MDAKRLS: Predicting human microbe-disease association based on Kronecker regularized least squares and similarities

Area of Science:

• Microbiology
• Computational Biology
• Bioinformatics

Background:

• Microbes are integral to human health and disease.
• Identifying microbe-disease associations is crucial for understanding disease mechanisms and developing treatments.
• Current knowledge of disease-related microbes is limited, necessitating efficient computational approaches.

Purpose of the Study:

• To develop a novel computational method, MDAKRLS, for discovering potential microbe-disease associations (MDAs).
• To enhance the accuracy and efficiency of identifying microbes linked to human diseases.

Main Methods:

• Developed MDAKRLS using Kronecker regularized least squares.
• Incorporated Hamming interaction profile similarity and Gaussian interaction profile kernel similarity.
• Utilized Kronecker products to construct microbe-disease pair similarities for improved prediction.

Main Results:

• MDAKRLS achieved high AUC values (0.9327 and 0.9023 ± 0.0015) in cross-validation, outperforming state-of-the-art methods.
• Demonstrated faster computing speed compared to existing methods.
• Case studies on IBD and asthma validated a high percentage of top predicted microbes through literature.

Conclusions:

• MDAKRLS exhibits effective and reliable prediction performance for microbe-disease associations.
• The method can serve as a valuable tool for discovering novel disease-related microbes.
• Facilitates biomedical research by aiding in the identification of microbes for further investigation.