a tool that integrates optimal transport with input convex neural networks to predict molecular responses of individual cells to various perturbations. By learning a map between the unpaired distributions of unperturbed and perturbed cells, CellOT outperforms current methods and generalizes the inference of treatment outcomes in unobserved cell types and patients.

This is a preview of subscription content, access via your institution

Access options

Rent or buy this article

Prices vary by article type



Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Overview of the CellOT model and its application.


  1. Dagogo-Jack, I. & Shaw, A. Tumour heterogeneity and resistance to cancer therapies. Nat. Rev. Clin. Oncol. 15, 81–94 (2018). This review article discusses tumor heterogeneity and subsequent differences in sensitivity and resistance to treatment.

    Article  CAS  PubMed  Google Scholar 

  2. Schiebinger, G. et al. Optimal-transport analysis of single-cell gene expression identifies developmental trajectories in reprogramming. Cell 176, 928–943.e22 (2021). This paper uses classical optimal transport to analyze cell differentiation processes.

    Article  Google Scholar 

  3. Makkuva, A. et al. Optimal transport mapping via input convex neural networks. Proc. 37th Intl Conf. Machine Learning 119, 6672–6681 (2020). This paper presents a neural optimal transport framework that parameterizes the map using the gradient of an input convex neural network.

    Google Scholar 

  4. Gut, G. et al. Multiplexed protein maps link subcellular organization to cellular states. Science 361, eaar7042 (2018). This paper discusses the multiplexed imaging technology 4i, a platform for simultaneous quantification of protein expression levels and cell morphology in single cells.

    Article  PubMed  Google Scholar 

  5. Bunne, C. et al. Supervised training of conditional Monge maps. Adv. Neural Inf. Process. Syst. Vol. 35 (eds Koyejo, S. et al.) (2022). This work introduces a neural optimal transport scheme that can be conditioned on a context such as patient metadata.

Download references

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This is a summary of: Bunne, C. et al. Learning single-cell perturbation responses using neural optimal transport. Nat. Methods (2023).

Rights and permissions

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Neural optimal transport predicts perturbation responses at the single-cell level. Nat Methods 20, 1639–1640 (2023).

Download citation

  • Published:

  • Issue Date:

  • DOI:


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing