Detecting Boolean Asymmetric Relationships With a Loop Counting Technique and its Implications for Analyzing...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Traditional gene-gene relationship analysis often focuses on symmetric correlations (positive/negative).
• Biclustering methods typically identify subsets of genes with correlated expression across samples.
• Genes can also exhibit asymmetric relationships, akin to 'if-then' logic.

Purpose of the Study:

• To develop a general method for detecting biclusters enriched for Boolean Asymmetric Relationships (BARs) in gene expression data.
• To identify asymmetric gene-gene interactions driving biological heterogeneity in specific sample subsets.
• To explore the utility of BAR-biclusters in single-cell RNA sequencing data.

Main Methods:

• Developed a novel biclustering approach to detect Boolean Asymmetric Relationships (BARs).
• Applied the method to single-cell RNA sequencing (scRNA-seq) data.
• Compared findings from BAR-biclusters with traditional boolean-symmetric biclusters.

Main Results:

• Statistically significant BAR-biclusters were identified, containing unique information not found in traditional biclusters.
• BAR-biclusters highlighted different subsets of cells and gene pathways compared to symmetric methods.
• Combining BAR and symmetric signals improved linear classifier performance.

Conclusions:

• BAR-biclusters offer a novel perspective on gene-gene interactions, capturing asymmetric regulatory effects.
• This method enhances the understanding of biological heterogeneity, particularly in complex datasets like scRNA-seq.
• Integrating asymmetric and symmetric signals leads to more robust predictive models in genomics.