TooManyCells

TooManyCells identifies and hierarchically organizes transcriptionally similar cells from single-cell transcriptomics data to enable multiresolution analysis of cellular diversity.

Key Features:

• Graph-Based Algorithms: Employs graph-based methodologies to identify and visualize cell clades from single-cell data.
• Orthogonal Visualization Model: Provides a visualization model that is orthogonal to traditional dimensionality-reduction methods for alternative views of cellular relationships.
• Matrix-Free Divisive Hierarchical Spectral Clustering: Utilizes a matrix-free divisive hierarchical spectral clustering approach to produce hierarchical, multiresolution partitions of cells.
• Multiresolution Exploration: Enables simultaneous examination of rare and common cell populations and detection of subtle variations within cellular states.

Scientific Applications:

• Transcriptional similarity analysis: Identifying and visualizing transcriptionally similar cells across resolutions, with reported improvements compared to popular clustering and visualization algorithms.
• Leukemic T cell studies: Analysis of drug-resistance acquisition in leukemic T cells using single-cell transcriptomic datasets.
• Cellular dynamics and heterogeneity: Exploration of cellular dynamics and heterogeneity by leveraging existing single-cell transcriptomic datasets.

Methodology:

Combines graph-based clustering with hierarchical spectral clustering techniques, specifically a matrix-free divisive hierarchical spectral clustering approach that operates without relying on fixed resolutions.