Inside the ‘Public Library of Blueprints’: How the ASAP Discovery Consortium Turned Open Science Into an Antiviral Pipeline

By Allison Proffitt 

October 21, 2026 | When the ASAP Discovery Consortium took home an Innovative Practices Award at this year's Bio-IT World Conference & Expo, the recognition capped a four-year experiment in doing drug discovery almost entirely in the open. In a winners' session at the event, Nick Lynch, Data Management Lead at Curlew Research and one of the project's team leaders, walked attendees through what it actually took to build a global, 20-institute antiviral pipeline that publishes its data as it goes—not after a paper clears peer review, not after a patent is filed, but essentially in real time. 

"ASAP was a collaborative science project involving upwards of 20 institutes, created after the initial [Covid-19] pandemic, but really looking at post-pandemic preparedness, with grants from the NIH AViDD centers," Lynch told the room, using the shorthand for the AI-driven Structure-enabled Antiviral Platform. The project's aim, he said, was simple to state and hard to execute: build clinic-ready antiviral candidates using an open-science model and make sure every piece of data generated along the way is treated as something the wider research community can use. 

From COVID Moonshot to a Multi-Virus Portfolio 

ASAP is the direct successor to the COVID Moonshot, the crowdsourced antiviral discovery effort that emerged during the pandemic's early days. According to the consortium's award entry, that earlier project's decision to openly disclose compound, structural, and assay data ultimately helped seed the discovery of Ensitrelvir, an antiviral now approved in Japan and fast-tracked by the U.S. FDA, with ASAP's predecessor data cited directly in the Journal of Medicinal Chemistry paper describing the drug (DOI: 10.1021/acs.jmedchem.2c00117). 

ASAP set out to generalize that success across a much broader threat landscape. Rather than chasing a single pathogen, the consortium built parallel discovery programs against coronaviruses, flaviviruses (including Dengue and Zika), and picornaviruses (including enteroviruses EV-D68 and EV-A71)—families chosen, per the entry, because they cause recurring epidemics and currently lack effective therapeutics. Cross-viral efficacy is a challenge, Lynch conceded, but that is one of the specific problems the consortium set itself, with at least one candidate now showing applicability against both Zika and Dengue. 

The numbers behind that effort are substantial. ASAP has collected more than 18,000 X-ray datasets, deposited over 1,250 structures into the Protein Data Bank, and in 2023 alone accounted for roughly 6% of all new X-ray protein structures added to the PDB worldwide—equivalent to two-thirds of that year's output from every U.S. synchrotron combined. The pipeline has produced a preclinical candidate now patented as ASAP-0017445, alongside eight hit-to-lead progressions and ten resistance-robust drug targets. 

Treating Data as a "First-Class Research Product" 

The philosophy Lynch described—and that runs through the award entry—is that the byproducts of drug discovery are worth as much as the drug itself. "Each one of those was considered a sort of first-class citizen in terms of how we share the data," he said of the consortium's core assets: structures, assay protocols, plasmids, chemical matter, and target enabling packages (TEPs), a concept the team borrowed from the structural genomics consortium tradition, that comprises the complete data package needed to enable a structure-based drug discovery campaign against a target. 

That meant building shared infrastructure most single-site biotechs never have to think about: compound registration rules that work identically across every chemistry partner, standardized assay protocols, and public identifiers that let a molecule or protein be tracked consistently from an internal database out to a public repository like ChEMBL or the PDB. "We have about 180 protocols that we've published out on protocols.io so far, both for crystallography and more traditional assays," Lynch said, noting that getting to a common best practice across dozens of labs "has taken some work." 

Lynch framed the project's approach to FAIR data principles—Findable, Accessible, Interoperable, Reusable—as clustered. The Findable and Accessible pieces, he argued, tend to arrive together because they depend on winning over scientists day to day, while Interoperability and Reusability demand a steeper, slower investment in ontologies and metadata. "You need to start early with that," he said, "but again, it comes with a whole extra burden of detail"—one the team tried to absorb early so data would be, in his phrase, "born fair" rather than retrofitted later. 

A Blind Challenge Built on Timestamped Data 

One of the session's more technical highlights was ASAP's collaboration with the OpenADMET consortium and Valence Labs to run a prospective "blind challenge", a benchmarking exercise made possible specifically because ASAP's data was already organized on a timeline. "Because we had temporal data for our projects ... we made use of that in a blind challenge with the OpenADMET consortium," Lynch explained. Working with Valence Labs, the team released a single project's data as three separate views—potency data, standard ADMET endpoints, and ligand pose data—then let outside groups train and test predictive models against real, held-back lead-optimization results from early 2025. 

"We were really grateful for all the groups that took part," Lynch said, adding that a paper covering the results was forthcoming. He argued the temporal-release format is inherently more useful than typical retrospective benchmarks, since it mirrors how a drug hunter actually has to make predictions before, not after, the answer is known. OpenADMET has since launched further rounds built on the same model. 

Even with the best planning, there are circumstantial challenges. Lynch singled out Enamine, the consortium's chemistry partner based in Kyiv, for particular credit: "during our work... the war with Russia affected them, but their resilience and ability to deliver it is truly amazing," he said. 

Lynch also left the audience with an unresolved tension rather than a tidy answer. Open science and intellectual property don't obviously coexist, and he didn't pretend otherwise. In practice, the consortium's entry notes a middle path: assay data is released quickly for early hit-to-lead work, then held briefly during the lead optimization phase to allow a single compound to be patented, preserving a route to affordable, equitable licensing downstream. 

Measuring Success in "Blueprints," Not Just Drugs 

ASAP is less a single drug program than "a public library for high-end blueprints," an investment in shared infrastructure whose payoff is measured not only in its own pipeline (one preclinical candidate, six lead-optimization campaigns, nine early lead packages) but in how many other labs enjoy a head start on lifesaving work.