Uncovering pseudotemporal trajectories with covariates from single cell and bulk expression data

Area of Science:

• Computational Biology
• Genomics
• Statistical Modeling

Background:

• Pseudotime algorithms extract temporal dynamics from static data, useful for studying cellular differentiation and tumor progression.
• Existing methods often assume data homogeneity, limiting their application to complex, large-scale datasets.
• There is a need for advanced pseudotime methods that can incorporate biological variability.

Purpose of the Study:

• To develop a novel statistical framework for pseudotemporal modeling that accounts for covariate-modulated trajectories.
• To extend the applicability of pseudotime analysis beyond single-cell genomics.
• To provide a method for analyzing dynamic biological processes in complex datasets.

Main Methods:

• A hybrid regression-latent variable model was developed.
• The framework integrates covariate information to modulate pseudotime trajectories.
• The model was applied to both single-cell and bulk gene expression datasets.

Main Results:

• The novel framework successfully recovered known and identified novel covariate-pseudotime interaction effects.
• The approach demonstrated its ability to handle complex biological datasets.
• The model effectively captures how external factors influence biological trajectories.

Conclusions:

• The proposed statistical framework offers a powerful extension to existing pseudotime methods.
• This approach enables more nuanced analysis of dynamic biological processes influenced by various factors.
• The method broadens the application of pseudotemporal modeling to diverse biological research areas.