Machine Learning for Prediction of Drug Targets in Microbe Associated Cardiovascular Diseases by Incorporating...

Area of Science:

• Bioinformatics
• Computational Biology
• Drug Discovery

Background:

• Host-pathogen interactions are fundamental to human disease, influencing invasion, infection, and immune responses.
• Identifying specific drug targets within these interactions is critical for developing effective therapies.

Purpose of the Study:

• To develop and evaluate machine learning algorithms for classifying host and pathogen proteins as drug targets.
• To identify key protein features that are predictive of drug target status.
• To predict potential drug targets for Microbe Associated Cardiovascular Diseases.

Main Methods:

• Four machine learning algorithms (Logistic Regression, K-Nearest Neighbor, Support Vector Machine, Random Forest) were employed.
• Models were trained on datasets of host and pathogen proteins with computed sequence, structure, biological, and network features.
• 10-fold cross-validation was used to assess model performance, with Random Forest showing superior accuracy.

Main Results:

• The Random Forest classifier achieved 99% accuracy and a ROC-AUC score of 0.99±0.01 for both host and pathogen datasets.
• Eigenvector Centrality of host-pathogen and host-host interactions emerged as the most significant feature for classifying pathogen and host targets, respectively.
• Other key features included catalytic/binding sites, instability index, and cellular location.

Conclusions:

• Machine learning, specifically the Random Forest algorithm, is highly effective for classifying drug targets based on protein characteristics.
• The study successfully predicted 331 host and 743 pathogen proteins as potential drug targets for Microbe Associated Cardiovascular Diseases.
• These predicted targets warrant experimental validation for therapeutic intervention in cardiovascular diseases linked to microbial infections.