Joint Estimation of Multiple Dependent Gaussian Graphical Models with Applications to Mouse Genomics

Area of Science:

• Statistics
• Computational Biology
• Genomics

Background:

• Gaussian graphical models (GGMs) are essential for depicting conditional dependencies between random variables.
• Existing GGM methods often assume independence among graphs, limiting their application to complex, interdependent datasets.
• Modeling gene expression across multiple tissues requires accounting for both tissue-specific and systemic (whole-body) dependencies.

Purpose of the Study:

• To develop a novel estimator for analyzing a group of dependent Gaussian graphical models.
• To address the limitations of existing methods that assume graph independence.
• To provide a robust framework for modeling complex biological data with inherent cross-tissue dependencies.

Main Methods:

• Proposed a novel estimator for dependent Gaussian graphical models.
• Decomposed the estimation problem into two layers: a systemic layer and a category-specific layer.
• Developed a graphical Expectation-Maximization (EM) technique for joint estimation of both layers.

Main Results:

• Established estimation consistency and selection sparsistency for the proposed estimator.
• Simulation studies demonstrated the superiority of the EM method over a simple one-step approach.
• Applied the technique to mouse genomics data, yielding biologically plausible results.

Conclusions:

• The proposed graphical EM technique effectively estimates multiple dependent Gaussian graphical models.
• The method accurately captures both systemic and category-specific variations in complex biological networks.
• This approach offers a significant advancement for analyzing interdependent biological data, such as multi-tissue gene expression.