Abstract
Coronaviruses, including severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2), continue to pose a significant public health challenge globally, even in 2024. Despite advancements in vaccines and treatments, the accurate classification of coronavirus protein sequences remains crucial for monitoring variants, understanding viral behavior, and developing targeted interventions. In this study, we investigate the efficacy of various classification methods in accurately classifying coronavirus protein sequences. We explore the use of K‐nearest neighbor (KNN), fuzzy KNN (FKNN), support vector machine (SVM), and SVM with particle swarm optimization (PSO‐SVM) algorithms for classification, complemented by feature selection techniques including principal component analysis (PCA) and random forest‐recursive feature elimination (RF‐RFE). Our dataset comprises 2000 protein sequences, evenly split between SARS‐CoV‐2 and non‐SARS‐CoV‐2 sequences. Through rigorous analysis, we evaluate the performance of each classification model in terms of accuracy, sensitivity, specificity, and receiver operating characteristic area under the curve (ROC‐AUC). Our findings demonstrate consistently high performance across all models, reflecting their efficacy in classifying coronavirus protein sequences. Notably, the PCA + PSO‐SVM model emerges as the top‐performing model, exhibiting the highest classification accuracy, specificity, and ROC‐AUC score, demonstrating its effectiveness in distinguishing between SARS‐CoV‐2 and non‐SARS‐CoV‐2 sequences. Overall, our study highlights the importance of employing advanced classification methods and feature selection techniques in accurately classifying coronavirus protein sequences. The findings provide valuable insights for researchers and practitioners in the field of bioinformatics and contribute to ongoing efforts in understanding and combating the COVID‐19 pandemic and its evolving challenges.