Microbial risk score for capturing microbial characteristics, integrating multi-omics data, and predicting disease...

Area of Science:

• Microbiome Research
• Computational Biology
• Genomics

Background:

• Microbiome-wide association studies generate vast data for disease research.
• Unique microbiome data features pose challenges for disease prediction.
• Developing robust methods is crucial for leveraging microbiome data in healthcare.

Purpose of the Study:

• To propose a microbial risk score (MRS) framework for disease prediction.
• To aggregate complex microbial profiles into a summarized risk score.
• To enable the integration of microbiome data with other omics for enhanced disease prediction.

Main Methods:

• Developed a two-step MRS algorithm: identifying signature microbial taxa and integrating them into a continuous score.
• Utilized microbial association tests and pruning/thresholding for taxa identification.
• Constructed a community-based MRS using alpha diversity and proposed multi-omics integration methods.

Main Results:

• Evaluated the MRS framework using COVID-19, GMHI, and type 1 diabetes cohorts.
• Developed and validated disease-specific MRSs for colorectal adenoma, colorectal cancer, Crohn's disease, and rheumatoid arthritis.
• Achieved high AUCs for Crohn's disease MRS, demonstrating its predictive power.

Conclusions:

• The MRS framework highlights the utility of microbiome data for disease prediction and multi-omics integration.
• Provides a valuable tool for understanding the microbiome's role in disease diagnosis and prognosis.
• Offers potential for advancing personalized medicine through microbiome-based risk assessment.