oncomix

oncomix identifies mRNAs overexpressed in subsets of tumors relative to normal tissue using an unsupervised statistical approach to capture transcriptional heterogeneity.

Key Features:

• Unsupervised statistical modeling: Models mRNA expression heterogeneity to detect subgroup-specific overexpression without prior subtype labels.
• Paired tumor–normal analysis: Compares tumor and matched normal samples from the same tissue across patients.
• Sample size suitability: Intended for use with datasets of paired samples, ideally more than 15 pairs.
• Oncogene candidate identification: Identifies mRNA-level oncogene candidates that are overexpressed in specific tumor subsets relative to normal tissue.
• Transcriptional heterogeneity focus: Captures differences in mRNA expression between previously unrecognized tumor subtypes.
• Associative analyses: Can reveal associations between gene overexpression and intronic DNA methylation, downstream cell cycle gene expression, and patient survival, as demonstrated for CBX2 in breast tumors.

Scientific Applications:

• Oncogene discovery: Identify candidate oncogenes that are overexpressed only in subsets of tumors compared with matched normal tissue.
• Subtype-specific biomarker discovery: Detect transcriptional differences that define previously unrecognized tumor subtypes in paired datasets.
• Epigenetic association studies: Link overexpressed genes to intronic DNA methylation, exemplified by CBX2 in breast cancer.
• Functional and clinical correlation: Associate gene overexpression with cell cycle gene programs and clinical outcomes such as 5-year survival (demonstrated for CBX2).
• Therapeutic target prioritization: Prioritize subset-specific overexpressed genes for experimental validation and potential therapeutic intervention.

Methodology:

Uses an unsupervised statistical approach to model mRNA expression heterogeneity in paired tumor–normal datasets and identify genes overexpressed in tumor subsets relative to matched normal tissue.