Research published in Nature Communications on 19 June 2026 provides a striking demonstration. Researchers used Denmark’s extensive COVID-19 surveillance infrastructure to analyse 738,944 high-quality SARS-CoV-2 genomic sequences collected between 2020 and 2024. From this enormous genomic resource, they identified 303 persistent infections and investigated which individuals were most at risk and how the virus evolved during prolonged infection.
The scale is impressive, but the most interesting aspect from a biological data perspective is not simply that Denmark sequenced hundreds of thousands of viral genomes. It is that researchers could connect genomic information with other forms of evidence. That distinction matters, because a genome tells us about a pathogen — a genome connected to its biological and epidemiological context can tell us considerably more.
Most SARS-CoV-2 infections are relatively short. In some individuals, however, infection can persist for considerably longer. These persistent infections are scientifically important because they create an environment in which a virus can continue replicating and evolving within the same host, and researchers have investigated whether prolonged infections could provide opportunities for viral evolution and potentially contribute to the emergence of new viral variants.
The Danish researchers wanted to answer two related questions: who is most at risk of persistent infection, and what happens to the virus while that infection persists? Answering those questions required considerably more than sequencing.
The researchers analysed high-quality viral sequences collected by the Danish COVID-19 Genome Consortium between 2020 and 2024. The dataset contained 738,944 sequences. They searched for individuals who had genetically related viral sequences sampled at least 26 days apart — but simply finding two positive tests separated by several weeks would not necessarily demonstrate persistence, since a person could have recovered and subsequently become reinfected.
The researchers therefore used genomic information to distinguish likely persistent infections: consecutive sequences needed to belong to the same viral variant and share rare single-nucleotide polymorphisms, substantially reducing the likelihood that the observations represented independent infections. Using their criteria, the researchers identified 303 persistent infections. Already we can see why genomic surveillance is different from conventional case counting — the genome itself becomes part of the epidemiological evidence.
The next question was who those 303 people were — not their names, but their relevant biological and health context. The researchers linked Denmark’s genomic surveillance information with national registry data containing individual-level health and sociodemographic information, allowing them to investigate characteristics associated with persistent infection.
They found that immunocompromised individuals had the highest risk of experiencing persistent SARS-CoV-2 infection. Importantly, this was not simply a general association between poor health and persistence — other disease groups associated with mortality, including diabetes, did not show the same association. That conclusion becomes possible because the researchers were able to connect pathogen genome → infection episode → individual → health information → longitudinal observations. The sequence is one component of the evidence chain.
Repeated genomic observations also allowed the researchers to examine what happened to the virus during persistent infections. The sequences showed evidence of positive selection — in simple terms, some mutations appeared to provide an evolutionary advantage under the conditions experienced during persistent infection. The researchers also observed recurrent mutations associated with resistance to antiviral treatment.
This is particularly important. A genomic sequence collected at one moment provides a snapshot; multiple linked sequences collected through time can become a record of evolution. The evidence chain now becomes: individual → persistent infection → sample 1 → viral genome 1 → treatment and biological environment → sample 2 → viral genome 2 → sample 3 → viral genome 3 → evolutionary change. The relationship between those observations is what creates the scientific value.
This is where the Danish study becomes particularly interesting for the future of genomic surveillance. Imagine that Denmark had generated exactly the same 738,944 viral genomes but had removed every reliable connection between those sequences and the wider surveillance system. Researchers would still possess an extraordinary genomic resource — they could study viral lineages, investigate mutations, reconstruct aspects of viral evolution. But questions about the relationship between pathogen evolution and the host would become much harder, or impossible, to answer.
The value therefore does not reside solely in either dataset. It exists partly in the relationship between them, which creates an important infrastructure problem: how do we preserve those relationships without simply creating an enormous unrestricted database containing sensitive biological and health information?
The Danish researchers did not simply make all of the linked information publicly downloadable. The SARS-CoV-2 consensus genome data used in the study are available through GISAID. The linked individual-level clinical, demographic and genomic information is subject to restricted access under Danish data-protection requirements, and researchers seeking access must go through the relevant Danish authorities and institutional processes.
That distinction is essential. Useful genomic infrastructure needs to support two objectives that can sometimes appear contradictory: preserve relationships, and control access to those relationships. A system should be capable of knowing that Sequence X came from Sample Y, which relates to Individual Z, and has associated Clinical Information A — without necessarily allowing every user who can see Sequence X to discover the identity or complete medical record of Individual Z. The relationship can exist without being universally visible. That is an architectural problem as much as a policy problem.
During the COVID-19 pandemic, countries built genomic surveillance systems at unprecedented scale. But SARS-CoV-2 is unlikely to be the last pathogen for which large-scale sequencing becomes useful. Sequencing costs continue to fall, laboratory capability is expanding, and portable sequencing technologies are becoming more capable. Genomic surveillance is increasingly being considered across human health, animal health, food systems and environmental surveillance.
That creates the possibility of enormous distributed biological datasets: potentially millions of pathogen genomes, generated by hundreds or thousands of laboratories, across multiple countries, connected to clinical, veterinary, epidemiological and environmental information. At that scale, sequencing may cease to be the hardest part. Managing the resulting evidence becomes the challenge.
A genomic surveillance system needs to answer deceptively simple questions. Which sample produced this sequence, and where and when was it collected? Which organism or individual did it originate from? Which laboratory processed it, on which sequencing platform, through which version of which analytical pipeline? Has the sequence subsequently been reanalysed? Which epidemiological records relate to it, which users are permitted to see those relationships, and can the history be reconstructed later?
If those answers live only in spreadsheets, laboratory notebooks and disconnected databases, large-scale genomic surveillance becomes increasingly difficult to manage. The infrastructure needs to preserve the chain.
This is precisely the type of problem <a href="”>The BioChain is being designed to address. The objective is not to replace sequencing platforms, laboratory information systems, national health registries or specialist analytical tools — those systems already perform important and highly specialised functions. The missing layer is often the infrastructure connecting them.
A BioChain-style architecture can treat biological information as a series of linked, provenance-aware objects: individual or animal → sample → laboratory event → sequencing run → raw sequence → analysis workflow → consensus genome → epidemiological record → derived analysis. Each object can retain its own identity, and the relationships between them can also be retained. That creates several useful capabilities.
One of the most basic requirements is ensuring that the same biological object remains identifiable across systems. A sample may have a collection identifier, a laboratory accession number, a sequencing identifier, a filename, a repository accession and an identifier inside an analytical platform. Without explicit mapping, those can gradually become disconnected. The BioChain approach is to preserve those relationships rather than assuming one identifier can serve every system.
Every transformation can create provenance: sample A produced sequence B, sequence B was processed by workflow C, workflow C produced genome D, genome D contributed to analysis E. Rather than documenting that history retrospectively, the infrastructure can capture it as the work occurs — which matters particularly when datasets contain hundreds of thousands or millions of sequences, since manual reconstruction does not scale.
Genomic results increasingly depend upon software. A consensus genome is not simply produced by a sequencer; raw information passes through computational pipelines, and those pipelines, reference datasets, software and parameters all change over time, with AI models increasingly entering the workflow. The BioChain can therefore associate an analytical result with the computational process that generated it — software, version, workflow, reference, parameters, execution, output — so that a result can be reconstructed rather than merely retrieved.
The Danish study demonstrates why connecting genomic and health information can be scientifically powerful, and simultaneously why those relationships require governance. BioChain infrastructure can separate the existence of a relationship from permission to resolve it: a researcher might be permitted to analyse pseudonymised genomic and phenotypic information without receiving identifying information, another authorised system may be permitted to resolve the underlying identity, and a public repository may expose the genomic sequence while restricted systems retain sensitive contextual information. The architecture therefore supports controlled linkage rather than indiscriminate centralisation.
When sensitive biological information is used, it should be possible to establish what happened to it: who accessed a dataset, which analysis used it, which version was used, was a result exported, was a dataset subsequently corrected, and which derived analyses depend upon the corrected record. At population scale, those questions become increasingly difficult to answer without explicit provenance. An auditable chain makes them part of the infrastructure.
The future of genomic surveillance is unlikely to consist of every country’s biological information being placed into a single global database. Different jurisdictions have different legal requirements, institutions have different governance arrangements, and some information cannot leave secure environments. The Danish study itself illustrates this: genomic information can be shared more broadly while linked individual-level information remains controlled.
A BioChain approach can therefore support federated biological infrastructure. Data can remain where governance requires it to remain; what travels between systems can instead include permitted data, identifiers, provenance information, queries or derived results. The chain can span organisational boundaries without requiring every component to be physically stored in one place.
The Danish study is impressive because of its scale — more than 700,000 genomes provide extraordinary statistical and evolutionary resolution. But its deeper lesson is about connection. The researchers could ask sophisticated questions because genomic observations could be connected to other observations: sequence to sequence, sequence to person, person to health information, health information to persistent infection, persistent infection to viral evolution. That is what turns a sequence archive into surveillance infrastructure.
As genomic sequencing becomes cheaper and more widespread, this distinction will become increasingly important. The question will no longer simply be how many genomes we can sequence. It will be whether we can preserve enough trustworthy context around those genomes to understand what they mean.
The next generation of genomic surveillance could operate at scales far beyond those created during COVID-19 — human pathogens, livestock pathogens, wildlife disease, antimicrobial resistance, foodborne outbreaks, wastewater surveillance, environmental genomics. Potentially billions of biological observations distributed across thousands of systems.
Generating those data will be an extraordinary scientific achievement. Making them useful over time will require something else: identity, provenance, interoperability, reproducibility, security, governance, and an infrastructure capable of preserving the relationships between biological observations without assuming that every piece of information should be visible to everyone.
The Danish surveillance programme shows what becomes possible when genomic information remains connected to context. The challenge now is designing infrastructure capable of doing that deliberately, reproducibly and at much greater scale. That is the problem The BioChain is working on — the same problem examined from the human side of the evidence chain in our recent <a href="”>Q&A on reconstructing outbreaks.
They analysed 738,944 high-quality SARS-CoV-2 genomic sequences collected in Denmark between 2020 and 2024, identifying 303 persistent infections and investigating who was most at risk and how the virus evolved during those infections.
By requiring that sequences sampled at least 26 days apart belong to the same viral variant and share rare single-nucleotide polymorphisms — genomic evidence, not just the timing of two positive tests, that substantially reduces the chance the two samples represent independent infections.
Because “at risk” requires health context the sequence doesn’t carry. The researchers had to link the genomic data to Danish national registry data to find that immunocompromised individuals were at highest risk — a link between two different datasets, not something visible in the sequence itself.
Not on its own. The article argues that 738,944 genomes without their connections to sample, host, health and timing information would still be a large resource, but many of the most useful questions — like what drives persistent infection risk — would become impossible to answer.
By separating access levels rather than centralising everything: the SARS-CoV-2 genome data are available via GISAID, while the linked individual-level clinical and demographic information stays under restricted access via Danish data-protection processes.
Not replacing sequencers, lab systems or national registries, but providing the connecting layer between them — persistent identity across systems, default provenance capture, reproducible computational pipelines, controlled (not indiscriminate) linkage, auditability, and federation across jurisdictions with different governance requirements.
If your organisation generates, manages, analyses or governs biological evidence and would be interested in participating in a UK or European provenance demonstrator, The BioChain would welcome the conversation.
Get in touch
Regulatory developments, technical notes and platform news — sent occasionally, straight to your inbox.