A Novel Human Microbe-Disease Association Prediction Method Based on the Bidirectional Weighted Network

Area of Science:

• Microbiology and Computational Biology
• Human Health and Disease
• Bioinformatics and Network Analysis

Background:

• Microbes play a crucial role in human physiology and disease.
• Understanding microbe-disease associations is vital for medical advancements.
• Existing methods for predicting these associations have limitations.

Purpose of the Study:

• To develop a computational model for predicting potential microbe-disease relationships.
• To construct a novel bidirectional weighted network integrating known microbe-disease associations.
• To evaluate the predictive performance of the developed model, BWNMHMDA.

Main Methods:

• Construction of a bidirectional weighted network using normalized Gaussian interactions and bidirectional recommendations.
• Development of the BWNMHMDA computational model based on the constructed network.
• Validation using Leave-One-Out Cross-Validation (LOOCV) and 5-fold cross-validation.
• Performance evaluation through case studies on asthma, colorectal carcinoma, and chronic obstructive pulmonary disease.

Main Results:

• BWNMHMDA achieved reliable Area Under the Curve (AUC) values of 0.9127 (LOOCV) and 0.8967 ± 0.0027 (5-fold cross-validation).
• The model outperformed several state-of-the-art methods in prediction accuracy.
• Case studies confirmed 10, 9, and 8 of the top 10 predicted microbes for asthma, colorectal carcinoma, and COPD, respectively, in related literature.

Conclusions:

• BWNMHMDA is a robust and accurate computational model for predicting microbe-disease associations.
• The model's performance is superior to existing methods.
• BWNMHMDA shows significant potential for identifying novel microbe-disease links relevant to human health.