Orange

Orange performs data mining, visualization, and machine-learning analyses of biological datasets, emphasizing cancer microarray data for gene-level pattern discovery and classification.

Key Features:

• Data mining and visualization (VizRank, radviz): Incorporates VizRank and radviz to score and rank visualizations by class separation for exploring patterns and gene interactions in microarray datasets.
• Outlier detection and feature scoring: Provides outlier detection and gene feature scoring to identify significant genes and anomalies in microarray measurements.
• Classification via visualization: Projects data into interpretable visualizations that can differentiate cancer types using small subsets of genes with performance comparable to supervised data mining techniques.
• Machine learning integration: Includes machine learning components for building predictive models to classify tissue samples and support discovery of gene-level patterns.

Scientific Applications:

• Cancer microarray analysis: Analyzing cancer microarray datasets to identify differential gene expression patterns and classify tumor types.
• Gene interaction discovery: Revealing gene interactions and candidate biomarkers through visualization ranking and feature scoring.
• Predictive modeling of tissue samples: Developing classifiers for tissue sample classification and biomarker-driven diagnostics.

Methodology:

Uses VizRank and radviz for visualization scoring and ranking based on class separation; performs outlier detection and gene feature scoring; and integrates supervised machine learning components for classification and predictive modeling, with comparisons to state-of-the-art supervised data mining techniques.