Using knowledge-driven genomic interactions for multi-omics data analysis: metadimensional models for predicting...

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Tumor heterogeneity leads to diverse molecular signatures in cancer patients.
• Existing methods for identifying genomic interactions lack systematic approaches for multi-omics data.
• Prior biological knowledge integration is key to understanding complex genomic interactions.

Purpose of the Study:

• To propose a novel methodological framework, metadimensional knowledge-driven genomic interactions (MKGIs), for identifying pathway interaction models from multi-omics data.
• To apply MKGIs to ovarian cancer data to predict clinical outcomes.
• To evaluate the performance of MKGIs compared to single knowledge-driven models.

Main Methods:

• Development of the MKGI framework integrating prior biological knowledge with multi-omics data.
• Application of MKGIs to The Cancer Genome Atlas ovarian cancer dataset.
• Prediction of grade, stage, and survival outcomes using MKGI models.

Main Results:

• MKGI models identified distinct pathway features from different genomic datasets, indicating varied contributions to ovarian cancer outcomes.
• MKGI models significantly outperformed single knowledge-driven genomic interaction models.
• Identified key pathway interactions, including MAPK and GnRH signaling, crucial for cancer pathogenesis.

Conclusions:

• MKGIs provide a robust framework for analyzing pathway interactions in multi-omics data.
• The integration of biological knowledge enhances the interpretability and predictive power of cancer models.
• Understanding molecular signature variability through pathway interactions can advance precision medicine for ovarian cancer.