Rmagpie

Rmagpie performs supervised classification of microarray datasets to train classifiers, predict class labels for new observations, identify discriminative gene subsets, and estimate predictive error rates for genomic analyses.

Key Features:

• Classifier Training and Prediction: Supports training and prediction using classifiers including Support Vector Machines (SVMs) and Nearest Shrunken Centroids (NSCs) on labeled microarray datasets.
• Predictive Error Estimation: Provides methodologies to estimate the predictive error rate of trained classifiers for assessment of classification reliability.
• Gene Subset Identification: Reports subsets of genes that discriminate between classes to highlight features driving classification.

Scientific Applications:

• Biomarker discovery: Identification of gene subsets associated with disease classes to support biomarker selection from microarray data.
• Gene expression pattern analysis: Classification of samples to characterize gene expression differences across conditions or phenotypic groups.

Methodology:

Implements Support Vector Machines (SVMs) and Nearest Shrunken Centroids (NSCs), performs predictive error estimation and gene subset selection, and is implemented in R within the Bioconductor project.