A hierarchical integration deep flexible neural forest framework for cancer subtype classification by integrating...

Area of Science:

• Computational biology
• Genomics
• Cancer research

Background:

• Accurate cancer subtype classification is crucial for personalized treatment.
• High-throughput sequencing generates multi-omics data (gene expression, miRNA, methylation).
• Existing methods often rely solely on gene expression data, limiting classification power.

Purpose of the Study:

• To develop a novel framework for integrating multi-omics data for cancer subtype classification.
• To improve the accuracy and effectiveness of cancer subtype identification compared to single-data approaches.

Main Methods:

• Proposed a hierarchical integration deep flexible neural forest (HI-DFNForest) framework.
• Utilized stacked autoencoders (SAE) for representation learning within each omics data type.
• Integrated learned representations for comprehensive data analysis and classification using deep flexible neural forest (DFNForest).

Main Results:

• HI-DFNForest effectively integrates gene expression, miRNA expression, and DNA methylation data.
• Multi-omics data integration significantly improves cancer subtype classification accuracy compared to using only gene expression data.
• The proposed framework demonstrated superior performance against conventional methods on BRCA, GBM, and OV datasets from TCGA.

Conclusions:

• The HI-DFNForest framework provides an effective method for multi-omics data integration.
• This approach enhances cancer subtype classification, aiding in diagnosis and personalized treatment strategies.