That distinction was highlighted this week by the Pan American Health Organization and World Health Organization during work to strengthen genomic-surveillance capabilities across Latin America. Specialists from human and animal health laboratories in 11 countries recently gathered in Panama City for a regional course in bioinformatics applied to genomic surveillance of pathogens with epidemic and pandemic potential.
The objective went considerably further than teaching people how to generate sequences. Participants worked on sequence quality, alignment, phylogenetic analysis, interpretation and ultimately the preparation of reports capable of communicating genomic findings to decision-makers. PAHO summarised the challenge clearly: generating the sequence is only one part of the process. The real objective is turning genomic information into evidence.
That raises an important question for biological-data infrastructure: what has to happen between a sample entering a laboratory and genomic information becoming trustworthy enough to support a decision?
It is easy to represent genomic surveillance as sample → sequence → answer. Real systems look very different. A more realistic representation might be: case or surveillance event → biological sample → sample metadata → laboratory accession → extraction → sequencing → quality control → bioinformatics → genome assembly or consensus → comparison with other sequences → phylogenetic analysis → epidemiological context → interpretation → report → decision.
Every stage contributes something, and every transition creates an opportunity for information to be lost. The sequence itself is therefore only one object within a much larger evidence chain.
Imagine receiving the genome of a pathogen. By itself, the sequence can tell us a great deal about that organism. But surveillance questions usually require additional information: where was the sample collected, and when? From which host — human, livestock, wildlife or another source? Was it associated with clinical disease, or was it part of routine surveillance or outbreak investigation? What geographical information is available, and was there relevant travel or animal-movement history? Which diagnostic test preceded sequencing, what sequencing technology was used, how complete is the genome, and what quality thresholds were applied?
A genomic sequence becomes considerably more informative when it remains connected to its epidemiological context. This is why genomic surveillance is fundamentally a data-integration problem as well as a sequencing problem.
Not every sequence should be treated equally. Coverage can vary, reads can be poor quality, contamination can occur, genomes may be incomplete, mixed infections may complicate interpretation, and laboratory or computational artefacts can appear. PAHO’s recent regional training explicitly included assessing sequence quality and alignment before interpreting downstream results, because a phylogenetic tree is only as meaningful as the information used to construct it.
Quality control therefore should not disappear once an analysis has passed. Ideally, the resulting genomic record remains connected to information describing the quality of the underlying sequence and the processes used to assess it. The final answer should retain a route back to the evidence supporting it.
Sequencing produces data. Bioinformatics transforms those data into something interpretable. That transformation may involve read processing, quality filtering, alignment, assembly, consensus generation, variant identification, lineage or clade assignment, and comparison against reference databases and previously sequenced isolates.
Each of those steps depends on choices: which reference genome, which alignment tool and parameters, which classification scheme, which comparator database and which version of it. Change any one of these and a downstream result — a lineage call, a cluster assignment, an estimate of relatedness between two cases — can shift. That is not a flaw in bioinformatics; it is simply what it means for a result to be computed rather than observed directly. But it means the bioinformatic pipeline itself needs to be treated as part of the evidence, not as an invisible step between the sequencer and the report.
A phylogenetic tree, a lineage assignment or a similarity score is not self-explanatory. Two sequences can look closely related for several different reasons — recent transmission, a shared common ancestor much further back, convergent evolution, or simply insufficient resolution in the data used to compare them. Distinguishing between these requires the same epidemiological context discussed above, combined with domain expertise in reading what a genomic result can and cannot support.
This is presumably why the Panama City course spent as much time on interpretation as on the technical steps that precede it. A laboratory that can generate a high-quality genome but cannot situate it within a plausible epidemiological narrative has produced data, not evidence. Building that interpretive capacity across 11 countries’ worth of human and animal health laboratories is arguably a harder and slower task than building sequencing capacity itself.
The course’s inclusion of report-writing as a distinct skill is telling. A genomic finding that never reaches a decision-maker in a usable form has limited public-health value, however rigorous the underlying analysis. A report intended to inform a decision needs to state, clearly and honestly, what was found, how confident that finding is, what it does and does not rule out, and what additional information would strengthen or change the conclusion.
That communication step is also part of the evidence chain. If a report drops the caveats and quality context that accompanied the underlying analysis, a decision-maker may end up trusting a result more, or less, than the evidence actually supports. Preserving the connection between the technical finding and its honest presentation is as much a provenance problem as preserving the connection between a sample and its sequence.
Regional genomic-surveillance capacity is being built, in part, because pathogens with epidemic and pandemic potential do not respect laboratory or national boundaries. When a finding from one country’s surveillance system needs to inform a regional or cross-border response, the evidence chain behind it has to survive that handover intact — the same sample metadata, quality information, analytical parameters and epidemiological context that made the finding trustworthy in the first place need to travel with it, not just the headline conclusion.
That is the same principle behind how we think about connecting <a href="”>provenance across systems and organisations at The BioChain, whatever the pathogen, species or sector: a result is only as useful as the chain of evidence a second party can still verify behind it.
PAHO’s framing of this training is a useful correction to a common assumption — that expanding genomic surveillance is mainly a matter of getting more sequencers into more laboratories. Sequencing capacity matters, but it is only the beginning of the chain that turns a biological sample into a decision a public-health authority can act on with confidence.
Quality control, bioinformatics, epidemiological context, interpretation and honest reporting all sit between the sequence and the decision, and each one can either preserve or erode the evidence behind that decision. Investing in the skills and infrastructure that hold that chain together — as this course did across 11 countries — is arguably as important to pandemic preparedness as the sequencers themselves. It also matters because that infrastructure has to keep pace with sequencing itself getting radically cheaper — see our piece on the £12.47-per-sample method reshaping who can afford to sequence.
A regional training bringing together specialists from human and animal health laboratories in 11 countries to build bioinformatics skills for genomic surveillance of pathogens with epidemic and pandemic potential — covering sequence quality, alignment, phylogenetic analysis, interpretation and report-writing for decision-makers.
A sequence only becomes useful evidence once it’s connected to context — where and when the sample was collected, from which host, under what clinical or surveillance circumstances — and once it has been through quality control, comparison and expert interpretation.
Yes. Close similarity between two pathogen sequences can reflect recent transmission, a more distant shared ancestor, convergent evolution, or simply limited resolution in the underlying data — distinguishing between these requires epidemiological context and expert interpretation, not just the similarity score itself.
Because choices made in the pipeline — which reference genome, alignment tool, classification scheme and comparator database version — can change a lineage call or cluster assignment, meaning the pipeline is part of the evidence behind a result, not just a neutral processing step.
Because a genomic finding that never reaches a decision-maker in a clear, honestly caveated form has limited public-health value however rigorous the underlying analysis — the report is where quality context and confidence levels either survive or get lost.
Pathogens don’t respect national borders, so when a finding needs to inform a regional response, the sample metadata, quality information and epidemiological context behind it need to travel with the finding — not just the headline conclusion — for the receiving authority to trust and act on it.
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.