HGC

HGC performs hierarchical clustering on single-cell datasets to reveal nested cellular heterogeneity across multiple resolutions.

Key Features:

• Hierarchical Clustering with Linear Time Complexity: Constructs a hierarchical tree on a shared nearest neighbor graph with linear time complexity for scalability to large datasets.
• Graph-Based Input: Accepts an adjacency matrix (G) of edge weights in dgCMatrix format (R Matrix package), where zero indicates no connection between node pairs.
• Multiresolution Exploration: Provides hierarchical information that enables investigation of cell states at multiple levels of granularity.
• State-of-the-Art Accuracy: Demonstrates state-of-the-art accuracy on benchmark single-cell datasets.

Scientific Applications:

• Single-Cell RNA Sequencing (scRNA-seq): Identifies distinct cell populations and developmental trajectories from scRNA-seq gene expression data.
• Cancer Research: Characterizes tumor heterogeneity by resolving composition and nested subpopulations of cancer cells.
• Developmental Biology: Analyzes cellular differentiation by revealing nested lineage relationships among cell types.

Methodology:

Operates on a shared nearest neighbor graph derived from single-cell data using an adjacency matrix (dgCMatrix) to construct a hierarchical tree that encodes nested relationships while achieving linear time complexity.