%0 Journal Article %A Dongwan D. Kang %A Jeff Froula %A Rob Egan %A Zhong Wang %T A robust statistical framework for reconstructing genomes from metagenomic data %D 2014 %R 10.1101/011460 %J bioRxiv %P 011460 %X We present software that reconstructs genomes from shotgun metagenomic sequences using a reference-independent approach. This method permits the identification of OTUs in large complex communities where many species are unknown. Binning reduces the complexity of a metagenomic dataset enabling many downstream analyses previously unavailable. In this study we developed MetaBAT, a robust statistical framework that integrates probabilistic distances of genome abundance with sequence composition for automatic binning. Applying MetaBAT to a human gut microbiome dataset identified 173 highly specific genomes bins including many representing previously unidentified species. %U https://www.biorxiv.org/content/biorxiv/early/2014/11/15/011460.full.pdf