GeneCoNet

GeneCoNet constructs patient-specific gene correlation networks to identify prognostic gene-gene interactions in cancer.

Key Features:

• Patient-specific networks: Builds gene correlation networks tailored to each cancer patient's genetic profile.
• Deviation from normal: Identifies gene-gene relationships that significantly deviate from those observed in normal samples.
• Pairwise prognostic testing: Assesses prognostic significance using Cox proportional hazards regression and log-rank tests for every possible gene pair.
• TCGA integration: Leverages tumor samples from The Cancer Genome Atlas (TCGA) and generates networks based on TCGA sample IDs.
• Cross-cancer analysis: Applied across six cancer types to reveal prognostic gene correlations specific to each cancer.
• Subnetwork analysis: Detects subnetworks containing prognostic gene pairs and relates their size and complexity to patient survival times.

Scientific Applications:

• Individualized network inference: Generate individualized gene correlation networks for single patients using TCGA sample IDs.
• Prognostic biomarker discovery: Identify prognostic gene-gene correlations and patient-specific oncogenic interactions.
• Tumor-versus-normal comparison: Detect gene-gene relationships in tumors that deviate from normal-sample correlations.
• Survival analysis of gene pairs: Perform pairwise survival testing with Cox proportional hazards regression and log-rank tests.
• Pan-cancer specificity assessment: Compare prognostic gene correlations across six cancer types to find cancer-type-specific pairs.
• Network-based prognosis interpretation: Analyze how subnetwork size and complexity containing prognostic pairs associate with shorter survival times.

Methodology:

Constructs gene correlation networks for each patient by identifying gene-gene relationships that deviate from normal samples; evaluates prognostic significance of every possible gene pair using Cox proportional hazards regression and log-rank tests on TCGA tumor samples referenced by TCGA sample IDs.