Asterias

Asterias analyzes gene expression and array comparative genomic hybridization (aCGH) data to perform normalization, differential expression, class and survival prediction model building, and pathway-oriented annotation for high-throughput genomic studies.

Key Features:

• Supported data types: Gene expression and array comparative genomic hybridization (aCGH) datasets.
• Statistical methods: Validated statistical techniques including methods inspired by random forest algorithms for classification and variable importance estimation.
• Parallel computing: Leverages multicore CPUs and computing clusters to accelerate computation.
• Analyses supported: Data normalization and preprocessing, differential gene expression analysis, class prediction and survival prediction model building, and aCGH analysis.
• Variable selection: Provides measures of variable importance to aid selection of minimal gene sets for classification.
• Biological annotation integration: Integrates annotations from PubMed, Gene Ontology, KEGG pathways, and Reactome pathways for functional interpretation.

Scientific Applications:

• Differential expression analysis: Identification of genes with significant expression changes across experimental conditions.
• Predictive modeling: Construction of class and survival prediction models from gene expression data.
• Feature selection for diagnostics: Selection of minimal gene sets using variable importance measures for sample classification.
• aCGH analysis: Analysis of array comparative genomic hybridization data for copy-number aberration detection.
• Pathway and functional interpretation: Mapping genes to Gene Ontology, KEGG, and Reactome pathways and linking to PubMed references.
• High-dimensional microarray analysis: Handling datasets with many more variables than observations and supporting multi-class classification tasks.

Methodology:

Implements statistical techniques including random forest algorithms that provide variable importance measures and support multi-class classification in high-dimensional (p>>n) microarray data, and uses parallel processing on multicore CPUs and computing clusters to accelerate computation.