scShaper

scShaper infers linear pseudotemporal trajectories from single-cell RNA-seq (scRNA-seq) data using Shortest Hamiltonian path PERmuted clustering to estimate discrete pseudotimes and produce continuous smooth pseudotimes for analysis of dynamic processes such as cell differentiation.

Key Features:

• Shortest Hamiltonian path PERmuted clustering: Estimates discrete pseudotimes using a permutation-based shortest Hamiltonian path clustering approach.
• Ensemble smoothing: Aggregates an ensemble of discrete pseudotimes to generate continuous smooth pseudotime values.
• Nonlinear trajectory robustness: Maintains accurate ordering for trajectories arising from nonlinear mathematical models where principal curves can fail.
• Trajectory-based differential expression: Identifies differentially expressed genes along inferred pseudotemporal trajectories.
• Benchmarking: Demonstrated superior accuracy in cell ordering relative to state-of-the-art trajectory inference methods.
• Computational efficiency and hyperparameter requirements: Operates with computational efficiency and minimal hyperparameter tuning.
• Implementation: Provided as an R package.

Scientific Applications:

• Pseudotemporal ordering in scRNA-seq: Reconstruction of linear cell trajectories from single-cell RNA-seq datasets.
• Cell differentiation and developmental studies: Analysis of dynamic processes such as differentiation and lineage progression.
• Trajectory-based differential expression analysis: Detection of genes with expression changes along inferred pseudotime.

Methodology:

Uses Shortest Hamiltonian path PERmuted clustering to compute discrete pseudotimes and aggregates an ensemble of discrete pseudotimes to produce continuous smooth pseudotime values; comparisons include principal curves and benchmarking against state-of-the-art trajectory inference methods.