Uncertainty-aware dynamic integration for multi-omics classification of tumors

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Multi-omics data integration is vital for understanding complex biological mechanisms and improving medical diagnosis.
• Challenges exist in handling data quality variations and capturing uncertainty across different omics types.

Purpose of the Study:

• To propose an uncertainty-aware dynamic integration framework for multi-omics classification.
• To address the challenge of dynamically capturing data uncertainty in complex omics datasets.

Main Methods:

• Developed a framework with deep embedding and confidence prediction modules.
• Introduced 'confidNet' to assign confidence values for dynamic multi-omics integration.
• Extracted key information into low-dimensional representations for downstream tasks.

Main Results:

• The proposed method retains more crucial biomedical information compared to existing integration techniques.
• Achieved reliable multi-omics integration with high accuracy, even on small sample datasets.
• Demonstrated effectiveness through extensive experimental validation.

Conclusions:

• The framework is applicable to high-dimensional omics data.
• Has significant potential for advancing medical decision-making and biological analysis.