Hierarchical sets: analyzing pangenome structure through scalable set visualizations.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• The increasing number of microbial genome sequences leads to larger pangenomes.
• Existing pangenome visualization tools are inadequate for current data scales.
• Novel approaches are needed to translate large pangenome data into biological knowledge.

Purpose of the Study:

• To explore scalable set visualization techniques for pangenome analysis.
• To develop new methods for visualizing large and complex pangenome datasets.
• To enhance the understanding of microbial genome evolution and relationships.

Main Methods:

• Development of a hierarchical clustering algorithm utilizing set arithmetics.
• Implementation of composite dendrograms and icicle plots for visualizing set sizes.
• Application of hierarchical edge bundles to identify outlying elements and potential horizontal gene transfers.

Main Results:

• A novel hierarchical clustering approach optimizes intersection sizes within pangenomes.
• Visualizations effectively display pangenome and core genome size evolution across a hierarchy.
• Hierarchical edge bundles highlight horizontal gene transfers and non-hierarchical genomic relationships.
• The method was successfully applied to a dataset of 113 Escherichia and Shigella genomes.

Conclusions:

• Hierarchical set visualization offers a powerful new tool for pangenome analysis.
• The approach enhances the interpretation of microbial genome evolution and gene flow.
• The `hierarchicalSets` R package provides an accessible implementation of these methods.