Vis-SPLIT: Interactive Hierarchical Modeling for mRNA Expression Classification

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Accurate patient classification is crucial for cancer research and treatment.
• Analyzing complex gene expression data presents significant challenges.

Purpose of the Study:

• To introduce Vis-SPLIT, an interactive visual analytics tool for partitioning individuals based on gene signatures.
• To enable users to build cancer classification models by exploring gene expression and correlation data.

Main Methods:

• Development of Vis-SPLIT, an interactive visual analytics framework.
• Utilizing visualization components to reveal gene expression and correlation patterns.
• Interactive exploration and exploitation of visual separations for model building.
• Case study demonstration and usability evaluation with domain experts.

Main Results:

• Vis-SPLIT effectively classifies patients based on their genetic signatures.
• The tool provides valuable insights into RNA sequencing data.
• Demonstrated superior performance compared to an existing classification system.
• Usability evaluation confirmed its effectiveness with domain experts.

Conclusions:

• Vis-SPLIT offers an effective approach for patient classification using gene signatures.
• The tool enhances the understanding of RNA sequencing data in cancer research.
• Interactive visualization aids in decision-making for partitioning and classification.