Flexible copula model for integrating correlated multi-omics data from single-cell experiments.

Area of Science:

• Computational Biology
• Genomics
• Biostatistics

Background:

• Single-cell multi-omics technologies generate integrated datasets (e.g., gene expression, DNA methylation, chromatin accessibility).
• Analyzing associations between these distinct data types, especially their changes across covariates like cell types, presents statistical challenges due to differing marginal distributions.

Purpose of the Study:

• To develop a flexible statistical framework for joint analysis of single-cell multi-omics data.
• To model covariate-dependent correlation structures independent of marginal distributions.
• To investigate dynamic relationships within multi-omics data during key developmental processes.

Main Methods:

• Proposed a copula-based framework to model joint distributions of multi-omics data.
• The framework accommodates diverse marginal distributions, including discrete, continuous, and zero-inflated types.
• Validated the approach using simulation studies and applied it to mouse gastrulation data.

Main Results:

• The copula framework effectively models covariate-dependent associations between different omics layers.
• Demonstrated the ability to integrate diverse data types, including gene expression, methylation, and accessibility.
• Identified dynamic relationships between these molecular layers during mouse gastrulation.

Conclusions:

• The proposed copula-based framework offers a flexible and robust solution for integrative single-cell multi-omics analysis.
• This approach facilitates the study of complex biological systems by revealing dynamic, covariate-dependent molecular interactions.
• The methodology is applicable to various biomedical studies involving integrated omics data.