CBP-JMF: An Improved Joint Matrix Tri-Factorization Method for Characterizing Complex Biological Processes of...

Area of Science:

• Computational biology and bioinformatics
• Genomics and molecular biology

Background:

• Multi-omics data analysis is crucial for understanding complex biological processes (CBPs).
• Identifying CBPs is essential for classifying disease subtypes and cell types, necessitating advanced computational tools.
• Existing methods lack comprehensive approaches for inferring CBPs linked to sample groupings.

Purpose of the Study:

• To introduce CBP-JMF, a novel Python-based tool for discovering complex biological processes (CBPs).
• To apply CBP-JMF for identifying CBPs that characterize distinct disease subtypes, using breast cancer as a model.
• To provide a practical framework for integrating multi-omics data to reveal disease mechanisms.

Main Methods:

• Development of CBP-JMF, a tool based on a joint non-negative matrix tri-factorization framework.
• Implementation of the tool in Python for practical usability and accessibility.
• Application of CBP-JMF to multi-omics data from four subtypes of breast cancer.

Main Results:

• CBP-JMF successfully identified key CBPs associated with different breast cancer subtypes.
• A significant overlap was observed between genes derived from the identified CBPs and established subtype-specific pathways.
• The tool demonstrated effectiveness in uncovering biologically relevant CBPs that explain disease heterogeneity.

Conclusions:

• CBP-JMF is an effective computational tool for discovering complex biological processes underlying sample groups, particularly disease subtypes.
• The findings highlight the utility of joint non-negative matrix tri-factorization in multi-omics data analysis for biological discovery.
• The application to breast cancer subtypes validates CBP-JMF's potential for advancing precision medicine through pathway and subtype identification.