Scientists have developed an AI process that can rapidly identify plant proteins capable of acting as emulsifiers, pulling nearly 800 promising candidates from tens of millions of initial possibilities. The work, published in Communications Chemistry, could cut years off the research process for developing natural, sustainable alternatives to animal-derived emulsifiers used in everything from mayonnaise to moisturizers.
The research came out of the Sarkar Lab at Leeds' School of Food Science and Nutrition, according to Phys.org. It was led by postdoctoral researcher Dr. Simha Sridharan and supervised by Professor Anwesha Sarkar. They worked in close collaboration with Dr. Rik Sarkar, a machine learning expert at the University of Edinburgh.
Emulsifiers bind oil and water into stable mixtures. They are used in lotions, medicinal creams, sauces, ice cream, mayonnaise, and paints. The most common emulsifiers in use today are synthetic or animal-derived, including milk proteins like caseins and whey. Consumer interest in natural, plant-based alternatives has been growing, but the science has lagged behind.
The core problem is scale. "As we want to shift toward more sustainable, plant-based ingredients, scientists face a major challenge: There are millions of potential plant proteins, but testing them all to identify the right emulsifier is expensive and involves a time-consuming trial-and-error approach. Until now, there has been no reliable way to predict which plant proteins are likely to behave as emulsifiers like animal proteins," said Dr. Sridharan.
To solve that problem, the team built a simulation model to understand how proteins attach at the boundary between oil and water, which is the key behavior that makes an emulsifier work. They then applied machine learning to identify specific segments of a protein that control that attachment. By combining machine learning with statistical physics, the team could screen enormous numbers of plant proteins quickly.
Dr. Rik Sarkar described the structural feature they were targeting. "Emulsifiers often have a characteristic chemical structure called diblocks. We were able to model this structure mathematically for plant proteins. Using machine-learning-based features obtained from statistical physics simulations, we can predict which plant proteins are most likely to work as natural emulsifiers."
The proteins they identified come from sources as common as peas and potatoes. The research spans a wide range of plant-based candidates, and the 800 identified so far represent only the beginning of what the tool can do as it continues to be applied.
The practical consequences extend beyond food. Emulsifiers appear in pharmaceuticals, cosmetics, and industrial products as well. A faster, more reliable way to find plant-based versions could accelerate the shift away from animal- and petroleum-derived ingredients across multiple industries.
The researchers say the tool could also help address salt-related challenges in agriculture by pointing toward more resilient plant varieties, though the immediate focus remains on food and cosmetics applications. The study marks an early step, and further lab testing will be needed to confirm which of the 800 identified proteins perform as predicted under real-world conditions.