When Open Science Becomes Prior Art: Implications for Bioinformatics and Computational Biology Inventions

The open science movement is changing the landscape of prior art for bioinformatics inventions. Researchers increasingly publish not only journal papers, but also source code, datasets, computational notebooks (e.g. Jupyter), preprints and software packages. Platforms such as GitHub and public data repositories can therefore contain technically significant disclosures long before a conventional journal article appears.

This creates a tension between open collaboration and commercial exclusivity. The research may need to be shared to attract collaborators or validate results, while the underlying technology may ultimately represent a valuable patent asset. An apparently routine decision to upload code or data to a public repository can therefore have consequences for patent protection, investment and licensing opportunities.

Code is Disclosure

Under the European Patent Convention (EPC), the state of the art encompasses everything made available to the public before the relevant filing or priority date, regardless of the medium through which it is disclosed. The EPO expressly recognises internet disclosures as potential prior art.

For bioinformatics inventions, this can encompass public source-code repositories, software documentation, computational notebooks and online databases. A GitHub repository may, for example, disclose an algorithm, implementation details, parameter settings and worked examples that together provide a technical teaching materially relevant to a later patent claim.

The potential sources of disclosure are broader still. A question posted on a forum such as Stack Overflow may itself contain technical information relevant to a later invention. A researcher seeking help with an implementation might inadvertently describe a novel processing step, algorithmic approach or optimisation technique in enough detail for the post, and any resulting discussion, to become relevant to the prior-art analysis.

But the analysis is not as simple as identifying a website and its publication date. Public availability and technical disclosure are separate questions.

A repository may have existed for years while the relevant functionality was only introduced later. The relevant question is therefore what technical information was publicly accessible at the relevant date, not merely when the repository itself was created.

Version History Matters

This makes version control potentially significant evidence in patent proceedings. Unlike a journal article, a software repository is dynamic. Files are added, modified and deleted; datasets are updated; algorithms evolve. A current repository may therefore bear little resemblance to the disclosure that existed when the patent application was filed.

The historical record may consequently be critical: commit dates, releases, tags, archived versions and contemporaneous documentation can help establish what was publicly available and when. The same discipline applies to biological databases and datasets, where accession numbers, database releases and dataset versions may determine precisely what information was publicly accessible at a particular point in time.

Disclosure Across the Digital Ecosystem

Bioinformatics inventions can be disclosed across a much wider range of sources than the conventional patent and academic literature. This is particularly relevant to machine-learning and computational biology inventions, where the distinction between algorithm, training data, model and implementation can be difficult to draw. A patentability search that focuses exclusively on patents and conventional scientific publications may consequently overlook highly relevant disclosures.

Disclosure of an invention may not be confined to a single source. A preprint may describe the methodology, while a code repository provides the implementation and a public dataset supplies the underlying data. Equally, a seemingly informal online discussion may contain a specific technical feature relevant to a later claim. Searching for a disclosure when considering patentability therefore needs to extend across the digital ecosystem, with attention to both the content and timing of each disclosure.

The commercial implications extend beyond initial patentability. A technology may appear differentiated when assessed against conventional literature; however, with an understanding of the way in which disclosures may be made across the digital ecosystem, there is an increased chance of detection of relevant disclosure which may affect patent scope, competitive positioning or the value attributed to the underlying IP in a transaction or investment.

Commercial Implications

Open science and commercialisation are not necessarily binary choices. The AlphaFold ecosystem illustrates how openly available scientific resources can coexist with commercially valuable applications and proprietary developments. AlphaFold-related resources have been made broadly available to the scientific community, while the underlying technology has also supported commercial applications in drug discovery and life sciences.

The lesson is that openness can be managed at different layers of the technology. Foundational research, models or datasets may be openly shared to accelerate collaboration and adoption, while specialised datasets, implementations, workflows, optimisations or downstream applications may remain confidential or be protected through patents and other IP rights.

In practice, consideration should be given to whether relevant information will be disclosed through a code repository, software release, computational notebook, preprint, dataset or public forum. Where patent protection is contemplated, systems should be implemented to establish the relevant filing or priority position before disclosure. Where early disclosure is commercially necessary, the key question of what should be open, what should remain proprietary, and what form of IP protection is most appropriate should be part of the IP strategy.

This is particularly important where IP represents a significant component of enterprise value. Public disclosure may prejudice patent protection and, where confidentiality is critical, undermine the basis for trade-secret protection. Conversely, carefully managed openness can facilitate collaboration and commercial adoption while preserving proprietary value in subsequent developments.

The implications also extend equally to prior-art searching. Searching can no longer stop at patents and papers. Relevant material may reside in code repositories, software releases, preprints, datasets, biological databases and developer forums and establishing the relevance of such disclosures may require reconstructing what was publicly available, in what form, and when. This is particularly relevant when considering patent validity, for example in opposition proceedings and/or during litigation.

This article was prepared by Patent Director Roxna Kapadia and Trainee Patent Attorney Claire Green.