Radial clustergrams: visualizing the aggregate properties of hierarchical clusters

Area of Science:

• Computer Science
• Data Visualization
• Bioinformatics

Background:

• Visualizing complex hierarchical data is challenging.
• Existing methods like dendrograms and treemaps have limitations in space efficiency and hierarchical clarity.
• Effective visualization is crucial for fields like molecular diversity and conformational analysis.

Purpose of the Study:

• To introduce a new radial space-filling method for visualizing cluster hierarchies.
• To present the 'radial clustergram' as an improvement over existing visualization techniques.
• To demonstrate the utility of radial clustergrams in scientific applications.

Main Methods:

• Developed a radial space-filling visualization technique called a radial clustergram.
• Arranges clusters in layers representing tree levels.
• Utilizes node adjacency for parent-child relationships and color-coding for node properties.
• Incorporates a fisheye lens for focused exploration within the global context.

Main Results:

• Radial clustergrams offer efficient space utilization compared to dendrograms and hyperbolic trees.
• They provide superior clarity in conveying hierarchical structure and node properties, especially for higher-level nodes, compared to treemaps.
• The method effectively balances global context with focused area exploration.

Conclusions:

• Radial clustergrams are a powerful and versatile tool for visualizing cluster hierarchies.
• This method overcomes limitations of traditional techniques, offering enhanced clarity and space efficiency.
• Demonstrated applicability in molecular diversity and conformational analysis highlights its scientific value.