CyStainer: A transformer-based variational autoencoder for robust marker imputation in high-parameter cytometry

Authors

Ivanov, K., Moussawy, M. A., Kirk, F., Samuli, R., Lohi, O., Olsen, L., Modvig, S., Hautamäki, V., Heinäniemi, M.

Abstract

High parameter cytometry is essential for clinical diagnostics through precise immune cell profiling, improved patient stratification, and monitoring, while also enhancing the understanding of cellular responses in disease and therapeutic contexts. The amount of cytometry data is growing fast, and with that, the need to merge different datasets for unified analysis. Here, we present CyStainer, a transformer-based variational autoencoder that demonstrates competitive or superior performance to existing methods on several key tasks related to marker prediction. As a key novelty, we demonstrate that CyStainer can impute markers without having a set of shared backbone markers. We performed several benchmarks using real-world FACS, CyTOF, InfinityFlow and CITE-seq datasets to show that CyStainer is a robust and flexible tool for panel merging, marker imputation, dataset integration and virtual staining of unseen samples.

Preprint server: bioRxiv
The authors list and abstract were imported from bioRxiv on 02 Jul 2026.

Stats

1-terrible, 9-excellent. How would you rate this preprint? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.