Large-scale 16S gene assembly using metagenomics shotgun sequences.

Abstract

Combining a 16S rRNA (16S) gene database with metagenomic shotgun sequences promises unbiased identification of known and novel microbes. To achieve this, we herein report reference-based ribosome assembly (RAMBL), a computational pipeline, which integrates taxonomic tree search and Dirichlet process clustering to reconstruct full-length 16S gene sequences from metagenomic sequencing data with high accuracy. By benchmarking against the synthetic and real shotgun sequences, we demonstrated that full-length 16S gene assemblies of RAMBL were a good proxy for known and putative microbes, including Candidate Phyla Radiation. We found that 30-40% of bacteria genera in the terrestrial and intestinal biomes have no closely related genome sequences. We also observed that RAMBL was able to generate a more accurate determination of environmental microbial diversity and yield better disease classification, suggesting that full-length 16S gene assemblies are a powerful alternative to marker gene set and 16S short reads. RAMBL first realizes the access to full-length 16S gene sequences in the near-terabase-scale metagenomic shotgun sequences, which markedly improve metagenomic data analysis and interpretation. RAMBL is available at https://github.com/homopolymer/RAMBL for academic use. zengfeng@xmu.edu.cn. Supplementary data are available at Bioinformatics online.