A Pseudotime-Dependent TWAS Framework Identifies Disease Genes along Cell Developmental Paths.

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Transcriptome-wide association studies (TWAS) integrate gene expression and genome-wide association study (GWAS) data to link genes with disease risk.
• Single-cell expression data allows for finer resolution analysis, potentially revealing genetic effects missed by bulk TWAS.
• Current single-cell TWAS methods often assign associations to discrete cell types, overlooking continuous cellular processes and dynamic gene effects.

Purpose of the Study:

• To develop a novel framework, pseudotime-dependent TWAS (pt-TWAS), that models gene expression as a continuous function of pseudotime.
• To capture dynamic gene effects along cellular developmental trajectories and improve the identification of causal cell stages for disease risk.
• To enhance statistical power and accuracy compared to existing single-cell TWAS methods.

Main Methods:

• Developed pt-TWAS, a computational framework modeling gene expression continuously across pseudotime.
• Utilized shared genetic effects across cell stages to increase statistical power in simulations.
• Constructed confidence bands for gene effect curves to identify causal cell stages.

Main Results:

• pt-TWAS demonstrated higher statistical power than existing single-cell TWAS methods in extensive simulations.
• The method successfully identified known risk genes for B-cell acute lymphoblastic leukemia.
• pt-TWAS pinpointed specific cell stages relevant to the identified genetic mechanisms of disease risk.

Conclusions:

• pt-TWAS offers a powerful approach to dissect gene-disease associations at a fine-grained, cell-stage-specific level.
• The framework effectively models dynamic gene expression along developmental trajectories, overcoming limitations of discrete cell type mapping.
• This method advances the understanding of genetic mechanisms underlying diseases by revealing cell-stage-specific effects.