16S Amplicon Sequencing · V3–V4 Microbiome Profiling

Characterise microbial communities. From gut to soil.

End-to-end 16S rRNA amplicon sequencing and microbiome data analysis targeting the hypervariable V3–V4 region — a robust, widely validated approach for characterising bacterial community composition across complex samples.

Plan Your Project → See the Workflow
V3–V4
Hypervariable region targeted
DADA2
ASV inference engine
3
Reference DBs: SILVA · GG2 · GTDB
PICRUSt2
Functional prediction available
About the Service

Microbiome profiling via 16S V3–V4

Nucleome Informatics offers end-to-end 16S rRNA Amplicon Sequencing and microbiome data analysis services for a wide range of environmental and biological sample types.

Targeting the hypervariable V3–V4 region of the bacterial 16S rRNA gene provides a robust and widely validated approach for characterising microbial community composition across complex samples.

Our pipeline combines rigorous quality control, DADA2-based amplicon sequence variant (ASV) inference, taxonomic classification against curated reference databases, and a comprehensive suite of diversity metrics and community-level comparisons. Functional potential prediction using PICRUSt2 is available as an extension for suitable study designs.

  • Sequencing data quality assessment and preprocessing
  • DADA2-based denoising and ASV inference
  • Taxonomic classification (SILVA, Greengenes2, GTDB)
  • Alpha diversity analysis — within-sample diversity metrics
  • Beta diversity analysis — between-sample community structure
  • Differential abundance analysis across sample groups
  • Functional potential prediction using PICRUSt2 (where applicable)
  • Phyloseq-based visualisation and community analysis
  • Secure delivery of raw, intermediate and final analysis files

DADA2-Based ASV Inference

High-resolution amplicon sequence variant calling resolves closely related taxa without the clustering artefacts of OTU-based approaches — delivering finer taxonomic discrimination.


Multi-Database Classification

Flexible taxonomic classification against your preferred reference database.

SILVA Greengenes2 GTDB
Technology Highlights

Built for microbiome science

From ASV inference through functional prediction — a complete toolkit for community analysis.

DADA2 ASV Inference

High-resolution ASV calling to resolve closely related taxa, without OTU clustering artefacts.

Multi-Database Taxonomy

Classification against SILVA, Greengenes2 or GTDB using DADA2 and/or QIIME2-compatible classifiers.

Alpha & Beta Diversity

Shannon, Chao1, Faith's PD within samples; Bray-Curtis and UniFrac between samples.

Differential Abundance

Identify taxa that differ significantly across groups using DESeq2, ANCOM or LEfSe.

PICRUSt2 Prediction

Predict functional gene family and pathway abundances from 16S community profiles.

Reproducible Reporting

Structured ASV tables, taxonomy summaries, diversity metrics and project reports.

Learning Path

From first principles to expert-level analysis

New to microbiome sequencing? Start with Foundations. Planning a study or reviewing results? Jump to Practitioner or Expert. Each level builds on the one before it, so you can read this page as a short course or dip in where you need to.

The big idea in plain language: every bacterial species carries a gene called 16S rRNA that works like a barcode. Reading that barcode from a sample tells you which bacteria are present and in what proportions.

LEVEL 1

A microbiome is a community

Your gut, a spoonful of soil or a river sample each hold thousands of microbial species living together. Their mix influences health, crop growth and ecosystems.

LEVEL 1

The 16S gene is a barcode

All bacteria and archaea share the 16S rRNA gene, but parts of it differ between species. Those differences let us tell groups of microbes apart.

LEVEL 1

PCR picks out one stretch

Short primers match the stable parts either side of the V3–V4 variable region. PCR copies only that stretch (about 460 bp) from every microbe in the sample at once.

LEVEL 1

Reads become variants, then names

The sequencer reads millions of copies. Software removes errors, groups identical sequences into amplicon sequence variants (ASVs), and matches each to a reference database to give it a name.

LEVEL 1

Diversity in two questions

Alpha diversity asks how rich and even one sample is. Beta diversity asks how different two samples are from each other.

LEVEL 1

Which microbes differ?

Differential abundance testing compares groups, such as treated versus control, and lists the taxa that are consistently more or less abundant.

For scientists planning a project or reading a report: how design choices and quality control decide whether the final community profile can be trusted.

LEVEL 2

ASVs beat 97% OTUs

Older pipelines clustered reads at 97% identity, lumping close relatives together. DADA2 models the error profile of each run and keeps exact sequences, so results can be compared across studies.

LEVEL 2

Truncation and overlap matter

Quality falls along each read, especially read 2. Reads are trimmed where quality drops, but must still overlap enough (typically 12+ bases) to merge into the full V3–V4 amplicon.

LEVEL 2

Depth and rarefaction

A rarefaction curve shows whether more sequencing would still reveal new ASVs. A curve that flattens means the sample is reasonably well covered.

LEVEL 2

Pick the database for your question

SILVA is widely used and richly curated, Greengenes2 links 16S to a genome-based tree, and GTDB follows a genome-based taxonomy. The names differ between them, so state which one you used.

LEVEL 2

PERMANOVA needs a second look

A significant PERMANOVA can reflect a true shift in the centre of a group or just a difference in spread. Pair it with a dispersion test and an ordination plot.

LEVEL 2

Controls and replicates

Extraction blanks, a mock community and enough biological replicates tell you how much of what you see is signal and how much is contamination or batch effect.

For method developers and reviewers: the statistical and biological caveats that decide whether an apparent result holds up.

LEVEL 3

Data are compositional

Sequencing returns proportions, not absolute counts. If one taxon blooms, all others look lower. Methods such as ANCOM-BC, ALDEx2 and CLR-based models are designed for this, though none is perfect.

LEVEL 3

No single best DA method

A large benchmark found that no differential abundance method was at once robust, powerful and flexible, so the best practice is to test sensitivity across more than one approach.

LEVEL 3

Copy number and PCR bias

Genomes carry anywhere from 1 to about 15 copies of the 16S gene, and primers amplify taxa unevenly. Both distort relative abundance away from true cell counts.

LEVEL 3

Phylogeny-aware distances

UniFrac uses a tree to weigh how related two communities are. A reference tree built from genomes, as in Greengenes2, can also align 16S and shotgun data in one framework.

LEVEL 3

Function is inferred, not measured

PICRUSt2 predicts gene content from taxonomy. Independent tests found such tools often lack the sensitivity to detect health-related functional shifts, so treat output as hypotheses.

LEVEL 3

Resolution has a ceiling

A roughly 460 bp amplicon often resolves genus well but cannot always separate species. Full-length 16S or shotgun metagenomics extends resolution when the question demands it.

Amplicon
A DNA stretch copied many times by PCR so it can be sequenced.
ASV
Amplicon sequence variant: an exact, error-corrected sequence, resolving down to a single nucleotide.
OTU
Operational taxonomic unit: a cluster of similar sequences, commonly at 97% identity.
Phred Q30
A base-call quality score meaning about 1 error in 1,000 bases.
Chimera
An artefactual sequence stitched from two parents during PCR, removed during denoising.
Alpha diversity
Richness and evenness within one sample (Observed ASVs, Chao1, Shannon, Faith's PD).
Beta diversity
Dissimilarity between samples (Bray-Curtis, UniFrac) shown by PCoA or NMDS.
CLR
Centred log-ratio: a transform that handles proportions by comparing each taxon with the sample's average.
The Science

How a barcode gene becomes a community profile

These schematics show what happens to a sample, from the gene we target to the way modern denoising separates real biological sequences from sequencing errors.

Figure 1 The 16S rRNA gene and the V3–V4 target schematic
V1V2V3V4V5V6V7V8V9 V3–V4 amplicon, about 460 bp forward primer (position ~341) reverse primer (position ~805) 5′ end3′ end · gene length about 1,500 bp conserved: primers bind herehypervariable: tells species apartregions covered by this service
Figure 1 | Conserved anchors, variable signatures. Region boundaries are approximate positions on the E. coli gene. The conserved stretches let one primer pair work across most bacteria, while the variable regions carry the information used to tell taxa apart. Exact primers and positions are confirmed per project.
Figure 2 From sample to sequence infographic
STEP 01

DNA extraction

Cells are broken open and DNA is purified from faeces, soil, water, swabs or tissue.

gDNA
blank control
STEP 02

Amplify V3–V4

PCR with primers flanking the target region copies the barcode from every microbe at once.

~460 bp amplicon
band size · yield
STEP 03

Index and pool

A second short PCR adds a unique index per sample, so many samples can share one run.

indexed library
library QC
STEP 04

Paired-end sequencing

Both ends of each amplicon are read, giving two overlapping reads that are later merged.

FASTQ
Q30 · cluster density
Figure 2 | Typical wet-lab path. Platform, read length and kits are agreed for each project. Projects that supply pre-generated FASTQ files enter the workflow at the data QC step.
Figure 3 Why ASVs resolve more than OTUs schematic
97% OTU clustering DADA2 ASV inference one unit: three close relatives and some errors merged together three exact sequences kept; grey dots = sequencing errors removed
Figure 3 | Lumping versus resolving. OTU clustering groups sequences by a fixed identity threshold. DADA2 learns the error rates of the run and keeps only sequences that are more abundant than errors can explain, resolving differences down to a single nucleotide (Callahan et al., 2016, 2017).
Figure 4 Why one sample is never the whole story schematic
1 gram of faeces or soilbillions of cells · thousands of taxa
~1,500 bpfull 16S gene
~460 bpV3–V4 amplicon
1–1516S copies per genome
Figure 4 | Sampling a crowded community. Each green mark stands for one sequence drawn from a much larger pool. Biological replicates, adequate depth and negative controls keep a small sample of reads from being over-interpreted. Copy-number range from rrnDB (Stoddard et al., 2015).
Analysis Workflow

From raw reads to community profiles

Each sample passes through a quality-controlled analysis workflow designed to move from raw sequencing reads to taxonomically classified, diversity-analysed microbiome profiles.

Figure 5 Quality-gated amplicon workflow infographic
STAGE 01

Raw data QC

Per-base quality, adapter and primer content are reviewed for every sample.

FASTQQC report
Q30 · read counts
STAGE 02

Primer removal & filtering

Primers are clipped, then reads are truncated and filtered on expected errors and ambiguous bases.

filtered FASTQ
reads retained
STAGE 03

Error learning & denoising

DADA2 learns the run's error model and infers exact sequence variants for each sample.

denoised ASVs
error-model fit
STAGE 04

Merge & chimera removal

Paired reads are merged into the full amplicon and PCR chimeras are detected and dropped.

ASV table
merge rate · chimera rate
STAGE 05

Taxonomy

ASVs are classified against SILVA, Greengenes2 or GTDB with a DADA2 or QIIME2-compatible classifier.

SILVA · GG2GTDB
taxonomy table
classified to genus
STAGE 06

Diversity & ordination

Alpha metrics, Bray-Curtis and UniFrac distances, PCoA or NMDS, and PERMANOVA.

metricsdistance matrices
depth · rarefaction
STAGE 07

Differential abundance & function

DESeq2, ANCOM-BC or LEfSe with FDR control, plus PICRUSt2 where it fits the design.

DA tablespathways
FDR-adjusted
STAGE 08

Reporting & delivery

Phyloseq-based figures and a project report, shared through secure storage.

reportshared drive
access control
Figure 5 | Decision points. Amber gates mark where metrics are reviewed before samples move on. Samples with unusually low read counts or merge rates are flagged to you rather than silently carried forward.
01

Data QC & Pre-processing

Quality assessment, primer removal, quality filtering and trimming prior to denoising.

02

DADA2 Denoising

Paired-end reads denoised into ASVs; merging, chimera removal and filtering.

03

Taxonomic Classification

ASVs classified against SILVA / Greengenes2 / GTDB to genus level and above.

04

Alpha Diversity

Observed ASVs, Chao1, Shannon index and Faith's PD within-sample metrics.

05

Beta Diversity & Ordination — Bray-Curtis / UniFrac distances, PCoA/NMDS, PERMANOVA.

06

Differential Abundance — DESeq2, ANCOM-BC or LEfSe across sample groups.

07

Functional Prediction & Reporting — PICRUSt2 (where applicable); QC, taxonomy, diversity & DE reports.

Quality Control

Trust the data before you trust the biology

Every downstream result depends on clean reads. These interactive plots use simulated data to show what good and poor quality look like, and why denoising matters.

Figure 6 Quality profile of paired reads simulated
Figure 6 | Read 2 degrades first. Mean Phred quality along 2 × 300 bp reads. Truncation points are chosen where quality drops (dashed lines), keeping enough overlap to merge the pair into the full amplicon.
Figure 7 Learned sequencing error model simulated
Figure 7 | DADA2 learns errors from your own run. Points show observed error frequency at each quality score and the line shows the fitted model. A good fit supports confident separation of real variants from noise (Callahan et al., 2016).
Figure 8 Read tracking through the pipeline simulated
Input readsafter demultiplexing
100%
Filtered & trimmedexpected-error filter
92%
DenoisedDADA2 error correction
90.5%
Merged pairsfull V3–V4 amplicon
85.3%
Non-chimericfinal ASV table
80.1%
Figure 8 | Where reads are lost. Illustrative retention for one typical sample. Each step has a purpose: removing low-quality reads, correcting errors, joining pairs and discarding PCR artefacts. Sudden drops at the merge step usually point to over-trimming.
Figure 9 Rarefaction curves simulated
Figure 9 | Is sequencing deep enough? Observed ASVs as reads are subsampled. Curves that plateau indicate most of the community has been captured; curves still rising suggest a rich sample or insufficient depth.
Diversity Analysis

Within samples and between them

Alpha diversity summarises one community. Beta diversity compares communities and is the basis of most group comparisons. All plots below are simulated to illustrate the outputs you receive.

Figure 10 Alpha diversity by group simulated
Figure 10 | Richness and evenness. Shannon index for three simulated groups. Boxes show the median and interquartile range, dots are samples. Groups are compared with non-parametric tests, with multiple-testing correction when several metrics are reported.
Figure 11 Beta diversity ordination (PCoA) simulated
Figure 11 | Community structure. Each point is a sample; closer points have more similar communities. Ellipses show 95% confidence regions per group. PERMANOVA tests whether the group centroids differ and is reported with a dispersion check (Anderson, 2001).
Figure 12 Taxonomic composition simulated
Figure 12 | Relative abundance of top genera. Stacked bars for simulated samples in three groups, with rarer taxa pooled as "Other". Bars show proportions, so they describe the make-up of each community, not absolute cell numbers.
Figure 13 Why relative abundance can mislead interactive · simulated
Figure 13 | The compositional trap. Only taxon A changes in the real community (left). Because sequencing reports proportions, every other taxon seems to fall (right). Compositionally aware methods and, where possible, absolute-abundance anchors such as qPCR or spike-ins help avoid false conclusions (Gloor et al., 2017).
Differential Abundance

Finding the taxa that really differ

Tests ask which taxa are more or less abundant between groups. Effect size and statistical confidence should be read together, with false discovery rate control across all taxa tested.

Figure 14 Volcano plot of differential taxa simulated
Figure 14 | Effect size versus confidence. Each point is a genus. Green and orange points pass both the FDR (q < 0.05) and log2 fold-change thresholds; grey points do not. Labels mark the strongest hits.
Table 1 Choosing a differential abundance method
MethodIdeaGood for
DESeq2Negative-binomial model on countsSensitive; widely used; watch composition effects
ANCOM-BCBias-corrected log-ratio modelCompositional data; good false-positive control
ALDEx2CLR on Monte Carlo drawsSmall samples; conservative
LEfSeRank tests plus LDA effect sizeExploratory biomarker discovery
No method wins everywhere. A 2022 benchmark in Microbiome found that compositional methods improved false-positive control but often lost power, and none was robust, powerful and flexible at once. Agreement across methods is the strongest signal.
Deliverables

Structured outputs for microbiome research

At project completion, clients receive a structured set of sequencing, analysis and reporting outputs via secure shared storage.

Raw & Processed Data

FASTQ files — raw and appropriately trimmed sequence data.

QC Reports

Sample-level sequencing and preprocessing QC summaries.

ASV Table

Denoised and filtered ASV abundance matrix with per-sample read counts.

Taxonomy Table

Taxonomic classification of ASVs from phylum to genus level.

Diversity Results

Alpha diversity metrics and beta diversity distance matrices with ordination outputs.

Differential Abundance

Tables of significantly abundant taxa across comparisons.

Functional Predictions

PICRUSt2 gene family and pathway abundance tables, where included in scope.

Summary Report

QC, rarefaction curves, diversity metrics, taxonomy summaries and community comparisons.

Secure Delivery

Project files via secure shared storage or cloud-based access.

Sample Requirements

What we accept

Flexible input options for sequencing-plus-analysis or analysis-only projects.

Broad Matrix Support

We accept DNA extracted from gut contents, faecal material, soil, water, plant-associated material, swabs, tissue and other biological or environmental matrices, as well as pre-generated sequencing data where analysis-only services are required.

Project-specific requirements for DNA quantity, concentration, purity, sequencing depth, replicate structure and sample grouping are finalised according to the study design.

Contact us for a project-specific sample submission checklist and recommended sequencing depth.

Gut · Faecal · Soil · Water · Plant
Swabs · Tissue · or pre-generated FASTQ

Case Studies & Performance

Proven across sample types

Production workflows applied to gut microbiome and soil/rhizosphere community studies.

Gut Microbiome Profiling

Human / Animal · Multi-group · V3–V4

Mean raw-data Q30 >90%
Chimera removal DADA2 filters applied
Taxonomy database SILVA
Alpha & beta diversity Full suite
Differential abundance FDR-corrected

Multi-group 16S V3–V4 study processed end-to-end: QC, DADA2 denoising, SILVA taxonomy, alpha/beta diversity and differential abundance testing across study groups.

Soil / Rhizosphere Microbiome

Land-use · Plant compartments · PICRUSt2

Sample types Soil & rhizosphere
Community structure Beta diversity + PERMANOVA
Ordination PCoA / NMDS
Functional prediction PICRUSt2
Workflow Same validated pipeline

Community structure differences across land-use types or plant-associated compartments characterised using beta diversity ordination and PERMANOVA. Functional potential predicted using PICRUSt2.

The Frontier

What 16S can tell you, and where the field is heading

Amplicon sequencing remains the most cost-effective way to profile many samples. Here is how it compares with longer reads and shotgun sequencing, and what recent research says about its limits.

Figure 15 Short-read V3–V4 versus full-length 16S published data
Genus · V3–V4 Illumina
94.8%
Genus · full-length PacBio
95.1%
Species · V3–V4 Illumina
55.2%
Species · full-length PacBio
74.1%
Genus-level results were nearly identical. Longer reads mainly improved species-level assignment, while samples still clustered by body site rather than platform.
Figure 15 | Share of reads assigned. Data from saliva, plaque and faecal samples of 9 volunteers (Buetas et al., BMC Genomics, 2024). Results depend on database and sample type.
Table 2 Choosing the right approach
ApproachStrengthsLimits
16S V3–V4Low cost per sample, large cohorts, mature toolsGenus-level focus; PCR and copy-number bias; function only inferred
Full-length 16SBetter species-level assignmentHigher cost per read; fewer reads per run
Shotgun metagenomicsStrain and gene content; includes fungi and virusesHigher cost; heavy host DNA in some samples
Pick by question. For comparing community structure across many samples, V3–V4 is usually efficient. If you need species or gene-level answers, ask us about extended designs.
REFERENCE TREES

Greengenes2

A single genome-based tree combining 16S and genomic data. In the original study, 16S and shotgun data from the same samples agreed in ordination, taxonomy and effect size when placed on this tree (McDonald et al., Nature Biotechnology, 2024).

COMPOSITION

Beyond relative abundance

Spike-in standards, qPCR or flow cytometry can anchor proportions to absolute cell counts, which changes some conclusions drawn from relative data alone.

FUNCTION

Prediction has limits

An independent evaluation of PICRUSt2 and three similar tools on simulated and paired 16S and metagenome data found they often lacked the sensitivity to detect health-related functional changes (Matchado et al., Microbial Genomics, 2024).

CONTAMINATION

Low-biomass samples

Reagent and environmental contaminants matter most when little DNA is present. Extraction blanks and statistical decontamination tools help flag them (Davis et al., Microbiome, 2018).

STATISTICS

Cross-checking methods

Because differential abundance methods can disagree, reporting taxa supported by more than one approach is a practical safeguard (Yang and Chen, Microbiome, 2022).

STANDARDS

Mock communities

A defined mixture of known species run alongside samples shows how accurately a workflow recovers composition, and makes batches comparable.

Scope note. Nucleome's 16S service covers the workflow and deliverables described on this page, from data QC through diversity, differential abundance and optional PICRUSt2 prediction. The frontier methods shown here are provided to place the service in its scientific context; contact our team to discuss extended or project-specific analyses.
Why Nucleome

Service highlights

Support for human health, agricultural, environmental and ecological microbiome projects.

End-to-End 16S Analysis

From raw sequencing data through QC, ASV inference, taxonomy, diversity and differential abundance.

DADA2 High-Resolution

ASV-level resolution avoids OTU clustering artefacts and provides finer taxonomic discrimination.

Multi-Database Classification

Flexible taxonomic classification against SILVA, Greengenes2 or GTDB.

Comprehensive Diversity

Alpha and beta diversity, ordination and statistical testing in an integrated workflow.

Functional Prediction

PICRUSt2-based metabolic pathway prediction for community functional interpretation.

Broad Sample Support

Analysis for gut, soil, water, plant-associated and other complex microbiome samples.

Questions

Frequently asked, in plain language

From the basics to the finer points of study design and statistics.

basicWhat is 16S amplicon sequencing?
It reads a short stretch of the 16S rRNA gene, a barcode shared by all bacteria, to find out which microbes are in a sample and in what proportions. It does not read whole genomes.
basicWhich samples can I send?
DNA from gut contents, faeces, soil, water, plant-associated material, swabs, tissue and other matrices. You can also send existing FASTQ files for analysis only.
basicWhat will I receive?
FASTQ files, QC reports, an ASV table, a taxonomy table, diversity results, differential abundance tables, optional PICRUSt2 predictions and a summary report with figures, through secure shared storage.
intermediateWhy ASVs instead of OTUs?
ASVs are exact, error-corrected sequences, so they are reproducible across studies and can separate close relatives that 97% clustering merges. OTUs depend on the dataset they were clustered from.
intermediateHow many reads per sample do I need?
It depends on how diverse the sample type is and what you want to detect. Rarefaction curves show whether depth is enough, and our team agrees sequencing depth and replicate numbers with you for each study design.
intermediateWhich database should I choose?
SILVA is a common default. Greengenes2 is useful when you want phylogenetic methods or to compare with shotgun data, and GTDB follows a genome-based taxonomy. Names can differ between them, so keep the database consistent within a study.
advancedCan 16S tell me what the microbes are doing?
Only indirectly. PICRUSt2 predicts gene content from taxonomy, and an independent evaluation found that such tools often lack sensitivity to detect health-related functional changes. Use predictions to generate hypotheses and confirm with shotgun metagenomics or metabolomics.
advancedWhy do different differential abundance tools give different answers?
They make different assumptions about zeros and about compositional data. A 2022 benchmark found no single method to be best in every setting, so we recommend confirming key taxa with more than one method.
advancedCan I measure absolute bacterial numbers?
Standard 16S gives relative abundance. Absolute estimates need an extra anchor such as qPCR, flow cytometry or spike-in standards, which can be considered at the study design stage.
References

Key literature & resources

Selected sources behind the methods and reference values on this page. Figures labelled "simulated", "schematic" or "illustrative" are generated for this page, not client data.

  1. Callahan BJ, et al. DADA2: high-resolution sample inference from Illumina amplicon data. Nature Methods 2016;13:581–583.
  2. Callahan BJ, McMurdie PJ, Holmes SP. Exact sequence variants should replace operational taxonomic units in marker-gene data analysis. ISME Journal 2017;11:2639–2643.
  3. Bolyen E, et al. Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nature Biotechnology 2019;37:852–857.
  4. Klindworth A, et al. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Research 2013;41:e1.
  5. Quast C, et al. The SILVA ribosomal RNA gene database project. Nucleic Acids Research 2013;41:D590–D596.
  6. Parks DH, et al. GTDB: an ongoing census of bacterial and archaeal diversity through a phylogenetically consistent, rank normalized and complete genome-based taxonomy. Nucleic Acids Research 2022;50:D785–D794.
  7. McDonald D, et al. Greengenes2 unifies microbial data in a single reference tree. Nature Biotechnology 2024;42:715–718. Record
  8. Stoddard SF, et al. rrnDB: improved tools for interpreting rRNA gene abundance in bacteria and archaea and a new foundation for future development. Nucleic Acids Research 2015;43:D593–D598.
  9. McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS ONE 2013;8:e61217.
  10. Lozupone C, Knight R. UniFrac: a new phylogenetic method for comparing microbial communities. Applied and Environmental Microbiology 2005;71:8228–8235.
  11. Anderson MJ. A new method for non-parametric multivariate analysis of variance. Austral Ecology 2001;26:32–46.
  12. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology 2014;15:550.
  13. Lin H, Peddada SD. Analysis of compositions of microbiomes with bias correction. Nature Communications 2020;11:3514.
  14. Fernandes AD, et al. Unifying the analysis of high-throughput sequencing datasets: characterizing RNA-seq, 16S rRNA gene sequencing and selective growth experiments by compositional data analysis. Microbiome 2014;2:15.
  15. Segata N, et al. Metagenomic biomarker discovery and explanation. Genome Biology 2011;12:R60.
  16. Gloor GB, et al. Microbiome datasets are compositional: and this is not optional. Frontiers in Microbiology 2017;8:2224.
  17. Yang L, Chen J. A comprehensive evaluation of microbial differential abundance analysis methods: current status and potential solutions. Microbiome 2022;10. Record
  18. Douglas GM, et al. PICRUSt2 for prediction of metagenome functions. Nature Biotechnology 2020;38:685–688.
  19. Matchado MS, et al. On the limits of 16S rRNA gene-based metagenome prediction and functional profiling. Microbial Genomics 2024;10(2). Record
  20. Buetas E, et al. Full-length 16S rRNA gene sequencing by PacBio improves taxonomic resolution in human microbiome samples. BMC Genomics 2024;25:310. Link
  21. Davis NM, et al. Simple statistical identification and removal of contaminant sequences in marker-gene and metagenomics data. Microbiome 2018;6:226.
  22. Salter SJ, et al. Reagent and laboratory contamination can critically impact sequence-based microbiome analyses. BMC Biology 2014;12:87.
  23. Vandeputte D, et al. Quantitative microbiome profiling links gut community variation to microbial load. Nature 2017;551:507–511.