Integrative Analysis for Identifying Co-Modules of Microbe-Disease Data by Matrix Tri-Factorization With Phylogenetic...

Area of Science:

• Microbiology
• Bioinformatics
• Computational Biology

Background:

• Microbe-disease association mining is crucial for understanding disease pathogenesis.
• Existing methods often focus on single microbe-disease pairs, neglecting complex microbial interactions.
• A systems-level approach is needed to explore microbe-disease co-modules.

Purpose of the Study:

• To develop a novel algorithm, MDNMF, for identifying microbe-disease association modules.
• To reveal connections between different levels of microbial and disease modules.
• To enhance the understanding of complex microbe-related diseases.

Main Methods:

• Proposed a two-level module identifying algorithm (MDNMF) based on nonnegative matrix tri-factorization.
• Integrated microbe-disease association matrix, disease similarity, and microbe similarity matrices.
• Incorporated human symptoms-disease networks and microbial phylogenetic distance to improve model performance.

Main Results:

• MDNMF demonstrated superior performance compared to existing NMF-based methods on the HMDAD dataset.
• Achieved better results in terms of enrichment index (EI) and significantly enriched taxon sets.
• Successfully identified microbial modules with significant biological functions.

Conclusions:

• MDNMF offers a powerful tool for uncovering complex microbe-disease relationships.
• The algorithm's ability to identify microbial modules has significant implications for disease research.
• This approach advances the systems-level study of microbe-related diseases.