oppar

oppar detects over- and under-expressed outlier genes in expression datasets by implementing modified Cancer Outlier Profile Analysis (mCOPA) to enable comprehensive outlier-based feature selection.

Key Features:

• Outlier Detection: Identifies both up-regulated and down-regulated gene expression outliers for informative feature selection.
• Comparison with Other Methods: Compares outlier-based selection to differential expression and variance-based feature selection and selects more-informative features.
• Clinical Relevance: Assigns samples to clinically annotated subtypes with improved recovery of clinically relevant subgroups.
• Pathway Analysis Integration: Supports gene set enrichment and pathway analyses to explore disrupted molecular mechanisms in individual tumors.
• Tumor Suppressor Identification: Detects under-expressed outliers to aid identification of known and novel tumor suppressor genes, with validation against Oncomine and the Cancer Gene Index.
• Application to Prostate Cancer: Demonstrated on prostate cancer expression data for clustering, pathway analysis, and tumor suppressor identification.

Scientific Applications:

• Cancer subtype discovery: Reveals novel and clinically relevant cancer subtypes by analyzing heterogeneity within expression datasets.
• Molecular mechanism exploration: Combines outlier detection with pathway and gene set enrichment analyses to investigate disrupted pathways in individual tumors.
• Clinical subtype classification: Improves accuracy of assigning samples to clinically annotated subtypes for diagnostic and research applications.

Methodology:

Uses a modified COPA algorithm to detect over- and under-expressed outliers, compares results to differential expression and variance-based feature selection, and performs clustering, gene set enrichment and pathway analyses with validation against Oncomine and the Cancer Gene Index.