CytoVI: deep generative modeling of antibody-based single cell data.

CytoVI enables comprehensive analysis of antibody-based single-cell data, accurately modeling 350 proteins and revealing disease-related immune cell states.

• Why it matters: Single-cell antibody technologies face challenges from technical noise, batch effects, and limited panels, hindering reliable interpretation and discovery in clinical and research settings.
• What they did: The authors developed CytoVI, a probabilistic generative model that produces cell embeddings, imputes missing data, tests differential expression, and automates cell annotation, applied to large B cell and lymphoma datasets.
• The result: CytoVI successfully identified proteins linked to immunoglobulin class-switching and uncovered disease-associated T cell states, facilitating deeper insights into immune cell dynamics and disease mechanisms.