New ASV Registry gives DNA metabarcoding data a permanent home

DNA metabarcoding has quietly transformed how scientists measure biodiversity. Instead of painstakingly sorting through trays of insects or sieving sediment by hand, researchers can now sequence the DNA in an environmental sample and identify the species it contains in bulk. The method routinely matches, and sometimes outperforms, conventional species identification by human experts in quantity, quality and speed, which is why environmental authorities are increasingly adopting it for biomonitoring. Yet the technique has long suffered from an awkward bottleneck at the end of the pipeline: the vast tables of sequences it produces are usually dumped into publication supplements or private repositories, where they languish largely inaccessible and unannotated. A new web platform called the ASV Registry, described in the journal Metabarcoding and Metagenomics, aims to change that by giving amplicon sequence variants a permanent, searchable and citable home.

To understand why the registry matters, it helps to grasp what an ASV actually is. When sequencing reads come back from the laboratory, bioinformatics tools can either cluster them by sequence similarity into molecular operational taxonomic units, known as OTUs, or extract single, quality-assured sequences that represent the exact variants present in the sample. These exact sequences are the amplicon sequence variants, or ASVs. The terms ESV and ZOTU describe the same concept and are used interchangeably in the literature. Because ASVs are single DNA sequences rather than clusters, they offer finer resolution and a higher degree of comparability across datasets. When multiple environmental samples are analysed together, the detected sequences are organised into an ASV table, which combines the extracted sequences of all sampling plots with the number of occurrences of each sequence in each plot. The table can then be enriched with taxonomic identifications, and it is this annotation step that reveals the species composition of the samples.

Taxonomic assignment is performed by matching ASV sequences against DNA barcode reference databases such as the Barcode of Life Data system, the German Barcode of Life library, or NCBI, using search tools like BLAST, BOLD’s identification engine or VSEARCH. Some sequences will have no species or genus name attached, either because they stem from species absent from the reference database or because they belong to populations genetically distinct from the reference material of the same species. The resolution of species detection therefore hinges on the quality and coverage of the reference libraries used. Crucially, these libraries are growing and improving continuously, which means the taxon assignments in any ASV table are subject to constant change. Repeated identification of a given set of ASV sequences is not only worthwhile to exploit this ongoing growth but necessary to keep identifications up to date and as informative as possible. Most published ASV tables, however, are frozen in time as static supplements, making such reannotation practically impossible for anyone outside the original research group.

The ASV Registry was developed to address precisely this gap. It is a full-fledged web application reachable through any browser, and full usage requires only creating an account; the services are entirely free of charge. Technically, the platform is built on the Pyramid web framework and sits atop a MySQL database, with an ElasticSearch indexer managing search and retrieval processes. The registry supports a specific life cycle for ASV tables designed to facilitate effective and efficient use of the data they contain. Before uploading anything, users must first create or be added to a project container, which acts as an organisational unit grouping uploaded data, supporting collaborative work and handling access management. Only members of a given project can upload, analyse, delete and publish the data within it, and project creators can dynamically add and remove members. The amount of data these containers may hold is not limited.

For upload, ASV tables must be in CSV, TSV or XLSX format and feature a basic layout consisting of a single column of DNA sequences, one for each ASV, together with one or more columns each representing a single sampling plot or event. Cells within the sample columns contain occurrence data as integers, and user-defined identifiers for each sequence can also be included. Once the file is selected, its layout is automatically analysed and the columns that will be read into the database are highlighted; if the automatic highlighting fails, the user is prompted to select the columns manually. If the table is a new version of one already in the registry, it can be linked as a child of the parent table, and a set of sequencing primers can be added or chosen from a list of previously used primers. Primer declaration is recommended but not obligatory, allowing datasets with unknown or undisclosed primers to be uploaded. Collection metadata describing the sampling process and locations should be uploaded separately as a collection table, where users assign an attribute to each column from a list of 26 available metadata attributes, including habitat description, coordinates, fixation methods, collection date and sample size.

Once uploaded, each table receives a unique persistent identifier in the form of a simple integer, by which it will be findable at any later point. Each sequence within the table is also saved as part of an ASV set, which consists of the sequence and the sequencing primer pair selected during upload and is assigned a unique identifier in the format GBOL_ASV_ID_X, where X corresponds to a continuously incremented integer. The combination of sequence data and corresponding primers allows for more precise data points while limiting redundancy within the database. Notably, two sequences are assigned the same identifier only if they match each other exactly and the selected primers are identical or no primers were selected. A length discrepancy between two otherwise identical sequences will result in the registration of two separate ASV sequences. The developers acknowledge that this may limit the comprehensiveness of data in some cases, but it allows sequences and corresponding annotations to be compared with a large degree of independence from the method or pipeline used to generate them.

After the initial upload, the table becomes enrichable in various ways. The sequences can receive taxonomic assignments against thirteen reference databases, including multiple SILVA, NCBI and PLANiTS datasets. Users can set various parameters for the annotation process and select whether the chosen databases should be queried in parallel or sequentially. When annotating a single ASV sequence, all hits returned by the respective algorithm are stored in the database together with the taxon found, the associated taxonomy, the E-value and the percent identity. With default settings, VSEARCH returns only the best hit, while BLAST and BOLD return lists of hits. The specimen in the reference database identified for a given sequence is also saved by its ID and accessible with an enclosed link, so further information held by the reference databases is easily reachable. Because assignments are subject to change over time, repeated taxonomic annotation is supported, and only the most recent taxon annotation from a specific reference database is displayed for an ASV sequence, on the assumption that the most recent annotation is the most accurate as reference quality improves.

ASV tables are available for download immediately after upload. If no taxonomic assignments have been performed, the download returns the original file annotated only with the unique identifier assigned to each sequence. If a table has been taxonomically annotated, users can tailor which annotation data to include and where, allowing taxonomy data from several runs and over time to be readily compared in the downloaded file. Users can also choose the output format, with CSV, TSV and XLSX available to accommodate downstream applications. Because of the potential size of uploaded and analysed tables, download preparation is managed by a queue so users can continue working in the registry during processing, and they are notified by email once their file is ready. The registry also provides a versioning utility: when uploading a table, users may choose to upload it as a new version of a previously uploaded table, with the lineage noted in the database. Such a series facilitates keeping track of changes made to the original data, which is useful when testing different workflows or pipelines, and each version can be independently annotated, potentially highlighting differences between datasets in terms of the discovered species.

Once satisfied with the state of a table, users can choose to publish it. First, a publication draft is created and a preliminary DOI is assigned that can already be used to point to the table within the registry. Users must provide a title and short abstract for the table, the full name and affiliation of one or more authors, and a licence declaration, with a CC BY licence recommended to foster unrestricted use of the published datasets. Optional metadata such as collection and processing parameters, keywords, links to INSDC datasets and citations for related publications can be added to enrich the published table or improve findability. The sum of these metadata is submitted to DataCite and the preliminary DOI is finalised. This moves the table out of its original project container, so access is no longer limited to project members; instead, it becomes available to all users, logged in or not, to view, download and use. The registry provides several methods to search through and retrieve project-internal and published data, including filtering sequences by taxon assignment, tables and sequences by project, and sequences by location where collection tables were uploaded, with a map and keyword search on the dedicated search pages.

The ASV Registry was developed within a dedicated work package of the GBOL III: Dark Taxa project, building upon two previous funding phases of the German Barcode of Life project to establish a barcode library for the flora, fauna and fungi of Germany while supporting research into so-called dark taxa, species-rich but understudied groups of insects. The developers plan to position the registry as a contributor to the global metabarcoding data infrastructure, aiming to connect uploaded occurrence data via the GBIF provider LIB, the Leibniz Institute for the Analysis of Biodiversity Change, and to implement the GBIF metabarcoding pipeline for automatic provision of uploaded occurrence data to the Global Biodiversity Information Facility. By providing a platform for researchers to archive, analyse and ultimately publish their ASV data under the principles of findability, accessibility, interoperability and reusability, the registry hopes to foster comparative and collaborative DNA metabarcoding research and biodiversity monitoring. Easier access to published data and structured publication of new data will certainly boost the collective usefulness of DNA metabarcoding, turning what was once a scattered archive of supplements into a living, growing resource for open science.

Subject of Research: A web platform for registering, managing, versioning and publishing amplicon sequence variant tables with DOIs

Article Title: The ASV Registry: a place for ASVs to be

Article References: Bräunig, C., Meid, S., Quast, B., Rduch, V., & Grobe, P. (2024). The ASV Registry: a place for ASVs to be. Metabarcoding and Metagenomics, 8, Article e124564. https://doi.org/10.3897/mbmg.8.124564

Image Credits: AI Generated

DOI: 10.3897/mbmg.8.124564

Keywords: DNA metabarcoding, ASV Registry, amplicon sequence variants, biodiversity monitoring, FAIR data, DataCite, taxonomic annotation, GBOL, reference databases, open science, environmental DNA, data versioning