A team of UK researchers is using artificial intelligence (AI) and statistical physics in a project to identify plant proteins that could replace animal-derived emulsifiers.

The model developed by the researchers, based at the University of Leeds’ School of Food Science and Nutrition, has identified nearly 800 plant protein ingredients that could potentially act as emulsifiers – many of which had not previously been considered for this purpose.

Due to emulsifiers’ vital stability function across a wide range of food and beverage product applications, such as sauces, condiments and dairy alternatives, the food industry is working to develop more natural alternatives to existing synthetic varieties, as well as those derived from animal proteins such as casein and whey.

The Leeds team noted that ‘millions’ of potential plant proteins could have potentially useful, unexplored functional properties – but that identifying which ones work well through conventional laboratory testing can be costly, time-consuming and heavily reliant on trial and error.

Led by Simha Sridharan and supervised by Anwesha Sarkar, the team collaborated with University of Edinburgh machine learning expert Rik Sarkar to explore a faster and more reliable way of predicting which plant proteins could behave as effective emulsifiers.

They used a simulation model based on statistical physics to understand how proteins interact with oil and water interfaces. This interaction is important because, for a protein to work as an emulsifier, it must attach at the interface between oil and water and help stabilise the mixture.

The team then used machine learning to identify specific sections and characteristics of proteins that influence this behaviour.

The results found that pea and potato proteins demonstrated effective emulsification properties, supporting the predictions made by the AI-driven approach. These findings could be valuable for companies developing plant-based, sustainable and ‘clean-label’ food products.

Researchers believe this computational approach provides a promising new way to identify functional ingredients from a significantly larger pool of protein sources, accelerating discovery and development of new ingredient solutions while reducing the time and resources required during early-stage research.

It also shows the potential of cross-disciplinary work, combining food science, protein chemistry, statistical physics and AI, to address sustainable food system and alt-protein industry challenges.