Generation of multimillion-sequence 16S rRNA gene libraries from complex microbial communities by assembling paired-end illumina reads

Microbial communities host unparalleled taxonomic diversity. Adequate characterization of environmental and host-associated samples remains a challenge for microbiologists, despite the advent of 16S rRNA gene sequencing. In order to increase the depth of sampling for diverse bacterial communities, we developed a method for sequencing and assembling millions of paired-end reads from the 16S rRNA gene (spanning the V3 region; ～200 nucleotides) by using an Illumina genome analyzer. To confirm reproducibility and to identify a suitable computational pipeline for data analysis, sequence libraries were prepared in duplicate for both a defined mixture of DNAs from known cultured bacterial isolates (>1 million postassembly sequences) and an Arctic tundra soil sample (>6 million postassembly sequences). The Illumina 16S rRNA gene libraries represent a substantial increase in number of sequences over all extant next-generation sequencing approaches (e.g., 454 pyrosequencing), while the assembly of paired-end 125-base reads offers a methodological advantage by incorporating an initial quality control step for each 16S rRNA gene sequence. This method incorporates indexed primers to enable the characterization of multiple microbial communities in a single flow cell lane, may be modified readily to target other variable regions or genes, and demonstrates unprecedented and economical access to DNAs from organisms that exist at low relative abundances.     

Phylogenetic stratigraphy in the Guerrero Negro hypersaline microbial mat

The microbial mats of Guerrero Negro (GN), Baja California Sur, Mexico historically were considered a simple environment, dominated by cyanobacteria and sulfate-reducing bacteria. Culture-independent rRNA community profiling instead revealed these microbial mats as among the most phylogenetically diverse environments known. A preliminary molecular survey of the GN mat based on only ∼1500 small subunit rRNA gene sequences discovered several new phylum-level groups in the bacterial phylogenetic domain and many previously undetected lower-level taxa. We determined an additional ∼119 000 nearly full-length sequences and 28 000 >200 nucleotide 454 reads from a 10-layer depth profile of the GN mat. With this unprecedented coverage of long sequences from one environment, we confirm the mat is phylogenetically stratified, presumably corresponding to light and geochemical gradients throughout the depth of the mat. Previous shotgun metagenomic data from the same depth profile show the same stratified pattern and suggest that metagenome properties may be predictable from rRNA gene sequences. We verify previously identified novel lineages and identify new phylogenetic diversity at lower taxonomic levels, for example, thousands of operational taxonomic units at the family-genus levels differ considerably from known sequences. The new sequences populate parts of the bacterial phylogenetic tree that previously were poorly described, but indicate that any comprehensive survey of GN diversity has only begun. Finally, we show that taxonomic conclusions are generally congruent between Sanger and 454 sequencing technologies, with the taxonomic resolution achieved dependent on the abundance of reference sequences in the relevant region of the rRNA tree of life.     

Low sequencing efforts bias analyses of shared taxa in microbial communities

The potential for comparing microbial community population structures has been greatly enhanced by developments in next generation sequencing methods that can generate hundreds of thousands to millions of reads in a single run. Conversely, many microbial community comparisons have been published with no more than 1,000 sequences per sample. These studies have presented data on levels of shared operational taxonomic units (OTUs) between communities. Due to lack of coverage, that approach might compromise the conclusions about microbial diversity and the degree of difference between environments. In this study, we present data from recent studies that highlight this problem. Also, we analyzed datasets of 16 rRNA sequences with small and high sequence coverage from different environments to demonstrate that the level of sequencing effort used for analyzing microbial communities biases the results. We randomly sampled pyrosequencing-generated 16S rRNA gene libraries with increasing sequence effort. Sequences were used to calculate Good's coverage, the percentage of shared OTUs, and phylogenetic distance measures. Our data showed that simple counts of presence/absence of taxonomic unities do not reflect the real similarity in membership and structure of the bacterial communities and that community comparisons based on phylogenetic tests provide a way to test statistically significant differences between two or more environments without need an exhaustive sampling effort.     

Comparison of two next-generation sequencing technologies for resolving highly complex microbiota composition using tandem variable 16S rRNA gene regions

High-throughput molecular technologies can profile microbial communities at high resolution even in complex environments like the intestinal microbiota. Recent improvements in next-generation sequencing technologies allow for even finer resolution. We compared phylogenetic profiling of both longer (454 Titanium) sequence reads with shorter, but more numerous, paired-end reads (Illumina). For both approaches, we targeted six tandem combinations of 16S rRNA gene variable regions, in microbial DNA extracted from a human faecal sample, in order to investigate their limitations and potentials. In silico evaluations predicted that the V3/V4 and V4/V5 regions would provide the highest classification accuracies for both technologies. However, experimental sequencing of the V3/V4 region revealed significant amplification bias compared to the other regions, emphasising the necessity for experimental validation of primer pairs. The latest developments of 454 and Illumina technologies offered higher resolution compared to their previous versions, and showed relative consistency with each other. However, the majority of the Illumina reads could not be classified down to genus level due to their shorter length and higher error rates beyond 60 nt. Nonetheless, with improved quality and longer reads, the far greater coverage of Illumina promises unparalleled insights into highly diverse and complex environments such as the human gut.     

Effects of Experimental Choices and Analysis Noise on Surveys of the “Rare Biosphere”

When planning a survey of 16S rRNA genes from a complex environment, investigators face many choices including which primers to use and how to taxonomically classify sequences. In this study, we explored how these choices affected a survey of microbial diversity in a sample taken from the aerobic basin of the activated sludge of a North Carolina wastewater treatment plant. We performed pyrosequencing reactions on PCR products generated from primers targeting the V1-V2, V6, and V6-V7 variable regions of the 16S rRNA gene. We compared these sequences to 16S rRNA gene sequences found in a whole-genome shotgun pyrosequencing run performed on the same sample. We found that sequences generated from primers targeting the V1-V2 variable region had the best match to the whole-genome shotgun reaction across a range of taxonomic classifications from phylum to family. Pronounced differences between primer sets, however, occurred in the “rare biosphere” involving taxa that we observed in fewer than 11 sequences. We also examined the results of analysis strategies comparing a classification scheme using a nearest-neighbor approach to directly classifying sequences with a naïve Bayesian algorithm. Again, we observed pronounced differences between these analysis schemes in infrequently observed taxa. We conclude that if a study is meant to probe the rare biosphere, both the experimental conditions and analysis choices will have a profound impact on the observed results.For nearly 3 decades, investigations of the distribution of microbes in complex environments have focused on the use of rRNA genes (1, 2, 4, 11, 16, 18, 19, 22, 24). Because the full-length 16S rRNA sequence can be obtained with paired-end reads via traditional Sanger sequencing, until recently most studies of the 16S rRNA gene captured most or nearly most of the 16S sequence length. New pyrosequencing technologies, however, have recently been introduced that greatly reduce the per base cost of sequencing but with shorter read lengths than traditional Sanger sequencing (17). This new approach has proven powerful, yielding a previously unobtainable view of rare taxa (7, 12-14, 25).The shorter reads produced by pyrosequencing require the choice of a particular region of the 16S rRNA gene to target for pyrosequencing as well as the choice of an algorithm to classify the taxonomy of the shorter reads. In their initial surveys of microbial diversity with pyrosequencing (12, 14, 25), Sogin and colleagues targeted the V6 variable region, in part because it is was small enough to be captured with the 100-bp reads of the pyrosequencing technology available at the time. Recently, the read length of 454 pyrosequencing machines has been increased to an average of ∼250 bp. This allows for more flexibility in primer design and opens up the possibility of targeting regions of the 16S rRNA gene other than V6. In recent work, Huse et al. took advantage of this new capability to compare the classifications made for the human gut microbiome with the V6 and longer V3 regions (13). Plotting the taxonomic abundance of these two sequence sets against each other yielded an excellent correlation (r² = 0.99), suggesting that the choice of which variable region to target makes little difference. In this report, we introduce a data set examining the performance of sets of primers targeting the V1-V2, V6, and V6-V7 regions. By using a sample for which we have also generated a whole-genome shotgun sequencing run with 250 bp reads, we were able to compare the observed 16S rRNA genes in samples with and without an initial PCR step targeting the 16S rRNA gene. Our results demonstrate that experimental choices such as which region of the 16S rRNA gene to sequence and which algorithm to use to classify taxa are much more likely to affect observations of the “rare biosphere” than more commonly observed taxa.     

The Effect of Primer Choice and Short Read Sequences on the Outcome of 16S rRNA Gene Based Diversity Studies

Different regions of the bacterial 16S rRNA gene evolve at different evolutionary rates. The scientific outcome of short read sequencing studies therefore alters with the gene region sequenced. We wanted to gain insight in the impact of primer choice on the outcome of short read sequencing efforts. All the unknowns associated with sequencing data, i.e. primer coverage rate, phylogeny, OTU-richness and taxonomic assignment, were therefore implemented in one study for ten well established universal primers (338f/r, 518f/r, 799f/r, 926f/r and 1062f/r) targeting dispersed regions of the bacterial 16S rRNA gene. All analyses were performed on nearly full length and in silico generated short read sequence libraries containing 1175 sequences that were carefully chosen as to present a representative substitute of the SILVA SSU database. The 518f and 799r primers, targeting the V4 region of the 16S rRNA gene, were found to be particularly suited for short read sequencing studies, while the primer 1062r, targeting V6, seemed to be least reliable. Our results will assist scientists in considering whether the best option for their study is to select the most informative primer, or the primer that excludes interferences by host-organelle DNA. The methodology followed can be extrapolated to other primers, allowing their evaluation prior to the experiment.     

DegePrime,a Program for Degenerate Primer Design for Broad-Taxonomic-Range PCR in Microbial Ecology Studies

The taxonomic composition of a microbial community can be deduced by analyzing its rRNA gene content by, e.g., high-throughput DNA sequencing or DNA chips. Such methods typically are based on PCR amplification of rRNA gene sequences using broad-taxonomic-range PCR primers. In these analyses, the use of optimal primers is crucial for achieving an unbiased representation of community composition. Here, we present the computer program DegePrime that, for each position of a multiple sequence alignment, finds a degenerate oligomer of as high coverage as possible and outputs its coverage among taxonomic divisions. We show that our novel heuristic, which we call weighted randomized combination, performs better than previously described algorithms for solving the maximum coverage degenerate primer design problem. We previously used DegePrime to design a broad-taxonomic-range primer pair that targets the bacterial V3-V4 region (341F-805R) (D. P. Herlemann, M. Labrenz, K. Jurgens, S. Bertilsson, J. J. Waniek, and A. F. Andersson, ISME J. 5:1571–1579, 2011, http://dx.doi.org/10.1038/ismej.2011.41), and here we use the program to significantly increase the coverage of a primer pair (515F-806R) widely used for Illumina-based surveys of bacterial and archaeal diversity. By comparison with shotgun metagenomics, we show that the primers give an accurate representation of microbial diversity in natural samples.     

BIPES,a cost-effective high-throughput method for assessing microbial diversity

Pyrosequencing of 16S rRNA (16S) variable tags has become the most popular method for assessing microbial diversity, but the method remains costly for the evaluation of large numbers of environmental samples with high sequencing depths. We developed a barcoded Illumina paired-end (PE) sequencing (BIPES) method that sequences each 16S V6 tag from both ends on the Illumina HiSeq 2000, and the PE reads are then overlapped to obtain the V6 tag. The average accuracy of Illumina single-end (SE) reads was only 97.9%, which decreased from ∼99.9% at the start of the read to less than 85% at the end of the read; nevertheless, overlapping of the PE reads significantly increased the sequencing accuracy to 99.65% by verifying the 3′ end of each SE in which the sequencing quality was degraded. After the removal of tags with two or more mismatches within the medial 40–70 bases of the reads and of tags with any primer errors, the overall base sequencing accuracy of the BIPES reads was further increased to 99.93%. The BIPES reads reflected the amounts of the various tags in the initial template, but long tags and high GC tags were underestimated. The BIPES method yields 20–50 times more 16S V6 tags than does pyrosequencing in a single-flow cell run, and each of the BIPES reads costs less than 1/40 of a pyrosequencing read. As a laborsaving and cost-effective method, BIPES can be routinely used to analyze the microbial ecology of both environmental and human microbiomes.     

Analysis of 23S rRNA genes in metagenomes - a case study from the Global Ocean Sampling Expedition

As an evolutionary marker, 23S ribosomal RNA (rRNA) offers more diagnostic sequence stretches and greater sequence variation than 16S rRNA. However, 23S rRNA is still not as widely used. Based on 80 metagenome samples from the Global Ocean Sampling (GOS) Expedition, the usefulness and taxonomic resolution of 23S rRNA were compared to those of 16S rRNA. Since 23S rRNA is approximately twice as large as 16S rRNA, twice as many 23S rRNA gene fragments were retrieved from the GOS reads than 16S rRNA gene fragments, with 23S rRNA gene fragments being generally about 100 bp longer. Datasets for 16S and 23S rRNA sequences revealed similar relative abundances for major marine bacterial and archaeal taxa. However, 16S rRNA sequences had a better taxonomic resolution due to their significantly larger reference database.Reevaluation of the specificity of previously published PCR amplification primers and group specific fluorescence in situ hybridization probes on this metagenomic set of non-amplified 23S rRNA sequences revealed that out of 16 primers investigated, only two had more than 90% target group coverage. Evaluations of two probes, BET42a and GAM42a, were in accordance with previous evaluations, with a discrepancy in the target group coverage of the GAM42a probe when evaluated against the GOS metagenomic dataset.     

Characterizing homologues of crop domestication genes in poorly described wild relatives by high‐throughput sequencing of whole genomes

Wild crop relatives represent a source of novel alleles for crop genetic improvement. Screening biodiversity for useful or diverse gene homologues has often been based upon the amplification of targeted genes using available sequence information to design primers that amplify the target gene region across species. The crucial requirement of this approach is the presence of sequences with sufficient conservation across species to allow for the design of universal primers. This approach is often not successful with diverse organisms or highly variable genes. Massively parallel sequencing (MPS) can quickly produce large amounts of sequence data and provides a viable option for characterizing homologues of known genes in poorly described genomes. MPS of genomic DNA was used to obtain species‐specific sequence information for 18 rice genes related to domestication characteristics in a wild relative of rice, Microlaena stipoides. Species‐specific primers were available for 16 genes compared with 12 genes using the universal primer method. The use of species‐specific primers had the potential to cover 92% of the sequence of these genes, while traditional universal primers could only be designed to cover 80%. A total of 24 species‐specific primer pairs were used to amplify gene homologues, and 11 primer pairs were successful in capturing six gene homologues. The 23 million, 36‐base pair (bp) paired end reads, equated to an average of 2X genome coverage, facilitated the successful amplification and sequencing of six target gene homologues, illustrating an important approach to the discovery of useful genes in wild crop relatives.     

Taxonomic Classification of Bacterial 16S rRNA Genes Using Short Sequencing Reads: Evaluation of Effective Study Designs

Massively parallel high throughput sequencing technologies allow us to interrogate the microbial composition of biological samples at unprecedented resolution. The typical approach is to perform high-throughout sequencing of 16S rRNA genes, which are then taxonomically classified based on similarity to known sequences in existing databases. Current technologies cause a predicament though, because although they enable deep coverage of samples, they are limited in the length of sequence they can produce. As a result, high-throughout studies of microbial communities often do not sequence the entire 16S rRNA gene. The challenge is to obtain reliable representation of bacterial communities through taxonomic classification of short 16S rRNA gene sequences. In this study we explored properties of different study designs and developed specific recommendations for effective use of short-read sequencing technologies for the purpose of interrogating bacterial communities, with a focus on classification using naïve Bayesian classifiers. To assess precision and coverage of each design, we used a collection of ∼8,500 manually curated 16S rRNA gene sequences from cultured bacteria and a set of over one million bacterial 16S rRNA gene sequences retrieved from environmental samples, respectively. We also tested different configurations of taxonomic classification approaches using short read sequencing data, and provide recommendations for optimal choice of the relevant parameters. We conclude that with a judicious selection of the sequenced region and the corresponding choice of a suitable training set for taxonomic classification, it is possible to explore bacterial communities at great depth using current technologies, with only a minimal loss of taxonomic resolution.     

Estimating Bacterial Diversity for Ecological Studies: Methods,Metrics, and Assumptions

Methods to estimate microbial diversity have developed rapidly in an effort to understand the distribution and diversity of microorganisms in natural environments. For bacterial communities, the 16S rRNA gene is the phylogenetic marker gene of choice, but most studies select only a specific region of the 16S rRNA to estimate bacterial diversity. Whereas biases derived from from DNA extraction, primer choice and PCR amplification are well documented, we here address how the choice of variable region can influence a wide range of standard ecological metrics, such as species richness, phylogenetic diversity, β-diversity and rank-abundance distributions. We have used Illumina paired-end sequencing to estimate the bacterial diversity of 20 natural lakes across Switzerland derived from three trimmed variable 16S rRNA regions (V3, V4, V5). Species richness, phylogenetic diversity, community composition, β-diversity, and rank-abundance distributions differed significantly between 16S rRNA regions. Overall, patterns of diversity quantified by the V3 and V5 regions were more similar to one another than those assessed by the V4 region. Similar results were obtained when analyzing the datasets with different sequence similarity thresholds used during sequences clustering and when the same analysis was used on a reference dataset of sequences from the Greengenes database. In addition we also measured species richness from the same lake samples using ARISA Fingerprinting, but did not find a strong relationship between species richness estimated by Illumina and ARISA. We conclude that the selection of 16S rRNA region significantly influences the estimation of bacterial diversity and species distributions and that caution is warranted when comparing data from different variable regions as well as when using different sequencing techniques.     

Design and Experimental Application of a Novel Non-Degenerate Universal Primer Set that Amplifies Prokaryotic 16S rRNA Genes with a Low Possibility to Amplify Eukaryotic rRNA Genes

The deep sequencing of 16S rRNA genes amplified by universal primers has revolutionized our understanding of microbial communities by allowing the characterization of the diversity of the uncultured majority. However, some universal primers also amplify eukaryotic rRNA genes, leading to a decrease in the efficiency of sequencing of prokaryotic 16S rRNA genes with possible mischaracterization of the diversity in the microbial community. In this study, we compared 16S rRNA gene sequences from genome-sequenced strains and identified candidates for non-degenerate universal primers that could be used for the amplification of prokaryotic 16S rRNA genes. The 50 identified candidates were investigated to calculate their coverage for prokaryotic and eukaryotic rRNA genes, including those from uncultured taxa and eukaryotic organelles, and a novel universal primer set, 342F-806R, covering many prokaryotic, but not eukaryotic, rRNA genes was identified. This primer set was validated by the amplification of 16S rRNA genes from a soil metagenomic sample and subsequent pyrosequencing using the Roche 454 platform. The same sample was also used for pyrosequencing of the amplicons by employing a commonly used primer set, 338F-533R, and for shotgun metagenomic sequencing using the Illumina platform. Our comparison of the taxonomic compositions inferred by the three sequencing experiments indicated that the non-degenerate 342F-806R primer set can characterize the taxonomic composition of the microbial community without substantial bias, and is highly expected to be applicable to the analysis of a wide variety of microbial communities.     

Is the whole greater than the sum of its parts? De novo assembly strategies for bacterial genomes based on paired-end sequencing

Background

Whole genome sequence construction is becoming increasingly feasible because of advances in next generation sequencing (NGS), including increasing throughput and read length. By simply overlapping paired-end reads, we can obtain longer reads with higher accuracy, which can facilitate the assembly process. However, the influences of different library sizes and assembly methods on paired-end sequencing-based de novo assembly remain poorly understood.

Results

We used 250 bp Illumina Miseq paired-end reads of different library sizes generated from genomic DNA from Escherichia coli DH1 and Streptococcus parasanguinis FW213 to compare the assembly results of different library sizes and assembly approaches. Our data indicate that overlapping paired-end reads can increase read accuracy but sometimes cause insertion or deletions. Regarding genome assembly, merged reads only outcompete original paired-end reads when coverage depth is low, and larger libraries tend to yield better assembly results. These results imply that distance information is the most critical factor during assembly. Our results also indicate that when depth is sufficiently high, assembly from subsets can sometimes produce better results.

Conclusions

In summary, this study provides systematic evaluations of de novo assembly from paired end sequencing data. Among the assembly strategies, we find that overlapping paired-end reads is not always beneficial for bacteria genome assembly and should be avoided or used with caution especially for genomes containing high fraction of repetitive sequences. Because increasing numbers of projects aim at bacteria genome sequencing, our study provides valuable suggestions for the field of genomic sequence construction.

Electronic supplementary material

The online version of this article (doi:10.1186/s12864-015-1859-8) contains supplementary material, which is available to authorized users.     

Mock microbial community meta-analysis using different trimming of amplicon read lengths

Trimming of sequencing reads is a pre-processing step that aims to discard sequence segments such as primers, adapters and low quality nucleotides that will interfere with clustering and classification steps. We evaluated the impact of trimming length of paired-end 16S and 18S rRNA amplicon reads on the ability to reconstruct the taxonomic composition and relative abundances of communities with a known composition in both even and uneven proportions. We found that maximizing read retention maximizes recall but reduces precision by increasing false positives. The presence of expected taxa was accurately predicted across broad trim length ranges but recovering original relative proportions remains a difficult challenge. We show that parameters that maximize taxonomic recovery do not simultaneously maximize relative abundance accuracy. Trim length represents one of several experimental parameters that have non-uniform impact across microbial clades, making it a difficult parameter to optimize. This study offers insights, guidelines, and helps researchers assess the significance of their decisions when trimming raw reads in a microbiome analysis based on overlapping or non-overlapping paired-end amplicons.     

Systematic Design of 18S rRNA Gene Primers for Determining Eukaryotic Diversity in Microbial Consortia

High-throughput sequencing of ribosomal RNA gene (rDNA) amplicons has opened up the door to large-scale comparative studies of microbial community structures. The short reads currently produced by massively parallel sequencing technologies make the choice of sequencing region crucial for accurate phylogenetic assignments. While for 16S rDNA, relevant regions have been well described, no truly systematic design of 18S rDNA primers aimed at resolving eukaryotic diversity has yet been reported. Here we used 31,862 18S rDNA sequences to design a set of broad-taxonomic range degenerate PCR primers. We simulated the phylogenetic information that each candidate primer pair would retrieve using paired- or single-end reads of various lengths, representing different sequencing technologies. Primer pairs targeting the V4 region performed best, allowing discrimination with paired-end reads as short as 150 bp (with 75% accuracy at genus level). The conditions for PCR amplification were optimised for one of these primer pairs and this was used to amplify 18S rDNA sequences from isolates as well as from a range of environmental samples which were then Illumina sequenced and analysed, revealing good concordance between expected and observed results. In summary, the reported primer sets will allow minimally biased assessment of eukaryotic diversity in different microbial ecosystems.     

Evaluation of PCR Primer Selectivity and Phylogenetic Specificity by Using Amplification of 16S rRNA Genes from Betaproteobacterial Ammonia-Oxidizing Bacteria in Environmental Samples

The effect of primer specificity for studying the diversity of ammonia-oxidizing betaproteobacteria (βAOB) was evaluated. βAOB represent a group of phylogenetically related organisms for which the 16S rRNA gene approach is especially suitable. We used experimental comparisons of primer performance with water samples, together with an in silico analysis of published sequences and a literature review of clone libraries made with four specific PCR primers for the βAOB 16S rRNA gene. With four aquatic samples, the primers NitA/NitB produced the highest frequency of ammonia-oxidizing-bacterium-like sequences compared to clone libraries with products amplified with the primer combinations βAMOf/βAMOr, βAMOf/Nso1255g, and NitA/Nso1225g. Both the experimental examination of ammonia-oxidizing-bacterium-specific 16S rRNA gene primers and the literature search showed that neither specificity nor sensitivity of primer combinations can be evaluated reliably only by sequence comparison. Apparently, the combination of sequence comparison and experimental data is the best approach to detect possible biases of PCR primers. Although this study focused on βAOB, the results presented here more generally exemplify the importance of primer selection and potential primer bias when analyzing microbial communities in environmental samples.     

Analysis of the Cystic Fibrosis Lung Microbiota via Serial Illumina Sequencing of Bacterial 16S rRNA Hypervariable Regions

The characterization of bacterial communities using DNA sequencing has revolutionized our ability to study microbes in nature and discover the ways in which microbial communities affect ecosystem functioning and human health. Here we describe Serial Illumina Sequencing (SI-Seq): a method for deep sequencing of the bacterial 16S rRNA gene using next-generation sequencing technology. SI-Seq serially sequences portions of the V5, V6 and V7 hypervariable regions from barcoded 16S rRNA amplicons using an Illumina short-read genome analyzer. SI-Seq obtains taxonomic resolution similar to 454 pyrosequencing for a fraction of the cost, and can produce hundreds of thousands of reads per sample even with very high multiplexing. We validated SI-Seq using single species and mock community controls, and via a comparison to cystic fibrosis lung microbiota sequenced using 454 FLX Titanium. Our control runs show that SI-Seq has a dynamic range of at least five orders of magnitude, can classify >96% of sequences to the genus level, and performs just as well as 454 and paired-end Illumina methods in estimation of standard microbial ecology diversity measurements. We illustrate the utility of SI-Seq in a pilot sample of central airway secretion samples from cystic fibrosis patients.     

Impact of training sets on classification of high-throughput bacterial 16s rRNA gene surveys

Taxonomic classification of the thousands–millions of 16S rRNA gene sequences generated in microbiome studies is often achieved using a naïve Bayesian classifier (for example, the Ribosomal Database Project II (RDP) classifier), due to favorable trade-offs among automation, speed and accuracy. The resulting classification depends on the reference sequences and taxonomic hierarchy used to train the model; although the influence of primer sets and classification algorithms have been explored in detail, the influence of training set has not been characterized. We compared classification results obtained using three different publicly available databases as training sets, applied to five different bacterial 16S rRNA gene pyrosequencing data sets generated (from human body, mouse gut, python gut, soil and anaerobic digester samples). We observed numerous advantages to using the largest, most diverse training set available, that we constructed from the Greengenes (GG) bacterial/archaeal 16S rRNA gene sequence database and the latest GG taxonomy. Phylogenetic clusters of previously unclassified experimental sequences were identified with notable improvements (for example, 50% reduction in reads unclassified at the phylum level in mouse gut, soil and anaerobic digester samples), especially for phylotypes belonging to specific phyla (Tenericutes, Chloroflexi, Synergistetes and Candidate phyla TM6, TM7). Trimming the reference sequences to the primer region resulted in systematic improvements in classification depth, and greatest gains at higher confidence thresholds. Phylotypes unclassified at the genus level represented a greater proportion of the total community variation than classified operational taxonomic units in mouse gut and anaerobic digester samples, underscoring the need for greater diversity in existing reference databases.