cellTree

cellTree infers hierarchical relationships among cells from single-cell RNA sequencing (scRNA-seq) data to model cellular differentiation and underlying gene expression programs.

Key Features:

• Hierarchical Tree Structures: Generates hierarchical tree structures that represent relationships among individual cells and developmental trajectories.
• Latent Gene Group Identification: Identifies latent groups of genes (gene programs) that explain expression patterns across cells.
• Biologically Grounded Models: Employs models grounded in biological principles rather than relying solely on generic dimensionality reduction techniques.
• Visualization of Differentiation Paths: Produces compact tree representations that facilitate visualization of complex cellular differentiation paths and trajectories.

Scientific Applications:

• Cellular differentiation and development: Mapping lineage relationships and developmental trajectories from scRNA-seq data.
• Disease progression and cancer biology: Characterizing cellular heterogeneity and progression-related expression programs in disease contexts such as cancer.
• Identification of regulatory genes and pathways: Discovering key regulatory genes and pathways underlying observed cell-state differences.
• Stem cell and regenerative medicine studies: Resolving differentiation hierarchies and gene programs relevant to stem cell biology and regeneration.

Methodology:

Adapts document analysis techniques from Latent Dirichlet Allocation (LDA) to model scRNA-seq expression as latent gene groups and to infer hierarchical tree structures; the approach handles the high-dimensional nature of scRNA-seq data and produces compact trees reflecting biological relationships between cells.