Comparing Machine Learning Models for Peptide Toxicity Prediction

Peptides are widely studied as potential therapeutic agents because they are specific, effective, and relatively inexpensive to produce. They are used in drug development, vaccines, and antimicrobial treatments. However, peptide toxicity remains a major concern, since toxic peptides can cause harmful side effects and limit their medical use. Experimental testing of peptide toxicity is often expensive and time consuming, making computational approaches more practical. Machine learning provides an efficient way to predict peptide toxicity by learning patterns from labeled data. In this study, two machine learning models, Support Vector Machine and Random Forest, were compared for their ability to predict peptide toxicity. A dataset containing over ten thousand peptide sequences labeled as toxic or non toxic was used. Peptide sequences were converted into numerical features using sequence length and counts of positively charged amino acids. The models were evaluated using 10 fold cross validation in the Orange data mining software. Results showed that Random Forest achieved higher accuracy and AUC compared to Support Vector Machine. These findings suggest that Random Forest is better suited for peptide toxicity prediction when using simple sequence based features.