Detecting drug-artemisinin resistance based on the DNA barcode sequence of plasmodium falciparum using a machine learning algorithm

Abundant data encourages research using a computational methods approach in solving many problems. Bioinformatics is a blended science between biology and informatics techniques that solving problems with molecular biological analysis. Many problems in health, both for diagnosis and treatment of diseases, use data related to viral or bacterial deoxyribonucleic acid (DNA) sequences. It is also used to avoid drug resistance in overcoming variants or mutations of DNA sequences. Several anti-malaria drugs are designed to reduce or eliminate the Plasmodium virus infection. Artemisinin is an anti-malaria drug intended to reduce or eliminate parasitic viral infections. However, these drugs often do not work effectively because the parasites are already resistant. Plasmodium DNA mutations result in a mistake in the desired target. The DNA barcode sequence is the unique aspect of the whole-genome DNA sequence. Therefore, this study aims to detect anti-malaria drug resistance using machine learning methods based on determining the DNA barcode of Plasmodium falciparum. This study provides an exhaustive assessment of a range of machine learning architectures, such as K-Nearest Neighbor (KNN), Naïve Bayes, decision tree, Support Vector Machine (SVM), and Random Forest Algorithms. The Random Forest algorithm exhibits superior performance in terms of accuracy of 0.97, precision of 0.94, and an f1-score of 0.95 consistently, surpassing its competitors. However, the highest recall of 0.98 is achieved by SVM algorithm and is slightly different if compared to Random Forest.