A robust and accurate single-cell data trajectory inference method using ensemble pseudotime.

Area of Science:

• Computational Biology
• Bioinformatics
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables high-resolution analysis of cellular heterogeneity and development.
• Existing trajectory inference methods often rely on graph-based approaches and pseudotime calculation, which can be sensitive to errors in trajectory construction.

Purpose of the Study:

• To develop a novel and robust framework for trajectory inference using scRNA-seq data.
• To improve the accuracy of pseudotime inference and subsequent trajectory refinement.

Main Methods:

• Proposed the single-cell data Trajectory inference method using Ensemble Pseudotime inference (scTEP).
• scTEP leverages multiple clustering results to infer a robust pseudotime.
• The inferred pseudotime is used to fine-tune the trajectory.

Main Results:

• Evaluated scTEP on 41 real scRNA-seq datasets with known developmental trajectories.
• scTEP outperformed state-of-the-art methods in trajectory inference accuracy across linear and non-linear datasets.
• Demonstrated superior average performance and lower variance compared to existing methods, with enhanced robustness to clustering and dimension reduction errors.

Conclusions:

• Ensemble pseudotime inference using multiple clustering results significantly enhances robustness in trajectory analysis.
• Accurate pseudotime estimation is critical for strengthening the overall accuracy of trajectory inference pipelines.
• The scTEP R package is publicly available for use.