safe

safe performs two-stage permutation-based testing of functional gene categories in gene expression experiments to assess category-level significance while accounting for unknown correlations among genes.

Key Features:

• Permutation-Based Methodology: Employs a two-stage permutation-based approach that enables valid testing of gene categories and accounts for unknown correlations among genes.
• Flexibility Across Experimental Designs: Applicable to 2-sample and multi-class comparisons and simple linear regressions and extensible to other experimental setups via user-defined functions.
• Error Rate Estimation: Provides permutation-based estimation of error rates to control type I error in gene category testing.
• Integration with Gene Ontology and Protein Family database: Leverages Gene Ontology and the Protein Family database to categorize genes for biological interpretation.

Scientific Applications:

• High-throughput genomic and proteomic experiments: Enables direct testing of functional gene categories in large-scale expression studies.
• Microarray analysis of human lung carcinomas: Applied to a microarray dataset of human lung carcinomas to identify significant gene categories across multiple comparisons.
• Tumor vs. Normal Tissue: Identifying gene categories that differentiate tumor tissues from normal ones.
• Multiple Tumor Subtypes: Distinguishing gene categories associated with different tumor subtypes based on expression profiles.
• Survival Times: Associating gene categories with patient survival times to identify potential prognostic markers.

Methodology:

Uses a two-stage permutation-based testing framework with permutation-based estimation of error rates, accommodates unknown correlations among genes, and supports 2-sample, multi-class and simple linear regression designs while allowing extension via user-defined functions.