An interdisciplinary group of computer and plant sciences experts at Ben-Gurion University of the Negev and University of California, Davis are using artificial intelligence techniques to help identify cellular reactions that indicate how tomatoes and other crops handle different climate conditions.
The research is ultimately aimed at making the crops more robust while improving their nutritional value at the same time.
In the new paper published in Nature’s Communications Biology journal, the researchers identified new metabolic pathways, which are a set of chemical reactions occurring in a cell that enable it to keep living, growing and dividing.
“The world is facing crop yield loss due to climate change, insects and other stresses,” the researchers say. “Identifying pathways that are activated in varieties of tomatoes or wheat that are more resistant to stress will enable farmers to make crops more robust.
“By identifying metabolic pathways targeted for nutraceuticals, we can also improve the nutritional value of crops.”
The researchers are using machine learning, a branch of artificial intelligence in which systems learn from data to identify patterns and make decisions combined with correlation-based network analysis (CNA). CNA illustrates the relationship between molecular components without prior knowledge of the underlying chemistry.
“Much of the information on known pathways that exists is often discovered through time- consuming, experimental processes,” explains Dr. Rami Puzis, lead researcher and a member of the BGU Department of Software and Information Systems Engineering.
“We hope that this faster computer-based approach to understanding how plants react to environmental (abiotic) stresses like climate change will help address food security and production issues.”
Working with the BGU researchers, the UC Davis team began by collecting existing, known metabolic pathways from public databases. They constructed “correlation-based networks” of metabolites – biochemicals found in a line of specially-bred tomato plants. These metabolites link similar molecules, but the connections are not always clear. The researchers then mapped the known pathways onto the correlation networks to create a set of partial metabolic networks.
Using machine learning and raw data from analyzing all the biochemicals in tomato fruit, the system was able to identify four entirely new pathways in a tomato. These are: beta-alanine degradation-I, tryptophan-degradation-VII-via-indole-3-pyruvate (previously unknown in plants), beta-alanine biosynthesis-III, and melibiose degradation.