Predicting drug-target interactions using Lasso with random forest based on evolutionary information and chemical...

Area of Science:

• Computational Biology
• Pharmacology
• Bioinformatics

Background:

• Traditional experimental methods for identifying drug-target interactions are expensive and time-consuming.
• Machine learning approaches are increasingly utilized for efficient prediction of these interactions.
• Accurate drug-target interaction prediction is crucial for pharmaceutical research and drug development.

Purpose of the Study:

• To develop and validate a novel machine learning method, LRF-DTIs, for predicting drug-target interactions.
• To enhance the accuracy and efficiency of drug-target interaction prediction compared to existing methods.
• To provide a valuable tool for new drug research and target protein development.

Main Methods:

• Feature extraction using pseudo-position specific scoring matrix (PsePSSM) and FP2 molecular fingerprinting.
• Dimensionality reduction with Lasso and handling imbalanced data using Synthetic Minority Oversampling Technique (SMOTE).
• Prediction of drug-target interactions using a random forest (RF) classifier.

Main Results:

• Achieved high prediction accuracies: 98.09% (enzyme), 97.32% (ion channel), 95.69% (GPCR), and 94.88% (nuclear receptor).
• Demonstrated effectiveness in predicting new interactions and performing well on new datasets.
• Outperformed other existing prediction methods in cross-validation trials.

Conclusions:

• The LRF-DTIs method significantly improves the accuracy of drug-target interaction prediction.
• This approach plays a vital role in accelerating new drug discovery and target protein development.
• The source code and datasets are publicly available for academic use.