Using matrix of thresholding partial correlation coefficients to infer regulatory network.

Area of Science:

• Genomics
• Systems Biology
• Bioinformatics

Background:

• DNA arrays enable simultaneous measurement of thousands of gene expression levels across various conditions.
• Gene expression data provides insights into complex biological processes.
• Inferring regulatory networks from expression profiles is crucial for understanding cellular mechanisms.

Purpose of the Study:

• To propose a novel method for regulatory network inference from gene expression profiles.
• To apply the method to Saccharomyces cerevisiae gene expression data.
• To validate the inferred network against existing experimental results.

Main Methods:

• A matrix of thresholding partial correlation coefficients (MTPCC) method was developed.
• The method incorporates hierarchical cluster analysis and cluster boundary establishment.
• Gene expression data from Saccharomyces cerevisiae was analyzed using MTPCC, partial correlation coefficient matrix (PCCM), and permutation testing.

Main Results:

• 2467 genes were grouped into 12 clusters using hierarchical clustering.
• The expression profiles of these clusters were analyzed for partial correlations.
• An undirected dependency graph (UDG) representing the regulatory network of S. cerevisiae was generated.

Conclusions:

• The proposed MTPCC method effectively infers gene regulatory networks from expression data.
• The inferred network for S. cerevisiae shows consistency with established experimental findings.
• This approach offers a robust tool for dissecting biological regulatory systems.