Parameterization of asymmetric sigmoid functions in weighted gene co-expression network analysis

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Examining gene expression at the genomic level offers more accurate insights than individual gene analysis.
• Weighted Gene Co-expression Network Analysis (WGCNA) is a common method for clustering transcriptomic data, utilizing a power function for soft thresholding.
• Standard WGCNA power functions can overemphasize minor correlations in gene expression.

Purpose of the Study:

• To present a systematic procedure for parameterizing asymmetric sigmoid functions for soft thresholding in WGCNA.
• To offer a more robust alternative to standard power functions in WGCNA.
• To facilitate the application of asymmetric sigmoid functions in gene expression network analysis.

Main Methods:

• Developed a systematic parameterization procedure for asymmetric sigmoid functions.
• Applied the enhanced soft thresholding method within the WGCNA framework.
• Validated the approach using transcriptomic datasets from COVID-19, yeast, and E. coli.

Main Results:

• The proposed parameterization method simplifies the use of asymmetric sigmoid functions in WGCNA.
• The approach was successfully applied to diverse biological datasets, including COVID-19.
• The results demonstrated the generation of biologically plausible gene co-expression modules.

Conclusions:

• The systematic parameterization of asymmetric sigmoid functions provides a valuable alternative for soft thresholding in WGCNA.
• This method enhances the biological interpretability of gene co-expression networks.
• The approach is effective across various species and disease contexts, including COVID-19 research.