To help safeguard bees against pesticides, researchers at Oregon State University’s College of Engineering have harnessed the power of artificial intelligence.
The project, which involved training a machine learning model to determine whether any proposed new herbicide, fungicide, or insecticide would be toxic to honey bees based on the compound’s molecular structure, was led by Cory Simon, assistant professor of chemical engineering, and Xiaoli Fern, associate professor of computer science.
The findings, featured on the cover of The Journal of Chemical Physics in a special issue, “Chemical Design by Artificial Intelligence,” are important because many fruit, nut, vegetable and seed crops rely on bee pollination.
Nearly 100 commercial crops in the United States would cease to exist without bees to spread the pollen required for reproduction. The annual estimated global economic effect of bees exceeds $100 billion.
“Pesticides are widely used in agriculture, which increase crop yield and provide food security, but pesticides can harm off-target species like bees,” Simon said. “And since insects, weeds, etc. eventually evolve resistance, new pesticides must continually be developed, ones that don’t harm bees.”
The algorithm declares two molecules similar if they share many walks with the same sequence of atoms and bonds. Our model serves as a surrogate for a bee toxicity experiment and can be used to quickly screen proposed pesticide molecules for their toxicity.
Ping Yang
To train an algorithm to determine if a new pesticide molecule would be hazardous to honey bees, graduate students Ping Yang and Adrian Henle examined honey bee toxicity data from pesticide exposure tests, involving roughly 400 different pesticide compounds.
“The model represents pesticide molecules by the set of random walks on their molecular graphs,” Yang said.
A random walk is a mathematical notion that represents any wandering path where each step is determined by chance, as if by coin tosses, such as on the complex chemical structure of a pesticide.
Imagine, Yang says, that you are taking a random stroll along the chemical structure of a pesticide, moving from atom to atom by means of the bonds that bind the compound together.
You move in any directions, but you maintain track of your path and the order of the atoms and bonds you pass by. Then you depart on a separate molecule while contrasting the many detours with what you’ve previously done.
“The algorithm declares two molecules similar if they share many walks with the same sequence of atoms and bonds,” Yang said. “Our model serves as a surrogate for a bee toxicity experiment and can be used to quickly screen proposed pesticide molecules for their toxicity.”
The National Science Foundation supported this research.
