An integrated metagenomics pipeline for strain profiling reveals novel patterns of bacterial transmission and biogeography.
Nayfach S, Rodriguez-Mueller B, Garud N, Pollard KS. · 2016
Read An integrated metagenomics pipeline for strain profiling reveals novel patterns of bacterial transmission and biogeography. by Nayfach S, Rodriguez-Mueller B, Garud N, Pollard KS. (2016) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc5088602.txt.
Introduction
Stephen Nayfach, Beltran Rodriguez-Mueller, Nandita Garud, and Katherine S. Pollard present MIDAS in Genome Research in 2016, a pipeline that reads strain-level detail from metagenomes. Species-level counts hide real biology, since strains of one species differ in genes and single-letter DNA changes. MIDAS measures species abundance plus gene content and those small variants, using a database of more than 30,000 reference genomes grouped into species.
Applied to stool samples from 98 Swedish mothers and their infants over a year, the pipeline tracked who gave strains to whom. Species mixes of mother and baby converged with time, but strains diverged: early colonizers often came from the mother, while later ones arrived from elsewhere and carried spore-forming genes. In 198 ocean samples worldwide, gene content showed population structure tied to geography. The lesson is that the finest scale of microbial variation carries patterns, of transmission and place, that coarser views erase.
Catalog record
| CREATOR | Nayfach S, Rodriguez-Mueller B, Garud N, Pollard KS. |
|---|---|
| YEAR | 2016 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 82 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
An integrated metagenomics pipeline for strain profiling reveals novel patterns of bacterial transmission and biogeography By Stephen Nayfach, Beltran Rodriguez-Mueller, Nandita Garud, Katherine S. Pollard Published in: Genome Research DOI: 10.1101/gr.201863.115 ABSTRACT We present the Metagenomic Intra-species Diversity Analysis System (MIDAS), which is an integrated computational pipeline for quantifying bacterial species abundance and strain-level genomic variation, including gene content and single-nucleotide polymorphisms (SNPs), from shotgun metagenomes. Our method leverages a database of more than 30,000 bacterial reference genomes that we clustered into species groups. These cover the majority of abundant species in the human microbiome but only a small proportion of microbes in other environments, including soil and seawater. We applied MIDAS to stool metagenomes from 98 Swedish mothers and their infants over one year and used rare SNPs to track strains between hosts. Using this approach, we found that although species compositions of mothers and infants converged over time, strain-level similarity diverged. Specifically, early colonizing bacteria were often transmitted from an infant’s mother, while late colonizing bacteria were often transmitted from other sources in the environment and were enriched for spore-formation genes. We also applied MIDAS to 198 globally distributed marine metagenomes and used gene content to show that many prevalent bacterial species have population structure that correlates with geographic location. Strain-level genetic variants present in metagenomes clearly reveal extensive structure and dynamics that are obscured when data are analyzed at a coarser taxonomic resolution. Microbial species play important roles in the different environments that they inhabit. However, different strains of the same species can differ significantly in their gene content (Greenblum et al. 2015; Zhu et al. 2015) and single-nucleotide polymorphisms (SNPs) (Schloissnig et al. 2013; Kashtan et al. 2014; Lieberman et al. 2014). These strain-level differences are important for understanding microbial evolution, adaptation, pathogenicity, and transmission. For example, strain-level differences have shed light on ecological differentiation of closely related bacteria (Shapiro et al. 2012), uncovered the presence of ancient subpopulations of marine bacteria (Kashtan et al. 2014), and highlighted extensive intra-species recombination (Snitkin et al. 2011; Rosen et al. 2015). Strain-level variation is also important for understanding microbial pathogenicity. Differences at the nucleotide level can lead to within-host adaptation of pathogens (Lieberman et al. 2014), and differences in gene content can confer drug resistance, convert a commensal bacterium into a pathogen (Snitkin et al. 2011), or lead to outbreaks of highly virulent strains (Rasko et al. 2011). Metagenomic shotgun sequencing has the potential to shed light onto strain-level heterogeneity among bacterial genomes within and between microbial communities, yielding a genomic resolution not achievable by sequencing the 16S ribosomal RNA gene alone (Sunagawa et al. 2013) and circumventing the need for culture-based approaches. However, limitations of existing computational methods and reference databases have prevented most researchers from obtaining this level of resolution from metagenomic data. Assembly-free methods that map reads to reference genomes to estimate the relative abundance of known strains (Francis et al. 2013; Tu et al. 2014) are effective for well-characterized pathogens like E. coli that have thousands of sequenced genomes. However, such methods cannot detect strain-level variation for the vast majority of known species that currently have only a single sequenced representative. Other assembly-free approaches have been developed that use reads mapped to one or more reference genomes to identify SNPs (Schloissnig et al. 2013; Lieberman et al. 2014) and gene copy number variants (Greenblum et al. 2015; Zhu et al. 2015; Scholz et al. 2016) of microbial populations. These approaches have not been integrated together and/or made available as software. Recently, several software tools have been developed (Luo et al. 2015; Sahl et al. 2015) that use SNP patterns to phylogenetically type strains, but these methods do not capture the gene content of these organisms and may not be able to resolve strains in communities with high population heterogeneity. Additionally, existing methods do not provide comprehensive up-to-date genomic databases of bacterial species, thus limiting their utility across different environments. Assembly-based methods (Nielsen et al. 2014; Cleary et al. 2015) that seek to reconstruct microbial genomes without using reference genomes are a powerful alternative to assembly-free methods. However, these often require many samples, struggle to deconvolve closely related strains, or require manual inspection. To address these issues, we developed the Metagenomic Intra-species Diversity Analysis System (MIDAS), which is a computational pipeline that quantifies bacterial species abundance and intra-species genomic variation from shotgun metagenomes. Our method integrates many features and leverages a comprehensive database of more than 30,000 reference genomes (for a comparison to existing methods, see Supplemental Table S1). Given a shotgun metagenome, MIDAS rapidly and automatically quantifies gene content and identifies SNPs in bacterial species, which is accurate for populations with a minimum of 1 and 10× sequencing coverage, respectively. These statistics enable quantitative analysis of bacterial populations within and between metagenomic samples. To demonstrate the utility of this approach, we used MIDAS to conduct novel strain-level analyses on two data sets. First, we applied MIDAS to stool metagenomes from 98 Swedish mothers and their infants and used rare SNPs to track vertical transmission and temporal stability of strains in infants over the first year of life. Second, we used MIDAS to quantify gene content of prevalent bacterial species in 198 globally distributed marine metagenomes and identified significant intra-species population structure associated with geographic location and environmental variables. These analyses reveal striking microbial dynamics and structure that are missed when metagenomes are analyzed at a coarser taxonomic resolution. RESULTS Identification of bacterial species with a consistent definition and efficient algorithm To quantify strain-level genomic variation broadly and accurately, we built a comprehensive database of 31,007 high-quality bacterial reference genomes obtained from the Pathosystems Resource Integration Center (PATRIC) (Wattam et al. 2014). We accurately clustered these genomes into species groups to avoid inconsistent, erroneous, and incomplete annotations that afflict some microbial taxonomies (Mende et al. 2013) and to expand and improve upon previous efforts to systematically delineate bacterial species (Mende et al. 2013; Schloissnig et al. 2013; Varghese et al. 2015). Toward this goal, we hierarchically clustered reference genomes using the average pairwise percent identity across a panel of 30 universal genes (Fig. 1A) that we selected from a panel of 112 candidates (Supplemental Fig. S1; Supplemental Table S2; Wu et al. 2013). We found that the best gene families for identifying bacterial species were less conserved and more widely distributed across the tree of life relative to other genes we tested (Supplemental Fig. S2). For example, many ribosomal gene families were too conserved to differentiate closely related species (Supplemental Table S2). We applied a 96.5% nucleotide identity cutoff, which produced genome clusters that were highly concordant with a gold standard definition of prokaryotic species based on 95% genome-wide average nucleotide identity (Supplemental Table S3; Konstantinidis et al. 2006; Richter and Rossello-Mora 2009). Our procedure clustered the 31,007 bacterial genomes into 5952 genome clusters, representing distinct bacterial species (Supplemental Tables S4, S5). We inferred the phylogenetic relationships of these species using a concatenated alignment of the 30 marker genes (Supplemental Fig. S3). Because our algorithm uses a small set of highly informative marker genes, rather than genome-wide sequence comparisons, it will be efficient to update these genome clusters as additional genomes are sequenced. [Figure: Construction