Nobel Prize celebrates AI’s role in protein structure innovation

The 2024 Nobel Prize in Chemistry was awarded to Demis Hassabis and John Jumper from Google DeepMind for developing AlphaFold2, and to David Baker from the University of Washington for his work in computational protein design. These innovations have revolutionised the understanding of protein structures using artificial intelligence.

Before AlphaFold, determining 3D protein structures was complex and time-consuming, and often required the use of experimental methods like X-ray crystallography and cryo-electron microscopy. AlphaFold is a neural network-based model, powered by a deep learning algorithm that incorporates vast amounts of physical and biological data on protein structures. By leveraging multiple sequence alignments, it can predict protein structures with near-experimental accuracy. Since its launch, AlphaFold has accurately predicted the structure of over 200 million proteins and been used by more than 2 million researchers globally​.  Despite its recent development, AlphaFold2 has already been applied in numerous studies across biology and medicine. Its applications include studying disease pathology to develop targeted therapeutics, visualising enzymes that can decompose plastics, and engineering solutions to antibiotic resistance, among many others. 

The second half of the prize was awarded to David Baker, a structural biologist from the University of Washington, for his research in computational protein design. Baker and his team developed Rosetta, an algorithm for ab initio protein structure prediction—that is, the design of entirely new proteins from scratch. The Rosetta algorithm has been incorporated into AI-powered tools like RoseTTAfold, which uses a three-track neural network to predict protein structures. Baker’s model could revolutionise drug development, synthetic biology, and nanotechnology, opening possibilities for new medical treatments, vaccines, nanomaterials, and sensors. Key applications include designing proteins to mimic those involved in photosynthesis for renewable energy, creating molecular detectors that activate upon detecting cancerous cells, and engineering proteins as microscopic ‘nanocages’ to transport drugs into the body.

AI-enabled inventions 

This is the first time a Nobel prize has been awarded for an AI-enabled scientific breakthrough, and it probably won’t be the last.  

AI is being harnessed across scientific fields to identify meaningful trends in large datasets, predict outcomes based on these data, and simulate complex physical and biological scenarios. AI is influencing all industries, but its impact will be especially strong in pharmaceuticals where it is expected to accelerate drug discovery and, in clinical settings, generate patterns and insights from vast quantities of patient data to enable more personalised treatments. As the life sciences and pharmaceutical industries increasingly rely on AI-driven discoveries, companies operating in this space will need to carefully consider how to obtain meaningful protection for both the AI platforms and the products generated by such platforms. 

AI inventions as intellectual property 

While DeepMind has made AlphaFold2’s code open source, allowing for widespread use and collaboration, its underlying methods are protected through patents. DeepMind filed three international (PCT) applications in 2019, which cover relatively broad methods of generating multiple structural predictions, updating parameters using neural networks, and selecting the final structure.  

At the European Patent Office (EPO) an invention must represent a non-obvious technical solution to a technical problem to be patentable. In this light, it can appear tricky to patent the ‘core AI’, that is, the fundamental building blocks of an AI model, as such algorithms are often regarded by the EPO as abstract mathematical methods lacking a technical effect. However, with the right guidance, certain aspects of AI inventions can be patented. These aspects include the way training data is generated for use in training an AI model, the way the AI model is trained using the training data, and the way the model analyses the data. A clear link must be established between these aspects and a technical effect or advantage – and the technical purpose must be plausibly served across the scope of the patent claims. 

A more obvious target for patent protection is the compounds that are discovered using AI. David Baker is listed as an inventor on many patents, focused on compositions and methods for treating particular diseases. Presumably, some of these compositions include proteins synthesised using Baker’s AI model. Another example is BenevolentAI, a company collaborating with AstraZeneca to enhance both drug development and drug repurposing using AI – they own several patents directed towards the products of their AI model (compounds and uses thereof). 

Whether directed to a product of an AI or the AI itself, securing a patent is advantageous because it gives the owner exclusive rights to manufacture, use, sell or license their invention for a specified period. With these exclusive rights, they can establish a competitive edge, control market entry, and potentially set higher prices due to the lack of competition. Patents can attract investors by assuring intellectual property protection. This boosts the inventor’s reputation, making it easier to secure funding for development and commercialisation. Finally, because patents require public disclosure of the invention, the inventor can inspire future innovation while retaining their exclusive rights. 

Patent or trade secret? 

Alternatively, some aspects of an AI-enabled invention may be kept as trade secrets. Unlike patents, which require public disclosure of the invention, trade secrets remain private. This can be advantageous in certain scenarios, such as when the invention is not eligible for patent protection, or in extremely fast-moving areas of industry where the risk of obsolescence outweighs the benefits of patenting.  

However, trade secret protection relies heavily on the ability to maintain confidentiality by enforcing internal security measures or non-disclosure agreements (NDAs). If the trade secret is leaked or disclosed, its protection is lost, potentially causing significant damage to the business. Further, trade secrets offer no protection against independent discovery or reverse engineering by competitors. This means trade secrets are often better as a complement to other forms of IP protection, eg, patents, rather than as a standalone solution. 

In the case of AlphaFold2, although much of the system is public, certain details about the algorithms, neural network architecture, or data handling processes could be kept as trade secrets. In many areas of technology, companies will often use a blend of patents and trade secrets to maximise protection for their innovations.