Phys.org — Chemistry · September 7, 2026

From peas to potatoes, AI identifies nearly 800 promising plant proteins for food and cosmetics

AgricultureChemistryMaterials Science

THE AI ANGLE

Screening candidate plant proteins to predict emulsifier functionality

Researchers combined machine learning with statistical physics simulations to screen tens of millions of plant proteins, identifying nearly 800 candidate molecules capable of functioning as natural emulsifiers. Validated experimentally using pea and potato proteins, this computational pipeline circumvents costly trial-and-error to accelerate the replacement of synthetic and animal-derived surfactants in food, cosmetic, and industrial formulations.

THE TEACHING ANGLE

Instructors can explore how modeling molecular diblock architectures and interfacial oil-water physics enables machine learning to predict macroscopic surfactant behavior directly from agricultural plant biopolymers.

Read the original at phys.org Generate teaching or study materials

Instructors get discussion guides, assignments, and mini-cases. Students and readers get a plain summary, class prep, and an exercise. All built from the full article. Three are free with an account.

From peas to potatoes, AI identifies nearly 800 promising plant proteins for food and cosmetics

More in Agriculture