Gene regulation network inference with joint sparse Gaussian graphical models.

Area of Science:

• Systems biology
• Bioinformatics
• Genomics

Background:

• Understanding gene regulation is crucial in systems biology.
• Microarray data allows genome-wide expression profiling under diverse conditions.
• Gaussian graphical models (GGMs) are used to infer gene networks based on gene expression dependencies.

Purpose of the Study:

• To develop a method for inferring multiple GGMs that capture condition-specific gene regulatory networks.
• To introduce a novel class of nonconvex penalty functions for joint sparsity in multiple GGM estimation.
• To apply the developed method to real-world gene expression data for pathway identification.

Main Methods:

• Proposed a class of nonconvex penalty functions for estimating multiple GGMs.
• Implemented a flexible joint sparsity constraint for network inference.
• Utilized simulation studies to validate the properties of the proposed penalty functions.
• Applied the method to gene expression data from the GenCord Project.

Main Results:

• The proposed nonconvex penalty functions demonstrated effectiveness in simulation studies.
• The method successfully identified prominent biological pathways across different experimental conditions.
• The joint sparsity constraint enabled the estimation of condition-specific networks.

Conclusions:

• The developed method provides a robust approach for inferring multiple gene regulatory networks.
• This technique enhances the understanding of biological pathways by analyzing condition-specific gene dependencies.
• The application to GenCord data highlights the practical utility of the method in systems biology research.