Taxonomy metagenomic analysis for microbial sequences in three domains system via machine learning approaches. (2018)

• Record Type: Journal Article
• Authors: Afify, Heba M.; Al-Masni, Mohammed A.
• Is Part Of: Informatics in medicine unlocked. Volume 13(2018)
• Page Start: 151
• Page End: 157
• Publication Date: 2018
• Subjects: Bioinformatics -- Genome sequences -- Supper vector machine (SVM) -- Deep belief network (DBN) -- Taxonomy; Medical informatics -- Periodicals; 610.285
• DOI: 10.1016/j.imu.2018.05.004
• Languages: English
• ISSNs: 2352-9148

Abstract

Abstract: The rapid advancements of using clinical microbiology and genome sequences encourage several taxonomic approaches, based upon both genome classification and bioinformatics surveys. This taxonomy arranges the tree of life among different organism databases and exploits the high similarly in biological information to access the best representation of genome sequences. However, there is still a challenge to find the entire hierarchy of this tree, due to the existence of the biodiversity databases, and the different classifiers, according to evolutionary or phylogenetic relationships. This paper presents the classification of three domains of microorganisms including Bacteria, Archaea, and Eukarya using two algorithms: the supper vector machine (SVM) and the deep belief network (DBN). The proposed approach utilized the alignment method and the code generation process as preprocessing steps on the EzBioCloud 16S rRNA database. In addition, this study accommodated the issue of choosing the proper reference sequence (RefSeq) and the appropriate code generation process of the genome sequences. Our results showed that the proposed method classifies the genome sequences with an overall classification accuracy of 99.99% and 99.93% for SVM and DBN classifiers using the standard RefSeq of each class, respectively. This paper enhanced the area of microbiological scientific classification through progress in using the character-based arrangement that will help in future evolutionary frameworks.