DTFLOW: Inference and Visualization of Single-cell Pseudotime Trajectory Using Diffusion Propagation.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Accurate inference of cellular developmental trajectories from single-cell data remains a significant challenge in biological research.
• Existing methods often struggle with complex, multi-branching differentiation processes, necessitating more robust computational approaches.

Purpose of the Study:

• To introduce DTFLOW, a novel computational method for accurately determining multi-branching pseudo-temporal cellular trajectories.
• To enhance the accuracy and robustness of single-cell data analysis for inferring developmental processes.

Main Methods:

• DTFLOW employs Bhattacharyya kernel feature decomposition (BKFD) for dimensionality reduction by establishing stationary distributions and utilizing a Bhattacharyya kernel matrix for pseudotime calculation.
• It incorporates Reverse Searching on k-nearest neighbor graph (RSKG) to effectively identify multi-branching cellular differentiation pathways.

Main Results:

• DTFLOW demonstrated superior accuracy and robustness in constructing pseudotime trajectories across four diverse single-cell datasets.
• Comparative analysis against state-of-the-art methods confirmed the enhanced performance of DTFLOW.

Conclusions:

• DTFLOW provides a powerful and accurate tool for inferring complex cellular differentiation trajectories.
• The method addresses key challenges in single-cell data analysis, advancing the understanding of developmental biology.