SCFA

SCFA performs consensus factor analysis to integrate multi-omics data and identify molecular cancer subtypes and predict patient risk scores correlated with survival outcomes and vital status.

Key Features:

• Noise Reduction: Filters out noisy signals to emphasize consistent molecular patterns for more reliable subtype identification.
• Data Integration: Integrates multi-omics data across multiple biological levels to capture complementary molecular signals.
• Risk Prediction: Predicts patient risk scores that correlate with survival outcomes and vital status.
• Performance Superiority: Demonstrated superior performance on 7,973 samples from The Cancer Genome Atlas (TCGA) across 30 cancer types, identifying novel subtypes with significantly different survival profiles.
• Improved Accuracy with Data Integration: Subtype discovery and risk prediction accuracy improve when additional data types are incorporated.

Scientific Applications:

• Cancer molecular subtyping: Refines molecular subtype definitions to reveal tumour heterogeneity relevant to prognosis.
• Prognostic modeling and risk stratification: Produces risk scores for survival analysis and patient stratification.
• Discovery of clinically distinct subtypes: Enables identification of novel subtypes with distinct survival profiles across cancer cohorts.

Methodology:

Applies consensus factor analysis to systematically integrate multi-omics data while minimizing statistical assumptions, yielding subtype classifications and patient risk scores.