engleski

De novo metagenomic assembly using Bayesian model-based clustering

Microbial communities influence almost every aspect of our lives. Metagenomics aims to expand our knowledge of those communities by analyzing DNA samples extracted directly from their environments. Metagenomic studies still rely mostly on manual interventions and single genome assembly tools which are unaware of the nature of metagenomic data. In this thesis, a Bayesian model-based hierarchical clustering approach to aid in metagenomic assembly called Sigma is presented. Sigma is combined with an optimal single genome scaffolder Opera to show that metagenomic assembly problem can be accurately and automatically reduced to single genome assembly problem by systematically exploiting assembly information. Comparisons on simulated and real datasets show that this pipeline (OperaMS) outperforms state-ofthe- art single genome (Velvet, SOAPdenovo) and metagenomic (MetaVelvet, Bambus2) assembly tools.

metagenomics; de novo assembly; hierarchical clustering; Bayesian information criterion

nije evidentirano