A biology-driven deep generative model for cell-type annotation in cytometry

Cytometry enables precise single-cell phenotyping within heterogeneous populations. These cell types are traditionally annotated via manual gating, but this method lacks reproducibility and sensitivity to batch effect. Also, the most recent cytometers-spectral flow or mass cytometers-create rich and high-dimensional data whose analysis via manual gating becomes challenging and time-consuming. To tackle these limitations, we introduce Scyan , a Single-cell Cytometry Annotation Network that automatically annotates cell types using only prior expert knowledge about the cytometry panel. For this, it uses a normalizing flow-a type of deep generative model-that maps protein expressions into a biologically relevant latent space. We demonstrate that Scyan significantly outperforms the related state-of-the-art models on multiple public datasets while being faster and interpretable. In addition, Scyan overcomes several complementary tasks, such as batch-effect correction, debarcoding and population discovery. Overall, this model accelerates and eases cell population characterization, quantification and discovery in cytometry.

Automated summary

Cytometry technology allows for precise single-cell phenotyping within heterogeneous populations. However, traditional manual gating for cell type annotation lacks reproducibility and sensitivity to batch effect. In this study, we introduce Scyan, a novel Single-cell Cytometry Annotation Network that automatically annotates cell types using prior expert knowledge. We demonstrate that Scyan outperforms current state-of-the-art models on multiple public datasets and provides faster and interpretable results. Scyan also addresses complementary tasks such as batch-effect correction, debarcoding, and population discovery.