CellAssign

CellAssign assigns cell types in single-cell RNA sequencing (scRNA-seq) datasets using a probabilistic model that leverages marker gene priors to produce batch-aware cell-type annotations.

Key Features:

• Probabilistic Assignment: Uses a probabilistic model to assign cells to predefined or de novo cell types based on marker gene priors, reducing reliance on unsupervised clustering and manual annotation.
• Scalability: Scales to high-throughput scRNA-seq datasets, enabling automated assignment across thousands of cells.
• Batch and Sample Effect Control: Incorporates mechanisms to control for batch- and sample-specific effects to maintain consistent assignments across datasets and conditions.
• Input Requirements: Requires a binary marker gene-by-cell-type matrix specifying known marker genes for each cell type as prior information.

Scientific Applications:

• Tumor microenvironment profiling: Profiles tumor microenvironments to decompose cellular composition and interactions using marker-based assignments.
• High-grade serous ovarian cancer: Applied to characterize cellular composition in high-grade serous ovarian cancer at single-cell resolution.
• Follicular lymphoma: Applied to analyze follicular lymphoma to resolve tumor and microenvironmental cell types.
• Tissue decomposition and heterogeneity analysis: Enables decomposition of complex tissues into functionally distinct cell types to study cellular heterogeneity.

Methodology:

Integrates a binary marker gene-by-cell-type matrix with scRNA-seq expression data within a probabilistic model to assign each cell to a cell type while accounting for batch- and sample-specific effects.