Source-linked AI summary
BOLD v4: A Centralized Bioinformatics Platform for DNA-based Biodiversity Data
Sujeevan Ratnasingham, Catherine Wei, Dean Chan, Jireh Agda, Josh Agda, Liliana Ballesteros-Mejia, Hamza Ait Boutou, Zak Mohammad El Bastami, Eddie Ma, Ramya Manjunath, Dana Rea, Chris Ho, Angela Telfer, Jaclyn McKeowan, Miduna Rahulan, Claudia Steinke, Justin Dorsheimer, Megan Milton, Paul D. N. Hebert
TL;DR
Biodiversity work relies on robust data sources, and the paper describes a need for standardized, accessible barcode records that support comparison and conservation decisions [7]. BOLD manages molecular and specimen-based data and provides an interface to collect, store, analyze, and publish DNA barcode records; BIN clusters sequences into persistent OTUs to address dark taxa lacking scientific names [5, 19]. By 2022, BOLD had published 794 datasets with attributed DOIs, and its BIN system had been cited in more than 2,000 publications.
Problem
Biodiversity work relies on robust data sources, and the paper describes a need for standardized, accessible barcode records that support comparison and conservation decisions [7].
Source: §1.1 · p. 2Method
BOLD manages molecular and specimen-based data and provides an interface to collect, store, analyze, and publish DNA barcode records; BIN clusters sequences into persistent OTUs to address dark taxa lacking scientific names [5, 19].
Source: §1.5 · p. 5, §7 · p. 39Results
By 2022, BOLD had published 794 datasets with attributed DOIs, and its BIN system had been cited in more than 2,000 publications.
Source: §1.5 · p. 5, §4.8 · p. 32Takeaways & Limitations
DOIs provide unique dataset identification and support sharing and citation, while BINs enable biodiversity evaluation in groups with little taxonomic attention.
Source: §1.5 · p. 5, §4.8 · p. 31The Neighbor Joining tree is limited to 25,000 records, although the ID engine scales linearly to handle all data records.
Source: §5 · p. 33
Abstract
from arXiv · showhide
BOLD, the Barcode of Life Data System, supports the acquisition, storage, validation, analysis, and publication of DNA barcodes, activities requiring the integration of molecular, morphological, and distributional data. Its pivotal role in curating the reference library of DNA barcodes, coupled with its data management and analysis capabilities, make it a central resource for biodiversity science. It enables rapid, accurate identification of specimens and also reveals patterns of genetic diversity and evolutionary relationships among taxa. Launched in 2005, BOLD has become an increasingly powerful tool for advancing understanding of planetary biodiversity. It currently hosts 17 million specimen records and 14 million barcodes that provide coverage for more than a million species from every continent and ocean. The platform has the long-term goal of providing a consistent, accurate system for identifying all species of eukaryotes. BOLD's integrated analytical tools, full data lifecycle support, and secure collaboration framework distinguish it from other biodiversity platforms. BOLD v4 brought enhanced data management and analysis capabilities as well as novel functionality for data dissemination and publication. Its next version will include features to strengthen its utility to the research community, governments, industry, and society-at-large.
1 Introduction
BOLD brings sequence, taxonomic, and occurrence data together as both an aggregator and repository, with tools for barcode identification, collaboration, and publication. Its BIN system provides a standardized way to group sequences for work on taxa without scientific names.
1 Introduction
- BOLD interfaces with sequence, taxonomic, and occurrence databases as both an aggregator and repository, focusing on standardized gene regions [1–4].The paper describes genic analysis as a way of defining species [5].Source: §1 · p. 2
1.1 A Central Database
- Centralized barcode records enable cross-region and cross-taxon comparison and data sharing, while standardized formats support biodiversity monitoring and conservation decisions [7].BOLD is presented as a place to access, store, and analyze these records.Source: §1.1 · p. 2
1.2 Data Standards
- Formal barcode status requires designated marker regions, fewer than 1% ambiguous bases, primer sequences, country of origin, and at least 75% of region length, though other records may still be stored [9–12].BOLD checks COI translations against an HMM, examines stop codons, and checks for known contaminants.Source: §1.2 · p. 3
1.3 DNA-Based Species Identification
- BOLD identifies unknown barcodes using sequence-similarity thresholds; animal queries use HMM alignment and reference-library distances, while other markers use BLAST+ [14–18].The paper describes similarity methods as computationally efficient and accurate enough for most use cases.Source: §1.3 · p. 3
- Reference libraries are updated weekly with BOLD and NCBI submissions, and private records are included in identification searches despite redacted public data.Users can access identification through the public Identification Page, registered Workbench batch processing, or an API for programmatic requests.Source: §1.3 · p. 3, Table 1 · p. 4
1.5 Registry for Dark Taxa
- The BIN system clusters sequences into persistent operational taxonomic units to address taxa lacking scientific names and support reproducible species delineation [5, 19].Its RESL algorithm combines single linkage and Markov clustering, and the registry is updated monthly.Source: §1.5 · p. 5
- BIN effectiveness as a proxy for species has been demonstrated across diverse taxonomic groups [20–23].BINs are also integrated with GenBank and GBIF and accessible through BOLD consoles and analytical tools.Source: §1.5 · p. 5
- Since launch, BINs have been cited in more than 2,000 publications and support biodiversity assessment in groups receiving little taxonomic attention [24, 25].The paper also describes BINs as accelerating taxonomic research and species descriptions.Source: §1.5 · p. 5
2 Data Model
BOLD's data model combines sequence records with specimen information and supports adding data over time without overwriting populated fields. Specimen and sequence pages consolidate record information and support editing and change tracking.
2 Data Model
- Each record starts with a unique Sample ID and high-level taxonomic assignment; further elements can be added later without affecting populated fields, with permissions managed by data type.Figure 2 distinguishes specimen-related fields in green from sequence-related fields in orange and marks fields required to instantiate a record.Source: §2 · p. 7, Figure 2 · p. 7
2.1 Specimen Data
- The Specimen Page consolidates record elements for direct editing, with versioning, activity logs, Delta Views, and tags or comments supporting curation and auditability.Specimen images support morphological inspection, quality assurance, and taxon profiles; standardized orientation and aspect ratios enable image-array comparison.Source: §2.1 · pp. 7–8
2.2 Sequence Data
- The Sequence Page presents marker sequences, primers, and trace files separately for each marker and supports direct editing with activity logs and Delta Views.Trace files and quality scores inform sequence quality; primer registration is required before assembling a submission package, and sequences become viewable and downloadable after upload.Source: §2.2 · p. 10, Figure 4 · p. 11
3 Data Management
BOLD organizes records through projects, folders, campaigns, and datasets, with access controls suited to different collaboration needs. Project and dataset consoles support record monitoring and inspection, while datasets add DOI-based data sharing for publications.
3 Data Management
- Projects are BOLD’s primary record-management units, with project-level access controls and additional folder, campaign, and dataset structures for organizing work across teams and networks.Folders add an independent access control list; campaigns provide API data feeds but lack an ACL, while datasets can pool records across projects and folders.Source: §3 · pp. 12–14
3.1 Projects
- Project consoles provide real-time monitoring and correction of record-quality issues, while activity reports log uploads and modifications for traceability.Record Consoles expose compliance flags and allow subsets of records to be downloaded, analyzed, or added to a dataset.Source: §3.1 · p. 14
- Project access control lists let project managers set and revise users’ permissions, and activity reports log project record uploads and modifications.The Project Console summarizes specimens, sequences, BINs, sequence recovery, BIN discordance, sequence lengths, and record issues.Source: §3.1 · p. 14, Figure 5 · p. 15
3.2 Datasets
- Datasets are independent collections that can assemble records across projects; private records require permission from their project manager.Dataset access is managed through an ACL.Source: §3 · p. 14, §3.2 · p. 17
- Datasets support analysis and record drill-down, and can generate DOIs that reviewers can cite to view and download publication-related data.The paper links dataset DOIs with data discoverability and traceability (Mannheimer 2016).Source: §3.2 · p. 17
4 Data Lifecycle
BOLD manages records through a flexible lifecycle that integrates specimen information with molecular data and supports updates over time. Specimen submission uses required fields and validation checks, with errors or unresolved information handled through correction, annotation, or a pause in submission.
4 Data Lifecycle
- BOLD’s lifecycle approach integrates specimen information and molecular data in a centralized repository, and each lifecycle phase can be performed as records and requirements evolve.The integrated record can include sequences, primers, traces, images, and maps.Source: §4 · p. 17, Figure 7 · p. 17
4.1 Specimen Submission
- A specimen record begins with a unique Sample ID, field or museum identifier, storing institution, phylum, and collection country; more data can be added later.Records can be submitted through an online form or spreadsheet, but require an existing or newly created project.Source: §4.1 · p. 18
- Record information can be updated at any stage; BOLD replaces submitted values while retaining history and tracking who changed what and when.Some submissions pause until criteria are met, with exceptions including GPS obfuscation for endangered species.Source: §4.1 · pp. 18–19
- BOLD checks submissions for consistency, completeness, and compliance, pausing submissions for unresolved issues and applying prescribed remedies.New names may require supporting references or staff review; typographic or data-entry errors must be corrected before submission proceeds.Source: §4.1 · pp. 18–20
4.2 Image Submission
BOLD accepts specimen images as part of record assembly and uses them to support taxonomic validation and taxon profiles. It supports individual or batch uploads, image descriptors, and varied specimen orientations.
4.2 Image Submission
- Specimen images support validation of taxonomic assignments and contribute images to Taxonomy Browser and BIN profiles; future AI-based classification is planned.The paper cites images as sources of trait data [29] and as useful for validating taxonomic assignments [27, 28].Source: §4.2 · p. 20
- Images can be submitted individually or in batches, with batch uploads using a zip package containing an Excel file and JPEG images.BOLD accepts images up to 20 megapixels and 200 images per package; larger submissions require prior contact with support.Source: §4.2 · p. 20, Table 5 · p. 21
- BOLD recommends diagnostic views and image descriptors, including dorsal, lateral, or ventral views for animals and whole-sheet or detail views for plants.Up to ten images may be associated with each specimen, and each image receives a license, defaulting to Creative Commons Attribution-Share Alike when none is supplied.Source: §4.2 · pp. 21–22, Figure 8 · p. 22
4.3 Sequence Submission
Sequence submission follows specimen registration and requires a registered marker; BOLD associates sequences with records and checks them for stop codons and possible contamination. Trace files and primers support sequence verification and reproducibility and are submitted together.
4.3 Sequence Submission
- Sequence data are added after source specimens are registered, and the marker region must be registered in BOLD’s marker database.Uploads require metadata including the marker and the laboratory that generated the sequences, and accept up to 2,000 sequences per upload.Source: §4.3 · pp. 22–23, Table 6 · p. 23
- BOLD associates uploaded sequences with records, translates protein-coding sequences to flag stop codons, and compares barcode regions with a contaminant library for review flags.Sequences can be uploaded through the Project Console or Main Console and edited, deleted, or replaced from the Sequence Page.Source: §4.3 · pp. 22–23
4.4 Trace and Primer Submission
- Trace files and primers are uploaded together after specimen records exist, supporting sequence verification, reproducibility, and BOLD data-standard compliance.Single-record uploads allow four electropherograms and four optional score files per locus; batch uploads allow up to ten trace files per record and 400 files per package.Source: §4.4 · pp. 23–24, Table 7 · p. 24, Table 8 · p. 25
4.5 Record Validation
BOLD validates records through automated checks at upload and comparison with existing records after upload, combining automated screening with user analysis. Its tools help identify taxonomic, morphological, and geographic inconsistencies.
4.5 Record Validation
- Upload validation checks completeness and controlled vocabularies, flags discrepancies, and may repair or remove erroneous specimen-data elements.Complex fields such as species names and Exact Site descriptors remain difficult to check automatically, leaving comprehensive inspection to submitters and project managers.Source: §4.5 · p. 25
- Sequence anomalies such as translated stop codons require investigation because they may indicate sequencing errors or pseudogenes.Resolution can involve inspecting the sequence and comparing it with sequences from related organisms.Source: §4.5 · p. 25
- After upload, validation compares records with project or published data and includes automated contaminant-library screening alongside user-led analysis.BOLD provides the Taxon ID tree, Distribution Map, and Batch ID Engine to support this work.Source: §4.5 · p. 25
- The Taxon ID tree pairs a Neighbor Joining tree labelled with record taxonomy and an image array ordered to match the tree, exposing misplaced nodes and suspicious image contrasts.The Matching Images feature supplies morphological context that can make contamination or misidentification apparent.Source: §4.5 · pp. 25–26, Figure 9 · p. 26
- The Distribution Map projects specimen coordinates and allows users to correct or remove erroneous coordinate data directly from map points.Source: §4.5 · p. 26
- The Batch ID Engine compares project sequences with all BOLD data to detect taxonomic discrepancies, assuming established BOLD records are accurate.Source: §4.5 · p. 26
4.6 Dataset Assembly
BOLD datasets assemble selected records across projects and support reuse and publication. Before release, validation tools help identify taxonomic discordance, sequence-distance anomalies, completeness gaps, and mismatched identifications.
4.6 Dataset Assembly
- Customizable datasets can pool public and permissioned private records across projects, including selections by taxonomic group or geographic region.The assembled datasets primarily streamline data publication.Source: §4.6 · p. 27
- Datasets can grow progressively with eligible records and include previously published records found through taxonomy-, geography-, or other parameter-based searches, supporting data reuse.Source: §4.6 · p. 27
- Records from diverse sources should be comprehensively validated before publication to uncover and rectify inconsistencies and preserve dataset integrity.The passage identifies poor molecular data, contamination, misidentification, sparse independent sequences, and incomplete annotations as validation challenges.Source: §4.6 · p. 27, §4.7 · p. 27
4.7 Dataset Validation
- BIN Discordance flags differing taxonomic assignments within BINs containing at least two records, while Barcode Gap Analysis compares each species’ intra- and interspecific divergences.Unexpected divergence patterns warrant investigation as possible errors or biological explanations, including recently diverged species.Source: §4.7 · pp. 27–28
- Project Summary reports collection-location distributions and species-level gaps in coordinates, taxonomy, and sequences to check record completeness.Its reports can provide supplementary materials for reference-library publications.Source: §4.7 · p. 28
- Batch ID Engine cross-validates dataset assignments against BOLD sequences; matches to different taxa may indicate analytical mishaps or errors in published datasets.The latter cases should be reported to BOLD support.Source: §4.7 · p. 28
4.8 Dataset Publication
BOLD publication makes assembled datasets publicly accessible and supports their identification and citation through DOIs. It also connects sequence submission to GenBank and records dataset publication outcomes.
4.8 Dataset Publication
- BOLD supports minting a DOI for each dataset at publication, providing unique identification, data-sharing access, and a route to appropriate dataset citation.Source: §4.8 · p. 31
- BOLD's NCBI data-exchange pipeline transforms records into GenBank formats and submits them, then shares accessions with submitters under a GenBank embargo during peer review.Source: §4.8 · p. 32
- After journal acceptance, authors can associate the citation with the BOLD dataset through the Dataset Console publication interface.This is described as boosting visibility of both the data and publication.Source: §4.8 · p. 32
- By 2022, BOLD had published 794 datasets with attributed DOIs.Source: §4.8 · p. 32
5 Integrated Analytics
BOLD integrates analytical tools with its data, enabling immediate analysis after records are added or updated and avoiding data-conversion steps. The tools support varied analyses and result sharing, with some limits on scale.
5 Integrated Analytics
- BOLD enables direct analysis of data in the system, allowing records to be scrutinized immediately after addition or update.Source: §5 · p. 32
- Cluster Sequences generates OTUs from identified or unidentified sequences with RESL, providing an initial estimate of species diversity and specimen affinities.Source: §5 · p. 32
- BOLD’s 11 tools analyze specimen and sequence data, including interactive Neighbor Joining taxon-ID trees, distance summaries, DNA-base composition emphasizing GC-content, barcode-gap summaries, and diagnostic-character analysis.Distance Summary reports divergence at conspecific, congeneric, and confamilial levels; Barcode Gap Summary reports nearest-neighbor distances for each species in a list of specimens.Source: §5 · p. 32, Table 9 · p. 33
- BOLD avoids data-conversion steps by aligning tools with its data model and supports sharing analytical results through stored results and URLs.Users can run analyses on data they can access.Source: §5 · p. 33
- Merged projects consolidate records across projects, while searches by identifiers, taxa, geography, tags, and collection dates can focus analysis on subsets.Source: §5 · p. 33
- The Neighbor Joining tree is limited to 25,000 records, although the ID engine scales linearly and can handle all data records.Source: §5 · p. 33
6 Data Access
BOLD provides public search and retrieval alongside bulk data packages, a taxonomy browser, and programmatic API access. These tools support access to public records, taxonomic coverage, and data subsets for analysis and sharing.
6 Data Access
- Public Data Portal and BIN data are open to users, while private data are masked and used only for statistical overviews.BOLD's public-facing search tools query four databases; the Public Data Portal and BIN database provide access to primary BOLD data.Source: §6 · p. 34
6.1 Public Data Portal
- The Public Data Portal exposes all public records, supports combined keyword searches, and offers downloads in TSV, DWC, XML, and FASTA, plus maps.A single download is limited to 100,000 records.Source: §6.1 · pp. 34–35
6.2 Data Packages
- Data Packages address large-scale access needs with compressed datasets and metadata using the Frictionless Data standard for sharing, validation, and interoperability.Public records are packaged in quarterly snapshots referenced by DOI, with weekly snapshots for users seeking more up-to-date data.Source: §6.2 · p. 35
6.3 Taxonomy Browser
- The Taxonomy Browser summarizes barcode coverage across taxa and links taxonomic hierarchy exploration to specimen and sequence records.Representative images support visual verification of taxonomic assignments, while monitoring coverage can reduce duplicated library-construction effort.Source: §6.3 · pp. 35–36, Figure 17 · p. 37
6.4 Application Programming Interface (APIs)
- The BOLD API retrieves sequence, specimen, and metadata subsets by criteria such as taxonomy, geography, and gene marker, and supports XML and JSON.Its analytical endpoint supports near-real-time barcode-based species identification; Table 10 lists nine API endpoints/functions.Source: §6.4 · p. 38, Table 10 · p. 39
7 Conclusion
BOLD consolidates and manages molecular and specimen-based data through a system for collecting, storing, analyzing, and publishing DNA barcode records. Over nearly two decades, it has expanded in sophistication, scope, and utility and become integral to numerous research initiatives and programs.
7 Conclusion
- BOLD provides an interface for collecting, storing, analyzing, and publishing DNA barcode records, linking genetic data with taxonomic knowledge.Source: §7 · p. 39
- Over nearly two decades, BOLD has expanded in sophistication, scope, and utility.Source: §7 · p. 39
- BOLD has become an integral component of numerous research initiatives and programs.Source: §7 · p. 39