Two very different pieces of genomic research published in June 2026 illustrate how radically our ability to reconstruct infectious disease has changed. In one, researchers used ancient DNA to identify two phases of plague among hunter-gatherer communities living around Lake Baikal in Siberia approximately 5,500 years ago. In another, researchers used Denmark’s extensive SARS-CoV-2 genomic surveillance resources to investigate persistent infections across a dataset containing hundreds of thousands of viral genomes.
One study reaches thousands of years into the past. The other examines disease using the infrastructure of modern population-scale genomic surveillance. Yet both demonstrate the same underlying principle: a sequence becomes much more informative when we can connect it reliably to the biological, temporal and contextual evidence surrounding it. For The BioChain, that raises some interesting questions about what genomic surveillance actually requires.
Q: Let’s start 5,500 years ago. What did researchers find at Lake Baikal?
Researchers led by Ruairidh Macleod, Frederik Seersholm and colleagues analysed ancient DNA from Late Neolithic hunter-gatherer remains recovered from four cemeteries around the Angara River near Lake Baikal in southeast Siberia. They generated shotgun-sequenced ancient DNA from 46 individuals. DNA from Yersinia pestis, the bacterium that causes plague, was detected in 18 individuals — a remarkably high detection rate of approximately 39% within the sampled group.
More importantly, the evidence did not represent one isolated infection. The researchers identified two distinct phases of plague, separated by approximately four to six centuries. The first occurred around 5,500 years ago. The genomic evidence therefore pushes confirmed plague outbreaks deep into prehistory.
Q: Why is that surprising?
We tend to associate major infectious-disease outbreaks with cities, trade networks, agriculture and dense populations. These people were hunter-gatherers. The communities around Lake Baikal maintained hunting, fishing and gathering lifeways; domestic plants and livestock, other than dogs, were absent from these communities during the period examined. Yet the researchers found evidence consistent with substantial outbreaks of plague.
That challenges a relatively intuitive assumption — that large, settled agricultural populations were necessary before pathogens such as plague could cause serious outbreaks. Apparently they were not. The study’s authors argue that increased population density and the changes associated with the Neolithic agricultural transition were not prerequisites for significant zoonotic plague outbreaks.
Q: How can researchers call something an “outbreak” after 5,500 years?
This is where the study becomes particularly interesting from a data perspective. Finding Y. pestis DNA in one ancient skeleton tells us that an individual was infected. Finding it in several skeletons still does not automatically reconstruct an outbreak. The researchers combined several forms of evidence: pathogen DNA, human genomic information, archaeological context, radiocarbon dating, burial information, biological kinship, mortality patterns and pathogen phylogeny.
By reconstructing family relationships, the researchers found small groups of biologically related people affected by plague. The first phase appears to have occurred within a single generation. The mortality evidence also suggested acute deaths, with children aged approximately eight to eleven appearing particularly affected among the most acute infections. So the conclusion does not rest upon the pathogen sequence alone. It rests upon relationships between multiple forms of evidence.
Q: So this is effectively provenance?
Exactly. Imagine that we retained only the reconstructed Y. pestis genome. It would still be scientifically valuable. But we would lose much of what allows researchers to interpret it epidemiologically. Instead, the evidence can be conceptualised as a chain: individual → burial → archaeological site → date → biological sample → ancient DNA → Y. pestis detection → pathogen genome → phylogenetic relationship → family relationship → outbreak reconstruction.
Every relationship adds context. That is why provenance matters so much in biological research. The sequence tells us what was present. The surrounding evidence helps tell us what happened.
Q: What does the pathogen genome itself tell us?
The recovered strains sit very early in the evolutionary history of Y. pestis. The researchers conclude that the bacterium’s emergence must have occurred before approximately 5,700 years ago. These early strains are important because prehistoric Y. pestis differed from the forms associated with later bubonic plague — in particular, early strains lacked some of the adaptations associated with the classic flea-mediated bubonic transmission cycle. The Lake Baikal evidence nevertheless indicates that these ancestral forms could be associated with lethal disease.
The researchers also suggest that recurrent zoonotic spillover from wild rodents, particularly marmot populations, may have played an important role. Rather than one continuous human epidemic stretching across Eurasia, prehistoric plague may have involved repeated spillovers from extensive animal reservoirs. That is a distinctly One Health story occurring five and a half millennia ago.
Q: And then we jump from 5,500 years ago to modern Denmark?
Yes — and technologically it is an extraordinary jump. Instead of recovering tiny amounts of degraded pathogen DNA from archaeological remains, modern genomic surveillance can generate enormous numbers of pathogen genomes as infections occur. Denmark built one of the world’s most extensive national SARS-CoV-2 genomic surveillance programmes during the COVID-19 pandemic, and researchers have subsequently been able to use that infrastructure to ask questions that go beyond identifying which viral variant was circulating.
One June 2026 study examined a surveillance resource containing more than 700,000 SARS-CoV-2 genomes to investigate persistent infections. That scale changes what genomic epidemiology can do.
Q: What is a persistent infection?
Most SARS-CoV-2 infections resolve within a relatively short period. Occasionally, however, infection can persist within an individual for a much longer time — particularly in people whose immune systems have difficulty clearing the virus. A persistent infection creates an unusual evolutionary environment: instead of repeatedly moving between hosts, the virus can continue replicating and evolving within the same person over an extended period, potentially allowing mutations to accumulate. Researchers are interested in these infections partly because they provide an opportunity to observe viral evolution occurring within an individual host.
Q: What did the Danish researchers find?
Using Denmark’s genomic surveillance resources, the researchers identified 303 persistent SARS-CoV-2 infections. They found that immunocompromised individuals had the highest risk of persistent infection. The genomic data also provided evidence of positive selection during persistent infections, and importantly, recurrent mutations included changes associated with resistance to antiviral treatment.
That is not simply a sequencing observation. It becomes scientifically meaningful because researchers can connect changes in the virus with information about the infection and its clinical context.
Q: Isn’t that fundamentally the same thing the ancient-DNA researchers were doing?
Conceptually, yes. The technologies and timescales are completely different, but look at the structure of the evidence. At Lake Baikal: person → burial → date → sample → ancient pathogen DNA → genome → kinship → outbreak. In modern surveillance: person → infection → sample → viral genome → longitudinal sequence → clinical context → treatment → viral evolution → epidemiological interpretation. In both cases, the genome becomes more valuable because it remains connected to other information.
Q: Does that mean collecting more genomes isn’t necessarily enough?
Exactly. Imagine a repository containing one million pathogen genomes but nothing else. We know their nucleotide sequences perfectly. But suppose we don’t reliably know where the samples originated, when they were collected, what host they came from, how they were sequenced, which laboratory generated them, which analytical pipeline processed them, whether multiple sequences came from the same individual, what treatments occurred between samples, or what epidemiological events surrounded them. We have an enormous sequence collection, but many of the questions we could ask of it disappear. Volume is not the same thing as evidence.
Q: So metadata are as important as genomic data?
They perform different functions, but genomic data without sufficient contextual metadata can lose enormous analytical value. The genome may tell us how an organism differs genetically; metadata allow us to ask why those differences matter. Where did it occur? When? In which host? Under what conditions? Was the host treated? Did the pathogen subsequently change? Was another case epidemiologically connected? Were two sequences generated using comparable methods? The power comes from connecting those things — which is why modern genomic surveillance is increasingly a data-infrastructure problem as much as a sequencing problem.
Q: Doesn’t connecting all this information create privacy problems?
Absolutely, and this is where the modern example becomes considerably more complicated than the archaeological one. Rich metadata increase scientific value, but information about living people can be sensitive. A genomic surveillance architecture therefore cannot simply maximise linkage — it has to provide controlled linkage. Different information may need different access permissions. Identifiers may need to be pseudonymised. Researchers may need access to derived information without receiving directly identifying records. National authorities may require capabilities that individual researchers do not, and audit records may need to show who accessed information and for what purpose.
The objective is therefore not “put everything into one database.” It is: preserve the relationships required for legitimate analysis while controlling who can resolve and access those relationships. Those are very different architectures.
Q: What does this tell us about the future of genomic surveillance?
Sequencing is becoming faster, cheaper and increasingly distributed. That is extremely positive, but it means the infrastructure surrounding sequencing becomes more important rather than less. If ten laboratories generate sequences, we need interoperability. If one hundred laboratories generate them, we need standards. If thousands of devices and laboratories contribute biological information, we need persistent identifiers, provenance, access controls and reproducible analytical workflows. And if AI increasingly interprets those sequences, we also need to know which model and computational process produced the resulting analysis. The sequencing machine is therefore only one component of genomic surveillance — the wider system has to preserve the chain from biological event to scientific conclusion.
Q: What does the Lake Baikal work tell us about The Ancient BioChain?
Quite a lot. Ancient pathogen genomics is planned as <a href="”>Stage 2 of The Ancient BioChain, and the Lake Baikal research demonstrates why we are interested in it. We don’t simply want to ask “can we detect an ancient pathogen?” The much more interesting question is: what can we reconstruct when pathogen DNA remains connected to host, chronology, location, archaeology and other biological evidence? That potentially allows us to move from ancient sequence, to ancient infection, to ancient transmission, and ultimately to ancient epidemiology. A genome becomes one component of a much larger biological record.
Q: Five and a half thousand years from now, could somebody analyse today’s outbreaks in the same way?
Perhaps that is the most interesting question of all. We are generating an unprecedented biological record of contemporary infectious disease — pathogen genomes, host genomes, clinical information, environmental surveillance, animal-health data, treatment information, geography, epidemiology, computational models. But whether future researchers can use that record will depend partly upon what we preserve now. File formats will change. Software will disappear. Databases will be replaced. Organisations will close. Identifiers will change. Analytical pipelines will evolve.
The irony is striking. We can reconstruct part of an outbreak that happened 5,500 years ago because biological material, archaeological context and modern analytical methods happened to come together. Meanwhile, we could conceivably create genomic datasets today that become uninterpretable within decades because the relationships surrounding them were not adequately preserved.
Q: So what’s the lesson from these two studies?
Genomics has transformed our ability to observe infectious disease. It allows us to watch a virus evolve within an individual today, and it allows us to reconstruct the spread of a bacterial pathogen among a family that lived thousands of years before anyone knew bacteria existed. But the genome alone is rarely the whole story. To reconstruct an outbreak we need relationships: who or what was infected, where, when, which sample produced the sequence, how it was analysed, what other biological or epidemiological evidence surrounds it, and whether we can reconstruct those relationships later.
Five thousand years of technological progress separate Lake Baikal from modern genomic surveillance. The underlying information problem is surprisingly similar. A sequence tells us about a pathogen. A connected chain of evidence can tell us about an outbreak.
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.