Application of Machine Learning Techniques in Drug-target Interactions Prediction

Area of Science:

• Computational chemistry
• Bioinformatics
• Machine learning

Background:

• Drug-target interactions (DTIs) are crucial for drug design and repositioning.
• Traditional experimental methods for identifying DTIs are costly and time-consuming.
• Computational methods, particularly machine learning, offer efficient alternatives.

Purpose of the Study:

• To review and categorize machine learning methods for predicting DTIs.
• To compare the advantages and limitations of different machine learning approaches.
• To identify key databases used in drug discovery.

Main Methods:

• Categorization of machine learning methods into supervised, semi-supervised, and unsupervised.
• Review of representative recent methods within each category.
• Comparison of the strengths and weaknesses of various prediction models.

Main Results:

• Machine learning methods show significant efficiency in predicting DTIs.
• Databases frequently used in drug discovery were identified.
• Comparative analysis of different machine learning categories was performed.

Conclusions:

• No single prediction model is universally superior; selection depends on application.
• Key challenges in DTI prediction include data scarcity, prediction biases, and appropriate use of regression models.