A viral-protein library the size of a small city

On 10 July 2026, a European-led consortium released the largest open catalogue yet of the proteins that viruses use to infect, hijack and reprogram human cells. The dataset, described in a study coordinated by the European Molecular Biology Laboratory and partners across nine countries, covers more than 7,500 proteins drawn from hundreds of viral species — including every family on the World Health Organization's watchlist of pathogens with pandemic potential.

The release is less a single discovery than a piece of public infrastructure. By mapping the full protein inventory of dangerous viruses in one searchable place, the project aims to give laboratories everywhere the same starting point when the next outbreak arrives, and to break the long-standing advantage that well-funded institutes in rich countries hold in the early weeks of a crisis.

What exactly was published

The catalogue is built around two ideas. The first is completeness: rather than sequencing and assembling one new virus at a time, the team expressed more than 7,500 viral proteins as physical samples — each cloned, purified and verified — and paired each one with structural and interaction data. The second is comparability. Every protein is run through the same workflow, against the same human-cell reference, so a coronaviridae spike protein and a filovirus polymerase can be compared on the same axis for the first time.

The release also includes the experimental assays needed to ask disease-relevant questions of each protein: where it binds on a human cell, what host machinery it recruits, whether antibodies from recovered patients recognise it. According to the project summary, the platform is designed so that a researcher in any participating country can request a specific protein and receive it under standardised terms, without negotiating a fresh material-transfer agreement for each request.

Why the timing matters

Public-health agencies have spent five years trying to shorten the interval between an outbreak and a deployable countermeasure. SARS-CoV-2 went from unknown sequence to first vaccine candidates in roughly 11 weeks, a pace widely treated as the benchmark for what is achievable. Project leaders argue the real bottleneck is not the vaccine chemistry but the labour of cataloguing, in advance, the proteins a new virus is likely to deploy — a task that historically gets repeated, virus by virus, lab by lab.

The new dataset collapses that pre-work. A research group studying Crimean-Congo haemorrhagic fever, for instance, can now skip months of cloning and characterisation and start directly with the relevant viral proteins in hand. The WHO's R&D Blueprint for Epidemics, which lists priority pathogens, is expected to use the catalogue as a reference for the next round of target nominations.

Who is excluded, and by design

The release is deliberately narrow on what it does and does not contain. The team excluded sequences supplied under restrictive commercial licences, several highly pathogenic avian influenza strains held under enhanced biosecurity, and any protein that could not be expressed as a soluble, well-behaved sample under standard biosafety conditions. Researchers working on those targets will still rely on the small set of reference laboratories authorised to handle them.

Funding structures shaped the architecture. The project sits inside the European Commission's Horizon Europe programme and counts the EMBL, the Institut Pasteur, the Helmholtz Centre for Infection Research and the Swiss Tropical and Public Health Institute among the lead sites, with bioinformatics pipelines hosted at EMBL-EBI. Roughly a third of the work was carried out in lower- and middle-income partner countries including Senegal, Brazil, Vietnam and South Africa, intended to address a complaint from the COVID-19 era that reagents and data flowed out of outbreak zones but rarely back in.

What it does and does not solve

A shared protein map is not a drug and not a vaccine. The most it can do is compress the early, unglamorous stage of outbreak response — the stage in which individual laboratories clone, express and characterise viral proteins before any therapeutic question can even be formulated — into a single, shared resource. Anyone who has watched the first eight weeks of a novel-pathogen response knows how much time lives in that stage, and how much of that time is spent re-doing work already done elsewhere.

The structural questions remain. Antibody recognition data is included for several viruses but not all; for the ones most likely to seed the next pandemic, coverage is uneven and reflects where clinical samples are available. The team has flagged this gap explicitly and is soliciting additional sera from recovered patients, with a stated preference for samples from regions where the relevant viruses are endemic rather than where the spillovers happen to be studied. The next release, expected in early 2027, is intended to fill in those holes.

Beyond the immediate science, the dataset is a test case for whether open research infrastructure can be built and maintained between crises rather than in the panic of one. The 7,500-protein figure is impressive; more consequential is the institutional arrangement behind it — a multi-country library, on a continent that has spent the past five years arguing over how to share vaccines and antivirals. If it functions, it becomes a model for the next shared instrument. If it doesn't, the post-mortem will be instructive in a different way.

The first independent papers drawing on the catalogue are expected within months. The watch, for now, is whether the early users are in Dakar, Lyon and São Paulo — or only in Cambridge and Heidelberg.

This publication treated the catalogue as public infrastructure rather than as a single breakthrough. The lead reflects the dataset's size and provenance, on the rationale that a shared research tool earns its place by who can use it, not by who built it.