iterativeBMA

iterativeBMA implements iterative Bayesian model averaging for variable selection and two-class classification of microarray data.

Key Features:

• Bayesian Model Averaging: Applies Bayesian model averaging to combine multiple models and account for model uncertainty by weighting model contributions with posterior probabilities.
• Iterative Refinement: Uses an iterative model-selection procedure to refine variable selection and improve two-class classification performance on high-dimensional data.
• Two-class Microarray Classification: Focuses on classification of two-class microarray samples and the selection of predictive genes for class discrimination.
• Integration with Bioconductor (MeV+R): Provides interoperability via MeV+R integration with Bioconductor packages.
• Extensibility: Enables extension to additional Bioconductor packages for broader analytical workflows.

Scientific Applications:

• Two-class sample classification: Distinguishes between two phenotypic or experimental classes in microarray studies.
• Differential expression and biomarker discovery: Identifies differentially expressed genes and candidate biomarkers associated with class differences.
• Biological process and disease mechanism analysis: Supports investigation of underlying biological processes or disease mechanisms through selected predictive variables.

Methodology:

iterativeBMA iteratively evaluates multiple models, weights each model by its posterior probability, and averages model predictions to produce classification results and measures of variable importance.