Tool assembly: Metabarcoding

What is the metabarcoding tool assembly?

The metabarcoding tool assembly gives a detailed overview of metabarcoding data management and complements the Biodiversity domain page. It aims to provide solutions to the challenges encountered in the life cycle of metabarcoding data. This page considers the following main data types:

• Sample metadata
• Sequencing reads (raw data)
• Curated occurrence data
• Curated sequence and taxa tables

Metabarcoding sequencing reads are processed into two major types of occurrence tables. Operational Taxonomic Units (OTUs) are generated by clustering reads based on sequence similarity [1], while Amplicon Sequence Variants (ASVs) are resolved at the single-nucleotide level without any clustering step, enabling finer taxonomic resolution and greater comparability across datasets [2]. The coexistence of these two output types is one of the major challenges in managing metabarcoding data.

The main challenges that are specific to metabarcoding are:

• the lack of a universal, community-accepted standard for ASV/OTU tables and of a repository dedicated to them (several institutions are developing solutions, which are mentioned on this page);
• the identification bias of reference databases, which is inherent to the fact that only what is already known can be identified.

How to access the metabarcoding tool assembly?

All the tools presented in this tool assembly are free, publicly available and mostly open source. Often registration is required to submit data.

What can you use the metabarcoding tool assembly for?

This tool assembly provides guidance and tool recommendations for each stage of the metabarcoding data life cycle: Planning, Collecting, Processing and Analysing, Preserving, Sharing and Reusing. Each stage is described below Figure 1.

Data management planning

As described on the Biodiversity domain page, a Data Management Plan (DMP) centralises all the information about a study. You can write it using the Data Stewardship Wizard (DSW) or any other platform that follows the Research Data Alliance (RDA) DMP Common Standard, such as DMPRoadmap (for example DMP OPIDoR). An ELIXIR instance of Data Stewardship Wizard (DSW) is available for researchers here. In addition to the funder and institutional requirements, the DMP should follow the RDA structure and ideally be shared publicly alongside the dataset upon publication. For more information, see the RDMkit page on data management planning.

Collecting data and metadata

Before sample collection, plan which data and metadata have to be collected. Ask your contact person which permits you might need to obtain under the Nagoya Protocol (see also biological resource management and compliance on the Biodiversity page). You can find more information on the Convention on Biological Diversity page About the Nagoya Protocol.

A comprehensive reference for the standards applicable to environmental DNA (eDNA), including metabarcoding, is provided by Takahashi et al. [3], which underpins the FAIRe metadata checklist.

For the sampling itself, protocols.io provides a growing collection of eDNA sampling protocols for different types of biological material, which can be reused or adapted. Protocols should be assigned a DOI and referenced in the project metadata under associated_resource. For recording extensive experimental information, see also the LabID tool assembly.

All contamination mitigation steps (for example, bleach protocols or physical separation of clean and dirty spaces) must be recorded. All controls and blanks must be collected, assigned a unique identifier, and linked to their metadata so that they can be traced throughout the entire process, from field collection to bioinformatic output.

The standards that apply to metabarcoding sequences are the Genomic Standards Consortium (GSC) family of minimum information standards, more specifically the core standard Minimum Information about any (x) Sequence (MIxS). Specifications of the MIxS standard for metabarcoding data from different types of samples (for example, water or soil) are available as sample checklists in the European Nucleotide Archive (ENA). Among the checklists that you can find in the ENA checklist browser, checklists ERC000012 to ERC000027 are among the most used for biodiversity assessment studies. Raw data (sequencing reads) should be deposited in FASTQ format.

MIxS is incorporated into version 1.0 of the FAIRe checklist, which provides 337 terms (38 of them mandatory and 51 highly recommended) organised in components (projectMetadata, sampleMetadata, experimentRunMetadata, otuFinal and taxaFinal). You can validate your data against the checklist using FAIRe-fier, a web-based tool that checks controlled vocabularies, mandatory fields, and formatting constraints such as ISO 8601 dates.

Using a laboratory information management system or electronic lab notebook such as eLabFTW or LabID (Lab Integrated Data) to record this information from the outset reduces errors and facilitates submission to public repositories. FAIRe-ator can generate a pre-formatted data template based on your assay type, sample type and number of assays, ensuring correct file naming from the very beginning.

The gene marker used (target_gene) should be entered using the controlled vocabulary specified by FAIRe (28 standardised values). This prevents inconsistent variants such as CO1, COXI, COX1 or Cytochrome oxidase I from entering the metadata record.

The Darwin Core Archive (DwC-A) format, based on the Darwin Core (DwC) standard, is recommended for describing occurrence data. The GBIF Metabarcoding Data Toolkit (MDT), developed by the Global Biodiversity Information Facility (GBIF), enables formatting the data and publishing it through GBIF. The MDT user guide promotes the use of controlled vocabularies, which make your research inputs and outputs interoperable, and lists local instances of the tool. In the long term, the new Darwin Core Data Package (DwC-DP) publishing profiles should replace DwC-A, as they are more practical.

Persistent identifiers (PIDs) at the sample level (BioSamples accession) and at the sequence level (accessions number from the European Nucleotide Archive - ENA or the US based Short Read Archive - SRA ) are mandatory to ensure that the data is truly findable and interoperable.

Processing and analysing data

Choice of pipeline

The choice of pipeline depends strongly on the study. For biodiversity monitoring assessments, reusing and comparing against existing pipelines is recommended. If you are undertaking a methodologically innovative study, existing and new approaches can be combined. If no suitable pipeline exists, a tailored pipeline may be necessary.

For a detailed overview of available pipelines, see the review by Hakimzadeh et al. [4], which covers tools appropriate for water, sediment and eukaryote-only datasets. For an ASV-based metabarcoding analysis, nf-core/ampliseq is a well-documented, reproducible workflow that wraps several established tools. MGnify can also be used for comparative microbiome analyses, although it is better known for metagenomics analysis. Galaxy Ecology, as well as other Galaxy Labs, provides a web-based platform that allows you to export your full analysis history which can be published as a Research Object Crate (RO-Crate) to e.g. Zenodo. If you have created a new workflow you can make it publicly available (for example, via WorkflowHub), which is valuable for reproducibility (see also the Galaxy tool assembly). WorkflowHub allows publishing workflows with a DOI, and bio.tools helps you identify suitable tools for specific analysis needs.

Bioinformatics code and tailored pipelines should be published in a code repository such as GitHub, registered in WorkflowHub with a DOI, and documented under the term code_repo in the projectMetadata component of the FAIRe metadata checklist.

Reference database selection

The choice of reference database for taxonomic assignment is one of the most challenging decisions in a metabarcoding analysis, as it directly determines what can be identified and what remains unassigned.

Public databases include:

• Barcode of Life Data System (BOLD), covering plants, metazoans, fungi and other species
• The International Nucleotide Sequence Database Collaboration (INSDC) databases: National Center for Biotechnology Information (NCBI), ENA and DNA Data Bank of Japan (DDBJ)

Curated databases specific to gene markers include:

• SILVA for 16S/18S rRNA gene markers (Bacteria, Archaea and Eukarya)
• Protist Ribosomal Reference database (PR2) for eukaryotic 18S rRNA
• UNITE for fungal Internal transcribed spacer (ITS)

ITSoneDB (Eukaryal Ribosomal Internal Transcribed Spacer 1) and Greengenes2 are also gene-marker-specific databases, but they have not been maintained since 2024.

None of these databases comprehensively covers all organisms or all target genes, and the database best suited to one study may be inappropriate for another. Before committing to a database, evaluate its taxonomic breadth, its completeness for your target group and gene marker, and the date of its last update. EMBL-EBI created a curated overview of reference databases for biodiversity studies for a training course, as a first step towards a common reference.

Where no curated database adequately covers your target taxon, a custom or supplemented database can be built. Its construction and curation process must be documented with the same rigour as for a standard database. It should be deposited in an open data repository and referenced in the FAIRe checklist field otu_db_custom.

It is also important to acknowledge that any database-based assignment carries an inherent identification bias: only what is already known and sequenced can be identified, meaning that a fraction of ASVs/OTUs will remain unassigned in virtually any study. Rather than discarding these, FAIRe recommends retaining them in both the otuFinal and taxaFinal tables, as they may become assignable as reference databases are extended.

For the taxonomic assignment step specifically, see the taxonomic information section of the Biodiversity domain page.

Preserving and sharing data

Making research data FAIR and, where applicable, compliant with the CARE principles [5] requires that all data and metadata produced throughout a metabarcoding study are made accessible, with persistent identifiers, in appropriate public repositories.

All genomic data produced should be submitted to ENA, part of INSDC, within a short time (ideally at most one month) after data generation by the sequencing facility. All samples registered with ENA receive a BioSamples accession as well as a standard ENA sample accession, which serves as the persistent identifier at the sample level. Sequence submission to ENA can be facilitated by MADBOT, a brokering tool developed by the French ELIXIR Node that handles formatting and submission (see also data brokering). PlutoF, a common data space for biodiversity, can also facilitate the deposition of metabarcoding data in ENA and other repositories. Once accepted and published by ENA, the data is automatically made available through the whole INSDC network (ENA, NCBI and DDBJ). You can also submit your sequences to BOLD, the registry for barcode sequences. It uses the Barcode Core Data Model (BCDM) as a standardised data schema and you can transform your metadata from this standard to a Darwin Core Archives (DwC-A) later on.

In parallel with ENA and BOLD submissions, occurrence data (taxonomic tables) should be submitted to biodiversity databases:

• The Ocean Biodiversity Information System (OBIS) for marine occurrence data
• Global Biodiversity Information Facility (GBIF) for all biodiversity occurrence data

Both OBIS and GBIF require data formatted as DwC-A. The GBIF Metabarcoding Data Toolkit (MDT) is a web application that reshapes tabular metabarcoding data (similar to the FAIRe format) and publishes it to both GBIF and OBIS. The R script FAIRe2MDT converts FAIRe-formatted templates to the MDT input format, bridging the two standards.

Data components that cannot be accommodated in the nucleotide or biodiversity databases mentioned above should be archived in generalist open data repositories such as Zenodo, Dryad or FigShare. Do not use generalist repositories, or supplementary materials in journals, as a replacement for the domain-specific databases, because they offer less discoverability and accessibility. Data archived this way must be assigned DOIs and referenced in the projectMetadata component of the study.

Licensing

To ensure reusability (the “R” in FAIR), a clear data licence must be applied when sharing data. CC0 or CC BY 4.0 are recommended for most metabarcoding datasets. If the repository does not include a dedicated licence field, state the licence in the most appropriate available field (for example, a comment or description field). For guidance on choosing a licence, see the overview of Creative Commons licences, the Creative Commons License Chooser or Choose a license, and the RDMkit page on licensing.

Bibliography

1. Blaxter, M. et al. Defining operational taxonomic units using DNA barcode data. Philosophical Transactions of the Royal Society B: Biological Sciences 360, 1935–1943 (2005).
2. Callahan, B. J., McMurdie, P. J. & Holmes, S. P. Exact sequence variants should replace operational taxonomic units in marker-gene data analysis. The ISME Journal 11, 2639–2643 (2017).
3. Takahashi, M. et al. A metadata checklist and data formatting guidelines to make eDNA FAIR (Findable, Accessible, Interoperable, and Reusable). Environmental DNA 7, e70100 (2025).
4. Hakimzadeh, A. et al. A pile of pipelines: an overview of the bioinformatics software for metabarcoding data analyses. Molecular Ecology Resources 24, e13847 (2024).
5. Carroll, S. R. et al. The CARE principles for indigenous data governance. Data Sci. J. 19, (2020).

Related pages

More information

Training

Skip tool table

Tools and resources on this page