SMS: A Novel Approach for Bacterial Strain Analysis in Multiple Samples

Abstract

The analysis of the bacterial strains is important for understanding drug resistance. Despite the existence of dozens of computational tools for bacterial strain studies, most of them are for known bacterial strains. Almost all remaining tools are designed to analyze individual samples or local strain regions. With multiple shotgun metagenomic samples routinely generated in a project, it is necessary to create methods to infer novel bacterial strain genomes in multiple samples. To fill this gap, we developed a novel computational approach called SMS to de novo reconstruct bacterial Strain genomes in Multiple Samples. Tested on 702 simulated and 195 experimental datasets, SMS reliably identified the strain number, abundance, and polymorphisms. Compared with two existing approaches, SMS showed superior performance.