Population-Level Cell Trajectory Inference Based on Gaussian Distributions.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Inferring developmental trajectories from single-cell data is a key challenge.
• RNA velocity analysis advances trajectory studies but struggles with high-dimensional, noisy single-cell RNA sequencing data.
• Existing methods often neglect cell distribution characteristics, limiting their performance.

Purpose of the Study:

• Introduce CPvGTI, a novel Gaussian distribution-based method for robust single-cell trajectory inference.
• Improve the accuracy and reliability of developmental trajectory analysis, especially with complex datasets.
• Address limitations of current methods in handling high-dimensional and noisy single-cell data.

Main Methods:

• Utilize a Gaussian mixture model, optimized via the Expectation-Maximization algorithm, to define cell populations.
• Integrate RNA velocity with Gaussian Process Regression for analyzing differentiation trajectories.
• Evaluate CPvGTI against state-of-the-art methods using diverse simulated and real single-cell datasets.

Main Results:

• CPvGTI demonstrates superior performance in pseudo-time prediction and structural reconstruction compared to existing methods in simulation studies.
• The method successfully identified novel branch trajectories in human forebrain and mouse hematopoiesis datasets.
• CPvGTI shows enhanced accuracy in capturing complex developmental dynamics.

Conclusions:

• CPvGTI offers a significant advancement in single-cell trajectory inference, particularly for complex and noisy biological data.
• The Gaussian distribution-based approach effectively models cell populations and their dynamics.
• This method provides a more reliable tool for understanding developmental processes from single-cell data.