Biomarker discovery by integrated joint non-negative matrix factorization and pathway signature analyses

Area of Science:

• Genomics
• Pharmacology
• Bioinformatics

Background:

• Predictive biomarkers are crucial for personalized medicine, guiding patient selection for targeted therapies.
• Mining biologically meaningful biomarkers from complex multi-omics data remains a significant challenge.
• Existing methods struggle to fully leverage integrated genomic, transcriptomic, and pharmacological datasets.

Purpose of the Study:

• To develop and validate an approach for identifying novel predictive biomarkers from multi-omics cell line data.
• To integrate joint non-negative matrix factorization (JNMF) with pathway signature analyses for biomarker discovery.
• To enhance the prediction of drug efficacy by identifying more precise biomarker signatures.

Main Methods:

• Utilized joint non-negative matrix factorization (JNMF) to analyze integrated multi-omics data from cell lines.
• Applied pathway signature analyses for multi-layer interpretation of JNMF-derived clusters.
• Validated identified biomarker-drug associations using known examples (e.g., BRAF mutation/PLX4720, HER2 amplification/lapatinib).

Main Results:

• JNMF successfully identified known biomarker-drug associations, confirming the method's validity.
• Discovered that combined BRAF mutation and MITF activation predict enhanced sensitivity to BRAF inhibitors.
• Demonstrated that the BRAF/MITF axis is a more potent biomarker for BRAF inhibitor efficacy than BRAF mutation alone.

Conclusions:

• The integrated JNMF and pathway signature analysis approach effectively mines multi-omics data for predictive biomarkers.
• This method offers a more refined prediction of drug response, exemplified by the BRAF/MITF axis.
• The approach holds promise for drug development, identifying pharmacodynamic biomarkers, and analyzing drug response in clinical settings.