CLASSIFICATION FRAGMEN METAGENOM MENGGUNAKAN PRINCIPAL COMPONENT ANALYSIS NEIGHBOR

Abstract

Metagenomics is a study of metagenom analysis which its genetic materials is obtained directly from environmental samples. The process of metagenome sequencing produce fragments from mixture organisms. Thus, assembling fragments directly will generate chimeric contigs. Furthermore, a bining process is required to classify these fragments into a particular taxonomic level. In this study, the classification of metagenome fragment were extracted using n-mers, reduced its dimension using principal component analysis and classified using knearest neighbor. The experiments were conducted from in the various fragment length from 0.5 Kbp to 10 Kbp. The best results were obtained using KNN with k=7 and implementing 4-mers frequency. The accuracies of classifying known organisms obtained using PCA 95% were ranged from 91.6% to 99.9%. Moreover, the accuracies were slightly decreased when classifying unknown organisms, from 89.64% to 99.32%.