- Open Access
Latitude in sample handling and storage for infant faecal microbiota studies: the elephant in the room?
- Alexander G. Shaw†1Email author,
- Kathleen Sim†1,
- Elizabeth Powell1,
- Emma Cornwell1,
- Teresa Cramer2,
- Zoë E. McClure1,
- Ming-Shi Li1 and
- J. Simon Kroll1Email author
© The Author(s). 2016
Received: 8 March 2016
Accepted: 21 July 2016
Published: 30 July 2016
In this manuscript, we investigate the “stones best left unturned” of sample storage and preparation and their implications for the next-generation sequencing of infant faecal microbial communities by the 16S ribosomal ribonucleic acid (rRNA) gene.
We present a number of experiments that investigate the potential effects of often overlooked methodology factors, establishing a “normal” degree of variation expected between replica sequenced samples. Sources of excess variation are then identified, as measured by observation of alpha diversity, taxonomic group counts and beta diversity magnitudes between microbial communities.
Extraction of DNA from samples on different dates, by different people and even using varied sample weights results in little significant difference in downstream sequencing data. A key assumption in many studies is the stability of samples stored long term at −80 °C prior to extraction. After 2 years, we see relatively few changes: increased abundances of lactobacilli and bacilli and a reduction in the overall OTU count. Where samples cannot be frozen, we find that storing samples at room temperature does lead to significant changes in the microbial community after 2 days. Mailing of samples during this time period (a common form of sample collection from outpatients for example) does not lead to any additional variation.
Important methodological standards can be drawn from these results; painstakingly created archives of infant faecal samples stored at −80 °C are still largely representative of the original community and varying factors in DNA extraction methodology have comparatively little effect on overall results. Samples taken should ideally be either frozen at −80 °C or extracted within 2 days if stored at room temperature, with mail samples being mailed on the day of collection.
Developments in massively parallel sequencing technology have transformed our ability to study complex bacterial populations. They offer penetrating insights into human, animal and environmental microbial ecology through the analysis (for example by 16S rRNA gene sequencing) of large collections of samples, often accumulated in the field over an extended period of time. The new technology presents new challenges in the handling of these samples. Samples—which may vary in size from milligrammes to grammes—may be in transit for hours to days under variable ambient conditions before reaching the laboratory, where whilst in some cases, DNA may be extracted at once and held for later analysis; in others, samples may be stored frozen for variable (and sometimes extended) periods before processing. There are concerns about the changes that these environmental factors introduce, different aspects of which have been addressed by various investigators [1–9].
In the course of a study of the developing infant faecal microbiota, extending over 3 years, we have faced many of these issues. Some of our early infant faecal samples have been minute. Whilst some have been collected in hospital, stored quickly at −20 °C and shortly afterwards at −80 °C, others have been collected at the infant’s home by a research nurse and transported back to the laboratory at room temperature (~20 °C) only reaching frozen storage at the end of the day. As the study progressed, a practical solution to the collection of samples from our increasingly dispersed subject population has been for parents to collect them at home and mail them (suitably packaged) via the UK mail system. With no environmental control in transit, these packages arrive in our Institution’s mail room, where the temperature regularly reaches 30 °C. By whatever means the samples reach the laboratory, they are subsequently stored at −80 °C, some for many months, before further processing.
Now we report the results of an investigation of the impact of these different factors on data describing the microbial communities of the samples on the basis of 16S rRNA gene sequence.
Data analysed in this study is derived from 103 faecal samples collected from two infant cohorts. The first cohort consists of infants participating in the Neonatal Microbiota (NeoM) study, who were born prematurely (<32 completed weeks of gestation). Faecal samples were collected between the time of the baby’s admission to the NICU at St Mary’s Hospital or Queen Charlotte’s and Chelsea Hospital and discharge. The cohort is fully described in Sim et al. . Samples used here were collected from eight infants aged between 2 and 11 weeks. Such samples represent relatively simple bacterial communities. The second cohort consists of healthy, full-term infants born at Saint Mary’s Hospital. Samples used were collected from six infants between the ages of 13 and 19 months. Compared to faecal samples from premature infants, these samples represent a richer, more complex bacterial community.
Samples from premature infants were collected by nursing staff from diapers using a sterile spatula, placed in a sterile DNAase- and RNAase-free Eppendorf tube. Samples from term infants were collected by research nurses from diapers using a sterile scoop, placed in a sterile DNAase- and RNAase-free storage pots. Samples were homogenised with sterile microbiology loops prior to DNA extraction, in light of potential within-sample variation .
Bacterial DNA extraction
Faecal samples (200 mg unless otherwise specified) were processed using one of two methods. Faecal samples from premature infants processed for the long-term freezer storage experiment were processed using “Protocol 1”. In brief, enzymatic digestion of the sample with 10 μl lysozyme (75 mg/ml, Sigma) for 30 min at 37 °C was followed by bead-beating on a FastPrep homogeniser using Lysing Matrix B tubes (MP Biomedicals), with a setting of 6.0 m/s for 45 s (in 3 × 15-s pulses)). Samples were incubated with 2.4 μl RNase A (10 mg/ml) and 6 μl proteinase K (20 mg/ml) for 56 °C for 90 min, followed by a further bead-beating 15-s pulse. DNA was recovered using a phenol-chloroform extraction and purified with the QIAamp DNA mini kit (QIAGEN). All other samples were processed with “Protocol 2”. Here, the FastDNA SPIN Kit for Soil (MP Biomedicals) was used, incorporating bead-beating on a FastPrep homogeniser and performed following the manufacturer’s protocol except that the final elution step was into TRIS (10 mM) low-ethylenediaminetetraacetic acid (EDTA) (0.1 mM) buffer. Negative controls (buffer, no added faecal sample) were processed similarly.
Polymerase chain reaction amplification and sequencing of variable regions of the bacterial 16S rRNA gene
In this study, we employed two next-generation sequencing platforms—Roche 454 FLX and Illumina Miseq.
For samples analysed using the 454 FLX platform (samples involved in the sample weight experiment), the V3–V5 region of bacterial 16S rRNA genes was amplified from each DNA sample in quadruplicate using a primer pair tagged with individually unique 12-bp error-correcting Golay barcodes [11, 12]. Polymerase chain reaction (PCR) was performed as previously described . Replicate amplicons were pooled and purified, and pyrosequencing runs were carried out on a 454 Life Sciences GS FLX (Roche) following the Roche Amplicon Lib-L protocol. Replicate samples spread over all sequencing runs acted as internal controls.
For samples analysed using the Illumina Miseq platform (all other samples), the V4 region was amplified using the primers 520F “AYTGGGYDTAAAGNG” and 802R “TACNVGGGTATCTAATCC” , following a dual barcoding approach  using 8-bp Nextera Version 2 barcodes (Illumina). Quadruplicate PCR reactions were performed as follows: reaction mixes consisted of 12.5 μl of Q5 Master Mix (New England BioScience), 1 μl of extracted DNA, 5 μl of each primer pair at 1.5 μM and 1.5 μl of ultrapure water (Cambio). Thermocycler was operated at 95 °C for 2 min, followed by 35 cycles of 95 °C for 20 s, 50 °C for 20 s and 72 °C for 5 min. The reactions were then pooled and purified with sequencing carried out on a Miseq desktop sequencer (Illumina) using a 10 pM library and a 15 % PhiX spike-in, following the standard protocol for Nextera dual-indexing sequencing with V2 kits.
Negative controls were included in all sequencing runs to identify potential contamination.
Four hundred fifty-four shotgun-processed data were denoised using AmpliconNoise  as part of the “Quantitative Insights Into Microbial Ecology” (QIIME)  package followed by chimera removal with Perseus  and demultiplexing. QIIME was also used to join dual-indexed Illumina reads into single reads and demultiplex, with chimera removal performed by USEARCH . Sequences were clustered at 97 % sequence identity using the UCLUST algorithm  into operational taxonomic units (OTUs) and aligned by reference to the SILVA rRNA database 119 release  (separate processing for each platform). The OTU tables were filtered to remove singletons (sequences present only once in the dataset). Rarefaction was performed, removing heterogeneity of sequencing reads per sample (1622 reads for 454 data, 4356 reads for Illumina). The 454 dataset comprised 294 OTUs in total, and the Illumina dataset comprised 5625 OTUs. Measures of alpha diversity (Shannon-Weaver index, Chao1, total OTUs and phylogenetic diversity) and beta diversity (unweighted and weighted UniFrac distances, Bray-Curtis dissimilarity and Jaccard dissimilarity) were calculated. The rarefied OTU tables were summarised separately to phylum level using the QIIME script summarize_taxa.py.
The datasets supporting the conclusions of this article are available in the European Nucleotide Archive repository, PRJEB6345 (454 data, http://www.ebi.ac.uk/ena/data/view/PRJEB6345) and PRJEB10940 (Illumina data, http://www.ebi.ac.uk/ena/data/view/PRJEB10940).
Alpha and beta diversity measures were analysed using general linear models (GLMs) to determine associations with tested factors. OTUs were analysed using GLMs with a negative binomial distribution. Analysis was performed for all OTUs comprising reads >1 % of total reads from all samples in the involved dataset and OTUs comprising >5 % of any individual sample’s reads. Where significant associations were found, further analysis compared the baseline (the earliest timepoint or the standard weight) and each subsequent category. These analyses were performed using the Wilcoxon signed-rank test for paired data and the Mann-Whitney U test for unpaired data (each due to small sample sizes and uneven degrees of variance in the data). The extraction similarity and mail comparison analyses were performed solely using the Mann-Whitney U test (rather than GLMs) due to the small numbers of samples involved. Beta diversity measures were compared to standard values derived from replica sequenced samples (see below) using the Mann-Whitney U test.
GLMs included sample infant identification number as an additional variable. P values are presented prior to multiple hypothesis correction (MHC) where applicable, with data tables indicating whether results would still be significant after a Bonferroni correction.
Establishing baseline variation
Illumina Miseq: ten replicate samples from infant 14 were sequenced on a single Miseq run.
Roche 454 FLX: 42 samples were sequenced on two different sequencing runs.
Beta diversity measures for these baseline samples (henceforth described as “standard” variation between the microbial communities of a single faecal sample that is sequenced twice) are shown in Additional files 1 (Illumina Miseq) and 2 (Roche 454 FLX).
To determine the amount of variation that can be expected due to the DNA extraction process, one faecal sample from a term infant (Infant 14) was homogenised and split into ten aliquots. Both sets of five aliquots were frozen at −80 °C, with the first set being extracted after storage for 24 days, and the second set being extracted after storage for 115 days. A separate researcher performed a DNA extraction on each set of aliquots, using extraction kits with a different lot number. The aim of this process was to simulate ‘typical’ laboratory operation, incorporating the differences that would occur when performing multiple DNA extractions (time delays, different staff members and different kit reagents).
To investigate the effects of varying sample weight, fresh faecal samples from four premature babies were each split into four aliquots (25, 50, 100 and 200 mg) prior to freezing at −80 °C for up to 3 weeks before DNA extraction and sequencing.
To investigate the effects of long-term freezer storage, fresh faecal samples from four premature babies were each split into five aliquots, left at room temperature for between 4 and 8 h (simulating the potential collection and processing time period) and then frozen at −80 °C for between 2 months and 2 before DNA extraction and sequencing.
To investigate the effects of room temperature storage, five faecal samples from term infants and four from preterm infants were split into aliquots and stored at room temperature. At regular timepoints, an aliquot of faeces from each infant was moved to −80 °C and the entire set underwent DNA extraction 2 months after the experiment began.
To identify variation that may arise from the mailing of faecal samples, samples from five term babies were split into three aliquots. One aliquot from each baby was mailed to the laboratory from varied locations across London. Two aliquots were left at room temperature, one of which was frozen at −80 °C after 4 h and the other (a mail match) was frozen along with the mailed aliquot when it arrived back at the laboratory.
Summary of study experiments
Faecal sample source
DNA extraction protocol
Variation between extractions
Effects of varied weight
Infants 1, 2, 3 and 4
Roche 454 GS FLX
Effects of long term freezing
Infants 5, 6, 7 and 8
Effects of room temperature storage
Premature and term infants
Infants 5, 6, 7, 8, 9, 10, 11, 12 and 13
Effects of mailing samples
Infants 9, 10, 11, 12 and 13
How much variation can be expected between extractions?
To investigate the validity of comparing samples undergoing separate DNA extractions over the course of a typical laboratory schedule (differing people, dates and kit lot numbers), a single sample was split into ten aliquots and processed in two batches.
How small an amount of sample is sufficient?
A common problem facing studies that make use of precious samples that are obtainable only in small quantities is the issue of standardisation of sample weight. Such samples are too valuable to be ignored, but their inclusion risks potential weight related bias through undersampling. Alternatively, studies may utilise samples that prove extremely difficult to portion into desired weights; 200–300 mg is a commonly used range of sample weight, although this may be impossible to obtain in some studies. To investigate potential effects, four faecal samples from premature infants were split into a variety of weights prior to DNA extraction and sequencing and the resulting bacterial communities compared.
What are the effects of long term freezing?
The current ‘gold standard’ for long-term storage of samples is assumed to be freezing at −80 °C. However, studies have not investigated the potential long term effects on the microbial community. Given that studies may take years to collect samples, and ideally will prove to be a reliable resource for years to come, it is essential to quantify any variation that may occur. We investigated the effects of storage at −80 °C for up to 2 years on the derived microbial community.
OTUs with differential abundances over long term storage at −80 °C
Mean reads (% of group total) at:
Significant after a MHC?
2.5 % CI
97.5 % CI
What is the effect of room temperature storage on microbial communities?
Concerns have been raised concerning the stability of DNA when samples are kept at room temperature and resulting effects on microbial communities. Whilst studies have looked at specific timepoints, the time prior to storage of samples at −80 °C will be likely to vary greatly in studies involving hundreds of samples. We investigated the effects of room temperature storage on both simple and relatively complex communities (pre-term and term faecal samples, respectively) over a detailed timecourse. The two sample sets were analysed and independently.
OTUs in the simple community dataset that displayed differential abundances over the course of two weeks of storage at room temperature
Mean reads (% of dataset total) at:
Significant after a MHC?
2.5 % CI
97.5 % CI
OTUs in the more complex community dataset that displayed differential abundances over the course of 2 weeks of storage at room temperature
Mean reads (% of dataset total) at:
Significant after a MHC?
2.5 % CI
97.5 % CI
Does the mailing of faecal samples lead to microbial community divergence?
Environmental or population-based studies often require the collection of samples over a wide-geographic area. In such scenarios, sample transfer via the mail system can prove to be and extremely useful method for returning samples to the laboratory. Whilst we have shown that if the microbial community is relatively stable at room temperature for up to 2 days, the mail system itself may be a source of considerable environmental variation—our own university mail room has been known to reach 30 °C (86 °F) on an otherwise cool September day.
Mailed aliquots of samples were compared to two paired aliquots. One was stored at room temperature for 4 h prior to −80 °C storage (baseline samples). The second (a mail match) was stored at room temperature until the paired mail aliquot arrived at the laboratory. No significant differences in alpha diversity were found between the mail and mail match sample groups. Measures of beta diversity between baseline samples and each group were not significantly different (see Additional file 7: Figures S5, S6, S7 and S8 for PCoA plots). No differentially abundant OTUs were found between the mail and mail match groups (see Additional file 9: Figure S10), and no differences were found in phyla abundance (see Additional file 10: Figure S11).
Discussion and conclusion
A number of studies have looked at sample compositional stability associated with different conditions of storage. Carroll et al.  have provided the current benchmark for long-term −80 °C storage, showing no significant changes in the faecal microbial community after 6 months of storage, with Bai et al.  reporting similar findings for the vaginal microbiota after a shorter 4 weeks of storage. For some studies, however, samples may need to be accrued over far longer periods, and the impact of longer term storage has not until now been systematically appraised. Here, we report that 2 years at −80 °C leads to few significant changes in the microbial community of infant faecal samples, with a small reduction in the observed number OTUs and shifts in the abundance of some specific OTUs, although these were often not shared amongst similar lineages. The increases in abundance of the Lactobacillus and bacilli OTUs are a notable exception to this trend, and it may be important to consider the effects of long-term storage of samples where studies focus on these organisms. We also note that the results presented may not be representative of more complex mixtures and that the effects of long-term storage have only been described for the specific organisms featuring in our dataset.
Carroll et al.  and Cardona et al.  have investigated sample stability at room temperature. Both groups demonstrated compositional stability of faecal samples over a 24-h period. A lack of significant change in the relative proportions of different bacterial phyla has been reported over a 3-day period of storage at room temperature by Dominianni et al. . Voigt et al.  and Flores et al.  have demonstrated stability out to 7 days when samples are preserved in the proprietary Stabilization Solution “RNAlater”. These findings have appeared to represent the limit. Cardona et al.  have shown that significant changes in the community are seen in samples analysed after 2 weeks at room temperature. Here, we have investigated the compositional stability of infant faecal samples held at room temperature for periods out to 2 weeks and confirm the findings of Carrol et al., Cardona et al. and Dominianni et al. [2, 4, 7] that by 1 week, significant skewing of the observed population has occurred. In our more complex community samples, this primarily affects the abundance of some elements of the bacterial community, with our analysis indicating significant shifts in the weighted measures of beta diversity by 48 h. The most extreme shifts appeared to be correlated with increasing relative abundance of Enterobacteriaceae (specifically an Enterobacter OTU). The less complex community samples were more strongly affected by loss and gain of low abundance OTUs, leading to significant increases in unweighted measures. Considerable differences in both bacterial OTUs and phyla were evident by 1 and 2 weeks of room temperature storage, and although no particular trends unified which groups are likely to be affected, increased proportional abundance of Bifidobacterium and decreased Veillonella were seen in both datasets.
We conclude accordingly that for preservation of bacterial community structure, faecal samples should be frozen within 2 days of collection. In the UK, this happily allows time for samples to be delivered to the laboratory by mail without loss of quality, provided that they are dispatched on the day of collection by first class mail and care is taken to avoid weekends. In settings where the mail service is unreliable, alternative means of sample delivery should be considered to ensure that quality is maintained.
By carrying out replicate DNA extractions (different researchers, different lots of reagents), we have concluded that this part of the overall protocol contributes negligibly to variation.
Reduction in faecal sample weight, even down to only an eighth of the generally applied ‘standard’ of 200 mg, also scarcely affected alpha diversity and the majority of beta diversity measures, noting that this related to our study of infant samples of inherently low diversity. The strongest effects were seen when considering weighted beta diversity measures, with OTU abundances suffering increased variation as lower amounts of faeces are sampled. However, despite variation being higher than expected, the actual magnitude of these changes appears to be relatively small, with samples still retaining close similarity. We consider that it would be prudent to establish on individual study basis a minimum sample size for DNA extraction to yield consistent results.
EDTA, ethylenediaminetetraacetic acid; GLM, general linear models; MHC, multiple hypothesis correction; OTU, operational taxonomic unit; PCoA, principal coordinates analysis; PCR, polymerase chain reaction; QIIME, quantitative insights into microbial ecology; rRNA, ribosomal ribonucleic acid; TRIS, tris(hydroxymethyl)aminomethane
We thank our colleagues at the Imperial College Healthcare NHS Trust and the participants and their families, for their contributions to this study.
This study was supported by a programme grant (to JSK) from The Winnicott Foundation and generous additional funding from Micropathology Ltd., Meningitis Now and the National Institute for Health Research (NIHR) Biomedical Research Centre based at Imperial Healthcare NHS Trust and Imperial College London.
AGS wrote the draft manuscript, processed samples for next-generation sequencing and analysed the data. KS conducted the NeoM study, designed the experiments and processed the samples for next-generation sequencing. EP conducted the DORMICe study, designed the experiments and processed the samples for next-generation sequencing. EC processed the samples for next-generation sequencing. TC designed the experiments and processed the samples for next-generation sequencing. ZEM collected the samples required for the study. MSL guided the experimental design and sample processing methodology. JSK led both the NeoM and DORMICe studies and guided the experimental design. All authors critically reviewed and approved the manuscript.
The authors declare that they have no competing interests.
Consent for publication
Ethics approval and consent to participate
The study ‘Defining the Intestinal Microbiota in Premature Infants’ (ClinicalTrials.gov Identifier NCT01102738) was approved by West London Research Ethics Committee 2, United Kingdom (Reference number: 10/H0711/39). Parents gave written informed consent for their infant to participate in the study.
The study ‘Development Of Respiratory Microbiota In Children’ was approved by Riverside Ethics Committee, London, UK (Reference number: 12/LO/1362). Parents gave written informed consent for their infant to participate in the study.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Voigt AY, Costea PI, Kultima JR, Li SS, Zeller G, Sunagawa S, Bork P. Temporal and technical variability of human gut metagenomes. Genome Biol. 2015;16:73.View ArticlePubMedPubMed CentralGoogle Scholar
- Carroll IM, Ringel-Kulka T, Siddle JP, Klaenhammer TR, Ringel Y. Characterization of the fecal microbiota using high-throughput sequencing reveals a stable microbial community during storage. PLoS One. 2012;7:e46953.View ArticlePubMedPubMed CentralGoogle Scholar
- Bai G, Gajer P, Nandy M, Ma B, Yang H, Sakamoto J, Blanchard MH, Ravel J, Brotman RM. Comparison of storage conditions for human vaginal microbiome studies. PLoS One. 2012;7:e36934.View ArticlePubMedPubMed CentralGoogle Scholar
- Cardona S, Eck A, Cassellas M, Gallart M, Alastrue C, Dore J, Azpiroz F, Roca J, Guarner F, Manichanh C. Storage conditions of intestinal microbiota matter in metagenomic analysis. BMC Microbiol. 2012;12:158.View ArticlePubMedPubMed CentralGoogle Scholar
- Fouhy F, Deane J, Rea MC, O'Sullivan O, Ross RP, O'Callaghan G, Plant BJ, Stanton C. The effects of freezing on faecal microbiota as determined using MiSeq sequencing and culture-based investigations. PLoS One. 2015;10:e0119355.View ArticlePubMedPubMed CentralGoogle Scholar
- Flores R, Shi J, Yu G, Ma B, Ravel J, Goedert JJ, Sinha R. Collection media and delayed freezing effects on microbial composition of human stool. Microbiome. 2015;3:33.View ArticlePubMedPubMed CentralGoogle Scholar
- Dominianni C, Wu J, Hayes RB, Ahn J. Comparison of methods for fecal microbiome biospecimen collection. BMC Microbiol. 2014;14:103.View ArticlePubMedPubMed CentralGoogle Scholar
- Bahl MI, Bergstrom A, Licht TR. Freezing fecal samples prior to DNA extraction affects the Firmicutes to Bacteroidetes ratio determined by downstream quantitative PCR analysis. FEMS Microbiol Lett. 2012;329:193–7.View ArticlePubMedGoogle Scholar
- Lauber CL, Zhou N, Gordon JI, Knight R, Fierer N. Effect of storage conditions on the assessment of bacterial community structure in soil and human-associated samples. FEMS MicrobiolLett. 2010;307:80–6.View ArticleGoogle Scholar
- Sim K, Shaw AG, Randell P, Cox MJ, McClure ZE, Li MS, Haddad M, Langford PR, Cookson WO, Moffatt MF, Kroll JS. Dysbiosis anticipating necrotizing enterocolitis in very premature infants. Clin Infect Dis. 2015;60:389–97.View ArticlePubMedGoogle Scholar
- Fierer N, Hamady M, Lauber CL, Knight R. The influence of sex, handedness, and washing on the diversity of hand surface bacteria. ProcNatlAcadSciUSA. 2008;105:17994–9.View ArticleGoogle Scholar
- Sim K, Cox MJ, Wopereis H, Martin R, Knol J, Li MS, Cookson WO, Moffatt MF, Kroll JS. Improved detection of bifidobacteria with optimised 16S rRNA-gene based pyrosequencing. PLoS One. 2012;7:e32543.View ArticlePubMedPubMed CentralGoogle Scholar
- Klindworth A, Pruesse E, Schweer T, Peplies J, Quast C, Horn M, Glockner FO. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res. 2013;41:e1.View ArticlePubMedGoogle Scholar
- Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the MiSeq Illumina sequencing platform. Appl Environ Microbiol. 2013;79:5112–20.View ArticlePubMedPubMed CentralGoogle Scholar
- Quince C, Lanzen A, Davenport RJ, Turnbaugh PJ. Removing noise from pyrosequenced amplicons. BMC Bioinformatics. 2011;12:38.View ArticlePubMedPubMed CentralGoogle Scholar
- Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Pena AG, Goodrich JK, Gordon JI, et al. QIIME allows analysis of high-throughput community sequencing data. NatMethods. 2010;7:335–6.Google Scholar
- Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 2010;26:2460–1.View ArticlePubMedGoogle Scholar
- Pruesse E, Quast C, Knittel K, Fuchs BM, Ludwig W, Peplies J, Glockner FO. SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 2007;35:7188–96.View ArticlePubMedPubMed CentralGoogle Scholar