University Links: Home Page | Site Map
Covenant University Repository

PREDICTION OF GENETIC VARIANTS ASSOCIATED WITH ANTIMALARIAL DRUG RESISTANCE USING SET COVERING MACHINE

APATA, OLUWABUKOLA RACHEAL and Covenant University, Theses (2022) PREDICTION OF GENETIC VARIANTS ASSOCIATED WITH ANTIMALARIAL DRUG RESISTANCE USING SET COVERING MACHINE. Masters thesis, COVENANT UNIVERSITY.

[img] PDF
Download (639kB)

Abstract

Antimalarial resistance (AMR) has become a major issue in malaria-endemic countries, and novel methods for identifying strains resistant or susceptible to specific medications are critical in the fight against antimalarial-resistant Plasmodium parasites. The growing availability of genetic information has enabled the application of computational methods in surveying resistance patterns. K-mer-based machine learning approaches have shown considerable potential as a diagnostic and research tool. In this work, Set Covering Machine (SCM) algorithm was applied to predict antimalarial drug response outcomes and their genetic determinants. The model predicted six antimalarial drugs (Chloroquine, Dihydroartemisinin, Lumafantrine, Primaquine, Pyrimethamine, and Mefloquine) response phenotype in Plasmodium falciparum. The model used the most compact set of k-mers generated from the genomes of the parasite isolates to learn and predict binary drug response outcomes. To avoid model overfitting, ten-fold cross-validation was conducted on the training set to choose the optimal hyperparameter values. Regardless of the resistance mechanism, whether acquired resistance or point mutations in the chromosome, the training accuracy (mean cross-validation score) and testing accuracy of SCM prediction of the six antimalarial drug resistance was above 85%. The model significantly classified the resistant isolates from the sensitive isolates of the parasite and could be used as potential tools in antimalarial resistance surveillance and clinical studies. A number of sequence k-mers associated with antimalarial drug resistance were identified. We identified several already known genes and loci associated with the six drugs, including those containing pfcrt and pfdhfr. Novel genes and loci were also discovered. Of particular interest are the variant regions on the var genes on chromosomes 6, 8, 10, and 13 containing the Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1). The PfEMP1 variant k-mers were found to be associated with chloroquine, dihydroartemisinin, and pyrimethamine resistance. The var genes encode PfEMP1. The genes have extreme variability and are a principal virulence factor of malaria parasite with extreme antigenic variability. The variations in these var genes were found to play a role in antimalarial drug resistance in P. falciparum.

Item Type: Thesis (Masters)
Uncontrolled Keywords: Machine learning, Malaria, Plasmodium falciparum, Genome-Wide Association Study, Phenotype prediction
Subjects: Q Science > Q Science (General)
Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: Faculty of Engineering, Science and Mathematics > School of Electronics and Computer Science
Depositing User: Mrs Hannah Akinwumi
Date Deposited: 16 May 2022 17:42
Last Modified: 16 May 2022 17:42
URI: http://eprints.covenantuniversity.edu.ng/id/eprint/15839

Actions (login required)

View Item View Item