A decade of peer-reviewed research in machine learning and perturbation modeling

Featured research

Scalable and universal prediction of cellular phenotypes

Ji Y, Tejada-Lapuerta A, Schmacke NA, Zheng Z, Zhang X, Khan S, Rothenaigner I, Tschuck J, Hadian K, Hornung V, Theis FJ

The model integrates multimodal functional readouts across assay types and cell types into a single predictive architecture, used to predict phenotypic response to compounds and genetic perturbations not seen during training.

Read the preprint on bioRxiv →

Selected publications

Nat Methods

Pertpy: an end-to-end framework for perturbation analysis

Nature Methods · 2025
→
Nat Biotech

Defining and benchmarking open problems in single-cell analysis

Nature Biotechnology · 2025
→
Nat Med

An integrated cell atlas of the lung in health and disease

Nature Medicine · 2023
→
Cell Syst

Machine learning for perturbational single-cell omics

Cell Systems · 2021
→

Open science

Our commercial platform builds on research we have contributed to the open-source community. We believe foundational tools should be accessible to all.

pertpy

End-to-end perturbation analysis framework.

View on GitHub →
Prophet

Research model codebase.

View on GitHub →

Advisors

Fabian Theis

Fabian Theis

Head of Computational Health, Helmholtz Munich

Theis is among the most cited researchers in computational biology and co-created Scanpy, now the standard toolkit for single-cell analysis, used in tens of thousands of published studies. His lab helped define the analytical methods the single-cell genomics field builds on today, and he holds a professorship in mathematical modeling of biological systems at the Technical University of Munich.

Google Scholar →
Chris Sander

Chris Sander

Professor of Cell Biology, Harvard Medical School

Sander built the first computational biology department in cancer research as founding chair at Memorial Sloan Kettering Cancer Center, and co-founded cBioPortal, now the standard resource researchers worldwide use to explore cancer genomics data. His decades-long career spans foundational contributions to protein structure prediction and network biology.

Google Scholar →

Questions about the model or the data behind it?

If you want more detail on the architecture, the training data, or how the published results apply to your program, talk to the team directly.

Talk to the team →