DeepProg

DeepProg applies an ensemble of deep-learning and machine-learning methods to integrate multi-omics data for cancer survival prediction and patient risk stratification.

Key Features:

• Ensemble Framework: Combines deep-learning and machine-learning algorithms into an ensemble predictive framework.
• Survival Subtype Prediction: Identifies optimal survival subtypes across various cancer types.
• Risk Stratification: Stratifies patients into distinct prognostic risk categories to inform treatment decisions.
• High Predictive Accuracy: Achieves concordance index (C-index) values of 0.68–0.80 on liver and breast cancer datasets.
• Pan-Cancer Analysis: Associates genomic signatures with biological processes such as extracellular matrix modeling, immune deregulation, and mitosis in poor-survival subtypes across cancers.
• Multi-Omics Integration: Integrates genomics, transcriptomics, and proteomics data for combined analysis.
• Computational Integration: Addresses computational challenges associated with integrating diverse omics datasets.

Scientific Applications:

• Prognostic Prediction: Predicts patient survival subtypes to support analysis of disease progression and prognosis.
• Multi-Omics Integration: Enables combined analysis of multiple omics layers (genomics, transcriptomics, proteomics) to derive biological insights.
• Research Insights: Links genomic signatures to cellular processes, supporting cancer research and potential therapeutic target identification.

Methodology:

Employs an ensemble approach that integrates deep-learning models and machine-learning algorithms to manage complex multi-omics datasets for survival prediction and risk stratification.