Metagenomics Analysis of Oral Microbiota associated with Periodontal Disease by Machine Learning Methods

10.29428/9789860544169.201801.0145

Wei-Ren Lin；Yaw-Ling Lin；Ming-Li Liou；Shih-Hao Chang

metagenomics ； microbiome community ； machine learning ； support vector machine ； convolutional neural networks

NCS 2017 全國計算機會議

2017（2018 / 01 / 01）

772 - 777

英文

Analysis of shotgun metagenomic data provides opportunity to understand the metabolic potentials, ecological roles, and evolutionary history of rare and uncultured microbes in natural communities by analyzing environmental DNA directly without prior cultivation. In this paper, we propose methods and implement tools to facilitate the bioinformatics analysis of metagenomic data. Several bioinformatics sequences analysis open-source softwares are integrated to construct accessible platforms for metagenomics analysis. By applying machine-learning- based approaches, we analyze the oral microbiota compositions to further understand the effects of periodontal treatments on patients with chronic periodontitis. These machine-learning approaches include the (unsupervised) Dirichlet multinomial mixture (DMM) regression models, t- SNE visualization of high-dimensional metagenomic data, as well as the supervised convolutional neural networks and support vector machines models.