16S Profiling Pipeline Documentation

Overview

This Nextflow pipeline performs taxonomic profiling of paired-end 16S rRNA amplicon data (V3V4 region). The pipeline integrates quality control, read preprocessing, and ASV inference and taxonomic assignment using multiple complementary tools (ASV inference: DADA2, DEBLUR and USEARCH-UNOISE; taxonomic assignment: QIIME BLAST, QIIME NAIVE BAYES, IDTAXA, ASSIGNTAXONOMY, SINTAX) to provide robust microbial community characterization.

Pipeline Version: 1.0.0
Author: Petra Polakovicova
Nextflow Version Required: ≥22.10.0

---

Testing Instructions

Instructions for testing the pipeline's reproducibility are available as Markdown, as well as in PDF version.

Workflow Description

The pipeline implements the following workflow:

flowchart TB
    INPUT["Raw FASTQ files"] --> FASTQC_RAW["FASTQC RAW"] & CUTADAPT["CUTADAPT"]
    CUTADAPT --> FASTQC_TRIMMED["FASTQC TRIMMED"] & DADA2["DADA2"] & MERGE["MERGING READS"]
    DADA2 --> NB_DADA2["QIIME Naive Bayes DADA2"] & BLAST_DADA2["QIIME BLAST DADA2"] & IDTAXA_DADA2["IDTAXA DADA2"] & ASSIGN_DADA2["assignTaxonomy DADA2"]
    MERGE --> QIIME_IMPORT["QIIME IMPORT"]
    QIIME_IMPORT --> DEBLUR["QIIME DEBLUR"]
    DEBLUR --> NB_DEBLUR["QIIME Naive Bayes Deblur"] & BLAST_DEBLUR["QIIME BLAST Deblur"] & IDTAXA_DEBLUR["IDTAXA Deblur"] & ASSIGN_DEBLUR["assignTaxonomy Deblur"]
    NB_DEBLUR --> METASTANDARD_16S["MetaStandard 16S"]
    BLAST_DEBLUR --> METASTANDARD_16S
    IDTAXA_DEBLUR --> METASTANDARD_16S
    ASSIGN_DEBLUR --> METASTANDARD_16S
    NB_DADA2 --> METASTANDARD_16S
    BLAST_DADA2 --> METASTANDARD_16S
    IDTAXA_DADA2 --> METASTANDARD_16S
    ASSIGN_DADA2 --> METASTANDARD_16S
    FASTQC_RAW --> MULTIQC["MULTIQC REPORT"]
    FASTQC_TRIMMED --> MULTIQC

Loading

Workflow Steps

1. Quality Control (Raw Reads)

   • Tool: FastQC v0.12.1
   • Assesses raw read quality metrics
   • Output: HTML reports and quality metrics

2. Read Preprocessing

   • Tool: Cutadapt v5.2
   • Performs primer & adapter trimming

3. Quality Control (Trimmed Reads)

   • Tool: FastQC v.0.12.1
   • Validates reads after preprocessing (primers & adapters should be gone)
   • Output: HTML reports and quality metrics

4. ASV inference - DADA2

   • Tool: DADA2 v1.38.0
   • Performs ASV inference using divisive algorithm
   • Provides raw counts of each ASV in dataset
   • Runs in parallel with DEBLUR & UNOISE

5. MERGING READS - BBMERGE

   • Tool: BBMAP-BBMERGE vXY
   • Merges paired-end reads to create single-end reads
   • Needed for downstream ASV inference using DEBLUR and UNOISE

6. ASV inference - DEBLUR

   • Tool: Deblur plugin inside QIIME2 v2026.1
   • Performs quality filtering, length trimming, positive filtering using SORTMERNA & ASV inference using deblur algorithm
   • Provides raw counts of each ASV in dataset
   • Runs in parallel with DADA2 & UNOISE

7. ASV inference - UNOISE   - TO DO

8. Taxonomic assignment - QIIME naive-bayes classifier

   • Tool: qiime feature-classifier classify-sklearn inside QIIME2 v2026.1
   • Performs taxonomic classification using pre-built classifier, which was made using REscript plugin (see extract_silva.sh)
   • This classifier is run on all three ASV-inference approaches (DADA2, DEBLUR, UNOISE)
   • Runs in parallel with four different taxonomic classifiers

9. Taxonomic assignment - QIIME BLAST classifier

   • Tool: classify-consensus-blast inside QIIME2 v2026.1
   • Performs taxonomic classification using pre-built classifier, which was made using REscript plugin (see extract_silva.sh)
   • This classifier is run on all three ASV-inference approaches (DADA2, DEBLUR, UNOISE)
   • Runs in parallel with four different taxonomic classifiers

10. Taxonomic assignment - IDTAXA

    • Tool: R function DECIPHER::IdTaxa() v3.6.0
    • Performs taxonomic classification using pre-built classifier, which was made using QIIME2's REscript plugin (see extract_silva.sh) and train_idtaxa.R script.
    • This classifier is run on all three ASV-inference approaches (DADA2, DEBLUR, UNOISE)
    • Runs in parallel with four different taxonomic classifiers

11. Taxonomic assignment - AssignTaxonomy

    • Tool: R function dada2::assignTaxonomy() v1.38.0
    • Performs taxonomic classification using pre-built classifier, which was made using QIIME2's REscript plugin (see extract_silva.sh).
    • This classifier is run on all three ASV-inference approaches (DADA2, DEBLUR, UNOISE)
    • Runs in parallel with four different taxonomic classifiers

12. Quality Report Aggregation

    • Tool: MultiQC v1.21
    • Aggregates QC metrics from FastQC and Fastp
    • Generates comprehensive HTML report

---

Tools and Versions

Tool	Version	Container Source	Purpose
FastQC	0.12.1	quay.io/biocontainers/fastqc:0.12.1--hdfd78af_0	Quality control assessment of raw and trimmed reads
Cutadapt	5.2	quay.io/biocontainers/cutadapt:5.2--py310hdfd78af_0	Primer and adapter trimming
DADA2	1.38.0	quay.io-biocontainers-bioconductor-dada2:1.38.0--r45ha27e39d_0	ASV inference using divisive algorithm
VSEARCH-fastq_mergepairs	2.30.5	quay.io-biocontainers-vsearch:2.30.5--h0bb26bb_0	Merges paired-end reads into single-end reads
QIIME2 Deblur plugin	2026.1	qiime2/core:2026.1	Quality filtering and ASV inference via Deblur
UNOISE	2.30.5	quay.io-biocontainers-vsearch:2.30.5--h0bb26bb_0	ASV inference using UNOISE3
QIIME naive-bayes classifier	2026.1	quay.io-qiime2-amplicon:2026.1	Taxonomic assignment using pre-trained sklearn classifier
QIIME BLAST classifier	2026.1	quay.io-qiime2-amplicon:2026.1	Taxonomic assignment using BLAST classifier
IDTAXA (DECIPHER)	3.6.0	quay.io-biocontainers-bioconductor-decipher-3.6.0--r45h01b2380_0	Taxonomic assignment using DECIPHER IdTaxa function
AssignTaxonomy (DADA2)	1.38.0	quay.io-biocontainers-bioconductor-dada2:1.38.0--r45ha27e39d_0	Taxonomic assignment using DADA2 assignTaxonomy function
MultiQC	1.21	quay.io/biocontainers/multiqc:1.21--pyhdfd78af_0	Aggregates QC metrics into a single report

Container Technology

The pipeline uses Singularity containers for reproducibility and portability. All containers are automatically pulled and cached in the directory specified by singularity_cache_dir.

---

Installation

Prerequisites

The pipeline uses Nextflow ≥22.10.0 and Singularity which is assumed to be pre-installed by the user. Users will need at least 8 GB of RAM, 5 GB of storage for downloading the classifiers and Singularity images, and additional storage for running the pipeline (this can vary depending on how many samples you have).

Quick Setup (Recommended)

The pipeline includes an automated setup script that downloads all required databases and containers.

Run the setup script:

# Interactive mode (will prompt for installation directory)
./setup_pipeline.sh

# Or specify directory directly
./setup_pipeline.sh /shared/pipeline_resources

What the script does:

1. Creates directory structure:

   • classifiers/ - pre-built taxonomic classifiers (~500MB)
   • singularity_cache/ - Container images (~5 GB)
   • logs/ - Setup logs

2. Downloads pre-builts taxonomic classifiers from Zenodo

3. Pulls all required Singularity containers

4. Generates a configuration file (pipeline_paths.config) with all paths

After setup completes:

# Run the pipeline - database paths are already configured!
nextflow run main.nf \
  --input samples.csv \
  --outdir results

# Or use the generated config for explicit paths
nextflow run main.nf \
  -c /path/to/pipeline_paths.config \
  --input samples.csv \
  --outdir results

Note

After running the setup script, the default database paths in nextflow.config point to ./classifiers. If you installed to a different location, either use the generated pipeline_paths.config file or specify paths via command-line parameters.

Input Requirements

Samplesheet Format

The pipeline requires a CSV samplesheet with the following columns:

Column	Description
sample	Unique sample identifier
read1	Absolute or relative path to R1 FASTQ file (can be gzipped)
read2	Absolute or relative path to R2 FASTQ file (can be gzipped)

Example samplesheet (samplesheet.csv):

sample,read1,read2
sample1,/path/to/sample1_R1.fastq.gz,/path/to/sample1_R2.fastq.gz
sample2,/path/to/sample2_R1.fastq.gz,/path/to/sample2_R2.fastq.gz
sample3,/path/to/sample3_R1.fastq.gz,/path/to/sample3_R2.fastq.gz

Note

• Sample names must be unique
• File paths can be absolute or relative to the working directory
• FASTQ files can be compressed (.gz) or uncompressed

---

Running the Pipeline

To solve potential issue of the singularity tmp directory, please export the SINGULARITY_TMPDIR in your run script

export SINGULARITY_TMPDIR=/path/to/directory/with/space

Basic Usage

If you used the setup script (databases in default locations):

nextflow run main.nf \
  --input samplesheet.csv \
  --outdir results

If classifiers are in custom location:

nextflow run main.nf \
  --input samplesheet.csv \
  --outdir results \
  --classifiers /path/to/classifiers_dir

Output Descriptions

Directory	Contents	Description
01_fastqc_raw/	FastQC HTML & ZIP reports	Quality metrics for raw reads
01_fastqc_trimmed/	FastQC HTML & ZIP reports	Quality metrics for trimmed reads (after Cutadapt)
02_cutadapt/	Trimmed FASTQ files	Reads after adapter and primer removal
03_dada2/	ASV tables & FASTA sequences	DADA2 ASV inference outputs
04_merged/	Merged FASTQ reads	Paired-end reads merged for Deblur/UNOISE
05_deblur/	ASV tables & FASTA sequences	Deblur ASV inference outputs
06_unoise/	ASV tables & FASTA sequences	UNOISE ASV inference outputs
07a_qiime_naive_bayes/	Taxa tables (DADA2 & Deblur)	Taxonomic assignment using QIIME naive-bayes classifier
07b_qiime_blast/	Taxa tables (DADA2 & Deblur)	Taxonomic assignment using QIIME BLAST classifier
07c_assigntaxonomy/	Taxa tables (DADA2 & Deblur)	Taxonomic assignment using DADA2 assignTaxonomy
07d_idtaxa/	Taxa tables (DADA2 & Deblur)	Taxonomic assignment using DECIPHER IdTaxa
08_metastandard/	Standardized outputs of all executed denoising + taxonomic tools
09_multiqc/	MultiQC reports & JSON/logs	Aggregated QC metrics from all tools and samples
pipeline_info/	HTML, DAG, trace, timeline	Pipeline execution reports and workflow DAG

---

Citations

If you use this pipeline, please cite:

• Nextflow: Di Tommaso, P., et al. (2017). Nextflow enables reproducible computational workflows. Nature Biotechnology, 35(4), 316-319.
• FastQC: Andrews, S. (2010). FastQC: A Quality Control Tool for High Throughput Sequence Data.
• MultiQC: Ewels, P., et al. (2016). MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics, 32(19), 3047-3048.
• DADA2: Callahan, et al. (2016). DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods 13, 581–583.
• DEBLUR: Amir, A., et al. (2017). Deblur Rapidly Resolves Single-Nucleotide Community Sequence Patterns. mSystems, 2(2), e00191-16.
• QIIME2: Bolyen E, et al.(2019). Reproducible, interactive, scalable and extensible microbiome data science using QIIME 2. Nature Biotechnology 37: 852–857.
• DECIPHER: ES Wright (2024). Fast and Flexible Search for Homologous Biological Sequences with DECIPHER v3."The R Journal, 16(2), 191-200.