AI-identified plant proteins: overview
- Leeds researchers used AI to identify nearly 800 emulsifying proteins
- The system screened tens of millions of plant candidates
- Identified proteins could replace dairy emulsifiers in food products
- New emulsifiers may improve texture stability and clean-label appeal
- The approach offers faster cheaper discovery for meat alternatives
AI may have shaped the next generation of plant-based meat products.
Researchers from the UK’s University of Leeds have developed an AI process that can rapidly identify plant proteins capable of acting as emulsifiers, from a choice of tens of millions of initial candidates. The process has already identified nearly 800 proteins.
These proteins could potentially be used widely in both food and cosmetics. They could be particularly useful in plant-based meat substitutes.
Why emulsifiers are important in food production
Emulsifiers are an important part of food production because they enable the blending of oil and water into stable mixtures. They also reduce stickiness and prevent separation of components in processed foods.
Before the development of this AI tool, explains Dr Simha Sridharan, one of the researchers, predicting which plant proteins could be used as emulsifiers was a costly trial-and-error process. There had been no reliable way to quickly predict which plant proteins would do the job.
Such plant proteins act as animal-free alternatives to widely used dairy-sourced emulsifiers, such as caseins and whey.
How the AI works
The researchers’ AI used a simulation model to explore the behaviour of proteins towards oil and water mixtures, in order to ascertain whether they could work as emulsifiers.
The AI was able to learn which segments of the protein were relevant in ascertaining whether it would act the way an emulsifier would. As a result of this, the researchers could screen plant proteins to test whether they would behave like animal-based emulsifiers.
“Emulsifiers often have a characteristic chemical structure called di-blocks,” explains Dr Rik Sarkar, one of the researchers. “We were able to model this structure mathematically for plant proteins. Using machine learning based on features obtained from statistical physics simulations, we can predict which plant proteins are most likely to work best as natural emulsifiers.” A di-block is a kind of polymer made of two ‘blocks’ of simpler polymers.
The researchers tested their method against existing commercially available plant proteins, such as those from pea and potato, and found that it was effective.
What this means for industry
Many of the proteins that the AI found had never previously been considered for this purpose. The AI could open up new doors for the plant-based sector.
The research could enable industry to get plant-based emulsifiers from a wide range of sources for use in plant-based products, including meat substitutes.
New plant-based emulsifiers are currently in high demand in the plant-based meat sector, according to the advocacy organisation ProVeg International.
Better plant-based emulsifiers could help manufacturers solve challenges around texture and stability, as well as meet clean-label demands. They may even have the potential to help products better mimic meat, ProVeg suggests.
While the true potential of the method remains unknown, what is sure is that it is a cheaper, more efficient way to find a highly important functional ingredient for the plant-based meat sector. A wide range of proteins have already been discovered, and it is likely that more are on the way.
