sigFeature

sigFeature implements a feature selection algorithm for binary classification that combines support vector machine recursive feature elimination (SVM-RFE) with t-statistic to identify features that are both differentially significant between two classes and capable of achieving high classification accuracy.

Key Features:

• Integrated Feature Selection: Combines SVM-RFE and t-statistic to select features that satisfy both statistical significance and classification performance criteria simultaneously.
• Binary Classification Focus: Provides automatic feature selection specifically optimized for two-class classification problems.
• Gene Expression Compatibility: Developed and validated using microarray data sets downloaded from Gene Expression Omnibus (GEO).

Scientific Applications:

• Differential Gene Expression Analysis: Identifies differentially significant genes between two biological conditions or treatment groups from gene expression data.
• Biomarker Discovery: Selects compact sets of features that achieve higher classification accuracy for downstream biological signature prediction.
• Gene Set Enrichment Analysis: Supports downstream evaluation of biological signatures from selected feature sets.

Methodology:

The algorithm integrates support vector machine recursive feature elimination (SVM-RFE) with t-statistic in a feature selection process where selected features are evaluated for differential significance between two classes and classification accuracy.