Journal article · 2026 · Open access

Deep learning perturbation models can outperform baselines on calibrated metrics

Henry E. Miller, Gabriel Mejia, Francis Leblanc, Brendan Swain, Bo Wang, Lucas Paulo de Lima Camillo

Nature Biotechnology · Published 1 October 2026 · DOI 10.1038/s41587-026-03307-w

Open PDFView at the publisherCite this paper

In brief

“Deep learning perturbation models can outperform baselines on calibrated metrics” is a 2026 journal article by Henry E. Miller and 5 co-authors, published in Nature Biotechnology. OpenAlex counts 26 citations as of 16 November 2026, which puts it in the top 1% of works published in 2026 by citations. It is open access under a CC BY licence: the full text is free to read at Nature Biotechnology.

Citations
26
Most cited in
2026 (26)
Field-weighted impact
32.93
References
33
Authors
6

Abstract

Recent benchmarks report that deep-learning-based genetic perturbation models fail to outperform uninformative baselines. Here we introduce a positive control baseline and metric calibration framework, showing—across 14 datasets and 18 metrics—that common benchmarking metrics (for example, mean squared error and Pearson Δ ) are frequently miscalibrated, with reduced sensitivity to measure positive model performance. Under well-calibrated metrics, we find that deep-learning-based genetic perturbation models can outperform uninformative baselines.

Cite this paper

Formatted from the catalogue record. Check names, capitals and page numbers against the paper before you submit.

APA 7

Miller, H. E., Mejia, G., Leblanc, F., Swain, B., Wang, B., & Camillo, L. P. D. L. (2026). Deep learning perturbation models can outperform baselines on calibrated metrics. Nature Biotechnology. https://doi.org/10.1038/s41587-026-03307-w

MLA 9

Miller, Henry E., et al. “Deep learning perturbation models can outperform baselines on calibrated metrics.” Nature Biotechnology, 2026. https://doi.org/10.1038/s41587-026-03307-w.

Chicago

Miller, Henry E., Gabriel Mejia, Francis Leblanc, Brendan Swain, Bo Wang, and Lucas Paulo de Lima Camillo. 2026. “Deep learning perturbation models can outperform baselines on calibrated metrics.” Nature Biotechnology. https://doi.org/10.1038/s41587-026-03307-w.

Harvard

Miller, H. E. et al. (2026) ‘Deep learning perturbation models can outperform baselines on calibrated metrics’, Nature Biotechnology. doi:10.1038/s41587-026-03307-w.

IEEE

H. E. Miller, G. Mejia, F. Leblanc, B. Swain, B. Wang, and L. P. D. L. Camillo, “Deep learning perturbation models can outperform baselines on calibrated metrics,” Nature Biotechnology, 2026, doi: 10.1038/s41587-026-03307-w.

BibTeX

@article{miller2026deep,
  title     = {{Deep learning perturbation models can outperform baselines on calibrated metrics}},
  author    = {Miller, Henry E. and Mejia, Gabriel and Leblanc, Francis and Swain, Brendan and Wang, Bo and Camillo, Lucas Paulo de Lima},
  journal   = {Nature Biotechnology},
  year      = {2026},
  doi       = {10.1038/s41587-026-03307-w},
  url       = {https://doi.org/10.1038/s41587-026-03307-w}
}

Download for a reference manager: BibTeX (.bib) · RIS for Zotero, Mendeley and EndNote (.ris)

Topic and keywords

Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene Regulatory Network Analysis

  • deep learning
  • metric calibration
  • benchmarking
  • model evaluation
  • mean squared error
  • Pearson correlation
  • gene expression perturbation

Work with this paper

Most-cited papers in Biochemistry, Genetics and Molecular Biology

All highly cited Biochemistry, Genetics and Molecular Biology papers

Questions about this paper

How do I cite this paper in APA style?

Miller, H. E., Mejia, G., Leblanc, F., Swain, B., Wang, B., & Camillo, L. P. D. L. (2026). Deep learning perturbation models can outperform baselines on calibrated metrics. Nature Biotechnology. https://doi.org/10.1038/s41587-026-03307-w

How many times has this paper been cited?

26 times, according to OpenAlex on 16 November 2026. Google Scholar and Scopus count different sets of sources, so their numbers differ.

Is this paper free to read?

Yes. It is open access at Nature Biotechnology: sign in and open the PDF from this page.

Who wrote this paper?

Henry E. Miller, Gabriel Mejia, Francis Leblanc, Brendan Swain, Bo Wang, Lucas Paulo de Lima Camillo.

What is the DOI of this paper?

10.1038/s41587-026-03307-w, permanently at https://doi.org/10.1038/s41587-026-03307-w. Sign in and open it from this page.

Bibliographic data from OpenAlex (CC0), last checked 16 November 2026. ResearchGuru does not host this paper; once you are signed in, the links open it at the publisher or the repository that does.