BOLD Genomics · Scientific Data Portal · BioTech / Genomics
Data portal redesign for the Centre for Biodiversity Genomics at the University of Guelph.
Scale: 20 million specimen records, 15.7 million specimens with DNA barcodes, 351,084 species with barcodes.
15+ surface types delivered across data exploration, scientific browsing, geographic navigation, dataset access, and governance. I designed the Data Portal and related parts of the Suite.
BOLD is a cloud-based system for DNA barcode data storage and analysis, developed at the Centre for Biodiversity Genomics in Canada. It consists of four modules: a data portal, an educational portal, a registry of Barcode Index Numbers, and an analytical workbench.
The community spans scientific researchers, educators, and citizens with an interest in DNA barcoding. BOLD functions as public scientific research infrastructure.
Publicly funded. Based at the University of Guelph. Part of the international Barcode of Life research community.
The database had grown beyond its interface. By the time of the engagement, 20 million specimen records sat behind navigation that had not kept up with the database beneath it. Every user could see the scale. Reaching a specific record, a specific taxon, or a specific geographic region required work the interface did not support.
Data retrieval combined multiple scientific criteria: geography, taxonomy, depository. The existing IA did not structure these as parallel entry paths. Individual records held heterogeneous data: identifiers, taxonomy at eight levels, sequence data, barcode indices, and collection metadata at 20 fields. The existing portal held this in rigid table structures that could not carry the mix.
I was brought in through Toptal to redesign the Data Portal and related parts of the Suite. Scope covered Information Architecture, UX, UI, and the Design System.
Discovery was structural and artefact-led. Expert interviews with the scientific community. Component-level UX audit. Data flow mapping.
Expert interviews with biodiversity genomics specialists at the University of Guelph framed the scope of the work and surfaced the taxonomic and methodological depth the redesigned interface had to carry.
The component-level UX audit mapped pain points across every reusable interface element in the existing portal: tables, galleries, area charts, donut charts, maps. Each element was audited for specific issues, with annotated screens.
The data flow map traced access permission logic across three datasets, nine record pages, and BIN pages. Access-denied states were surfaced explicitly in the mapping.
Two sitemap iterations followed. The first was an annotated map of the existing portal, with yellow stickies calling out duplicated nodes, dead files, and misplaced items. The second was a clean restructuring of the full hierarchy, each flag resolved into a structural decision.
The final Information Architecture spanned every major section of the portal, from data exploration to governance, and resolved the hierarchy as one coherent tree.
The architecture drew on prior big-data product design experience and applied it to a scientific database at 20-million-specimen scale. Three decisions shaped it: a taxonomical Information Architecture, a modular component system, and parallel geographic navigation.
The governing structure was the taxonomical IA, designed for progressive depth of engagement. A casual visitor could enter through a country map or a blog article. A researcher could reach individual record and primer pages. The same tree served external scientists, science students, and the broader public, without forcing expert complexity on casual visitors or reducing depth for researchers.
Data drill-down followed the taxonomic hierarchy across eight levels: phylum, class, order, family, subfamily, genus, species, subspecies. The Taxon Treemap carried the primary interaction pattern. The Taxonomy landing page surfaced four kingdoms as parallel entry points: Animals, Plants, Fungi, Protists.
The 'Bits' component system replaced rigid table structures with modular containers. Each Bit held heterogeneous scientific data: identifiers, sequence, metadata, imagery, taxonomy. The format encapsulated different content types without forcing a common shape. Copy-to-clipboard actions were built into the component layer for direct use of BOLD data in papers and citations.
Geographic navigation was designed as a parallel entry to the same records that taxonomy reached, accessed through a different path. Land and Sea collection-site maps carried bubble-density layers at regional and global scale.
The redesigned portal covered 15+ distinct surface types, grouped around five functions.
Data exploration covered Public Data search, BIN detail, and Record detail. Record pages carried six information blocks: Identifiers, Specimen Images, Taxonomy, Sequence with illustrative barcode visualisation, Barcode Index Number, and Collection Data. The Collection Data block alone held 20 fields for location, depth, habitat, collector, and sampling protocol.
Scientific browsing covered Taxonomy and Primer detail. Taxonomy had landing and Browser levels. Primer pages held genetic sequence data, target marker, citation references, and quality-run statistics.
Geographic navigation covered Countries & Oceans at landing, country, and ocean levels. Each country or ocean page held a bubble-density map, attribution donuts, sequence-upload area charts, and collection-activity charts.
Dataset access covered the Datasets landing, curator profiles, and individual dataset pages.
Governance covered Home, Blog with listing and authored articles, News & Events, the Public Data API with parameter documentation, About, and Data Packages.
Data exploration
Record detail with six information blocks and 20-field Collection Data.
Scientific browsing
Primer detail with genetic sequence data, target marker, and quality-run statistics.
Geographic navigation
Country and ocean pages with bubble-density maps, donut attribution, activity charts.
Dataset access
Datasets landing, curator profiles, and individual dataset pages.
Governance
Home, Blog, News & Events, Public Data API, About, Data Packages.
The portal shipped six visualisation types: donut charts for Attribution across sequencing centres and specimen depositories, area charts for sequence uploads and collection activity over time, bubble-density maps at country and global scale, the hierarchical Taxon Treemap, multi-column taxonomy tables, and progress bars. Accessibility improvements extended across galleries, maps, data visualisations, and graphs.
A modular design system and handoff documentation closed the engagement, prepared for development continuation.
20,009,310
Specimen records
15,676,313
Specimens with DNA barcodes
351,084
Species with DNA barcodes
20 million specimens. Three audiences. One big-data drill-down architecture built to carry all three.