scHiCPTR

scHiCPTR infers pseudotime trajectories from single-cell Hi-C contact matrices using an unsupervised graph-based pipeline to order cells along developmental processes.

Key Features:

• Unsupervised workflow: Operates without predefined labels or training datasets for de novo pseudotime inference from single-cell Hi-C data.
• Comprehensive pipeline: Integrates imputation and embedding, graph construction, dual graph refinement, pseudotime calculation, and result visualization.
• Imputation and embedding: Processes sparse and noisy contact matrices to produce lower-dimensional representations for downstream analysis.
• Graph construction: Builds a graph representation of cells based on embedded contact-matrix features.
• Dual graph refinement: Applies two parallel graph pruning procedures to reduce spurious cell links and reinforce global developmental directionality.
• Pseudotime calculation: Computes temporal ordering of cells along inferred developmental trajectories.
• Trajectory topology handling: Supports multiple topologies including linear, bifurcated, and circular trajectories.
• Biological validation: Comparative analyses report superior pseudotime inference performance and biologically meaningful trajectories relative to other methods.

Scientific Applications:

• Cellular differentiation analysis: Ordering cells by pseudotime to study differentiation processes using single-cell Hi-C.
• Chromosomal organization dynamics: Investigating temporal changes in 3D genome conformation across developmental processes.
• Complex trajectory systems: Characterizing non-linear and multi-branch developmental programs where linear progression is insufficient.

Methodology:

Imputation and embedding of contact matrices; graph construction from embedded representations; dual graph refinement via graph pruning to remove spurious links and establish global directionality; pseudotime calculation from the refined graph; and visualization of inferred trajectories.