Gene differential coexpression analysis based on biweight correlation and maximum clique.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Differential coexpression analysis (DCEA) is crucial for understanding gene network changes.
• Pearson correlation is commonly used but sensitive to outliers.
• Robust alternatives are needed for accurate gene expression profile similarity measurement.

Purpose of the Study:

• To introduce Biweight Midcorrelation as a robust measure for gene expression similarity in DCEA.
• To develop a novel DCEA approach utilizing Biweight Midcorrelation and a half-thresholding strategy.
• To improve the identification of biologically relevant gene coexpression patterns.

Main Methods:

• Calculation of biweight midcorrelation coefficients for all gene pairs.
• Application of a 'half-thresholding' strategy to filter non-informative correlation pairs.
• Differential coexpression value calculation and maximum clique analysis on identified gene subsets.

Main Results:

• The novel Biweight Midcorrelation-based DCEA approach demonstrated superior performance on simulated data compared to three existing methods.
• Maximum clique analysis revealed additional discoveries when applied to gene subsets identified by the new method.
• The approach showed potential for identifying novel gene relationships relevant to diseases like Type 2 Diabetes (T2D).

Conclusions:

• Biweight Midcorrelation offers a robust alternative to Pearson correlation for gene differential coexpression analysis.
• The proposed DCEA method enhances the accuracy and discovery power of gene network analysis.
• This approach facilitates the identification of novel gene interactions and disease-related genes.