DrugClust: A machine learning approach for drugs side effects prediction

Area of Science:

• Computational biology
• Pharmacology
• Machine learning

Background:

• Understanding drug side effects is critical in drug discovery.
• Machine learning (ML) methods are essential for predicting and understanding complex drug-induced adverse events.
• Identifying mechanisms of drug side effects is a key challenge in pharmaceutical research.

Purpose of the Study:

• To introduce DrugClust, a novel ML algorithm for predicting drug side effects.
• To provide a tool for clustering drugs based on their features for improved side effect prediction.
• To enable biological validation of drug clusters and explore new drug-pathway interactions.

Main Methods:

• DrugClust employs a pipeline involving drug clustering based on features.
• Bayesian scores are utilized for predicting drug side effects.
• Enrichment analysis is integrated for biological validation of clusters and pathway interaction studies.

Main Results:

• The DrugClust algorithm was evaluated using 5-fold cross-validation.
• Performance was compared against established datasets (Zhang et al., 2015; Liu et al., 2012; Mizutani et al., 2012).
• DrugClust demonstrated superior performance in most comparative analyses.

Conclusions:

• DrugClust offers a promising approach for predicting drug side effects.
• The algorithm facilitates the understanding of mechanisms underlying drug toxicity.
• The R package is publicly available for use in drug discovery and research.