MPEMDA: A multi-similarity integration approach with pre-completion and error correction for predicting microbe-drug...

Area of Science:

• Microbiology
• Pharmacology
• Computational Biology
• Bioinformatics

Background:

• Predicting microbe-drug associations is crucial for understanding mechanisms.
• Wet lab experiments are time-consuming; computational methods offer an alternative.
• Existing models neglect individual similarities, impacting predictive accuracy.

Purpose of the Study:

• To develop a novel computational method, MPEMDA, for predicting microbe-drug associations.
• To overcome limitations of existing models by incorporating both integrated and individual similarities.

Main Methods:

• MPEMDA pre-completes the microbe-drug association matrix using various similarity combinations.
• A label propagation algorithm with error correction is employed for prediction.
• Similarity Network Fusion (SNF) is used to obtain integrated and individual similarities.

Main Results:

• MPEMDA outperforms state-of-the-art methods in 5-fold cross-validation and de novo tests.
• Experimental results on three benchmark datasets demonstrate superior performance.
• Case studies show potential for identifying novel microbe-drug associations.

Conclusions:

• MPEMDA offers a more accurate approach to predicting microbe-drug associations.
• The method effectively utilizes diverse similarity information for enhanced predictions.
• MPEMDA has strong potential for discovering new microbe-drug relationships.