Generating weighted and thresholded gene coexpression networks using signed distance correlation

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Functional gene annotation is crucial for understanding biological systems, yet many genes remain unannotated.
• Gene coexpression networks are valuable tools for inferring gene function and biological relationships.
• Existing methods for constructing gene coexpression networks, such as using Pearson correlation, may lose information by converting continuous correlation values into unweighted networks.

Purpose of the Study:

• To develop a principled method for constructing weighted gene coexpression networks.
• To leverage signed distance correlation for a more informative network construction.
• To compare the performance of signed distance correlation-based weighted networks against existing methods.

Main Methods:

• Developed a method to construct weighted gene coexpression networks using signed distance correlation.
• Weighted edges were assigned to gene pairs with correlation values exceeding a specified threshold.
• Analyzed gene expression data from multiple organisms.

Main Results:

• Networks constructed using signed distance correlation-based weighted approach are more stable.
• These networks capture significantly more biological information than those derived from Pearson correlation.
• Signed distance correlation-based weighted networks outperform unweighted networks derived from the same metric.

Conclusions:

• Weighted gene coexpression networks built with signed distance correlation offer improved stability and biological insight.
• This method provides a more comprehensive representation of gene relationships compared to traditional unweighted networks.
• The approach is broadly applicable to network construction in various biological and non-biological domains.