Plant scientists representing a cross-section of academic and private sector research institutions joined colleagues from the Department of Energy’s Oak Ridge National Laboratory at a July workshop to review the Advanced Plant Phenotyping Laboratory (APPL) shared-use facility at ORNL, designed to quickly analyze new plant materials and cultivars, and to accelerate breeding of new biofuels, bioproducts and food crops.
Workshop participants discussed the science that APPL’s high-throughput, automated, and AI-assisted platform empowers for the discovery, development and deployment of the next generation of plants for fuels, food, and more. Boosting research community access to the APPL facility promotes open science, the use and sharing of AI-ready data, and can significantly speed up the crop breeding process.
APPL’s automated system moves as many as 5,000 plants through a broad array of imaging systems 24 hours a day. The platform captures massive amounts of data on plant shape, structure and physiology, how they change and grow, react to different light, water, temperature and soil conditions, and their biochemical composition. Data are interrogated using unique AI models, replacing the painstaking, slow process of manual measurements and field phenotyping and leading to accurate predictive tools for plant performance and genes-to-traits linkages. ORNL is currently rolling out a new capability in APPL to view and analyze the plant root environment, or rhizosphere, where interactions with microorganisms play an essential role in plant health and productivity.
“Phenotyping is a vexing problem for those running breeding programs because it’s expensive and slow and sometimes inaccurate, yet it is the most valuable information we can gather,” said Valerio Hoyos-Villegas, assistant professor of plant breeding and genetics and lead for the Dry Bean Breeding and Genetics Laboratory at Michigan State University.
“Saving time in the field is invaluable because it’s labor-intensive, you need a lot of hands, and sometimes conditions can be hot, resulting in decreased concentration, less accuracy and precision,” said Thomas Pendergast, plant scientist at the University of Georgia. “The idea of being able to pass a whole bunch of genotypes of plants through APPL rapidly and getting the amount of data you receive is mind-boggling. It means we’ve almost instantly switched from the bottleneck of doing phenotyping to conducting analysis.”
Bridging the genes-to-traits gap
APPL enables rapid plant phenomics — the science of analyzing all the visible characteristics of plants and how they change over time based on their genetics and environment.
The lab’s broad imaging capabilities and AI-assisted analysis help overcome a critical gap: scientists have built a rich library of genetic markers for desired plant traits using high-throughput genotyping. But they lack access to rapid phenotyping capabilities that assess whether genetic improvements are successful. It can sometimes take multiple years to field-validate whether traits are successfully obtained and measured in new plant cultivars.
“While we’ve made great advances in sequencing, it’s the phenotyping that’s really holding us back,” Pendergast added.
APPL’s advanced phenotyping platform “offers a transformative solution,” said Jerry Tuskan, director of DOE’s Center for Bioenergy Innovation and a Corporate Fellow at ORNL. “By integrating multiple spectral imaging systems with AI-powered analytics, APPL’s high-throughput, cost-effective platform makes it easier to provide services ranging from descriptive assessments of plant performance to accelerating the process of definitively linking genes to traits.”
By integrating multiple spectral imaging systems with AI-powered analytics, APPL’s high-throughput, cost-effective platform makes it easier to provide services ranging from descriptive assessments of plant performance to accelerating the process of definitively linking genes to traits.
Workshop participants explored, among other topics, how APPL can:
- play a larger role in the plant research community, including broader, low-cost access for plant scientists across the nation;
- set up and run experiments in parallel for faster analysis;
- ensure a range of conditions for plant testing;
- provide AI-guided analytics to customers, including access to ORNL’s world-leading supercomputers and foundation models;
- release data in a timely manner and support publications to foster open science; and
- modify APPL’s imaging modalities for use on drones to facilitate field-scale rapid phenotyping.
“Early investment in AI such as we see in APPL will help establish the new paradigm by which AI is utilized to study biological systems, and that will pay off not only in the activities of APPL but in the ability of the entire plant community to access and understand the biology of plants,” said Jose Dinneny, professor of biology at Stanford University.
Jack Wang, associate professor and director of the Forest Biotechnology Group at North Carolina State University, said “If we rely on the resource and expertise that APPL has available to improve our phenotyping throughput, I believe we can significantly advance our capability to build the next generation of AI models and genetic insights, and ultimately the next generation of tree varieties that will align with more efficient, productive, and high-quality conversion of valuable woody tissue into energy, bioproducts, biomaterials and biochemicals.”
Offering a path to energy, food, economic security
“APPL to me is far more than just a phenotyping facility. It is a tool that connects what a plant represents to how it can benefit society,” Wang said. “This connection will allow scientists, governments and society to better understand and utilize our natural resources in ways that will create a stronger economy, ensure food security and develop more plant-based products.”
From an industry perspective, Kevin Falk of Corteva Agriscience said, “High-throughput phenotyping is the next critical step in plant research. APPL is a one-stop shop for many modalities, shoot and root, that can all correlate together. Very few if any other institutions have the capacity to bring all these modalities together and enable many different ways of looking at a plant over time.”
APPL is a real opportunity to accelerate our ability to improve the productivity of plants for food, energy and health,” said David Hanson, associate vice president for research at the University of New Mexico. “APPL is leading the charge in standard-setting for high-throughput phenotyping facilities, which will one day enable the adoption and replication of such systems around the world.”
UT-Battelle manages ORNL for DOE’s Office of Science. The single largest supporter of basic research in the physical sciences in the United States, the Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science. —Stephanie Seay