Salk’s unique partnership uses deep learning software known as SLEAP to study plant traits and accelerate the development of plants that can combat climate change.

The Intergovernmental Panel on Climate Change (IPCC) said removing carbon is critical to combating climate change and preventing global temperature increases. Accordingly, scientists at Salk strengthen the root systems of plants by using their natural ability to absorb carbon dioxide. This optimization aims to increase the amount of accumulated carbon and extend the storage period.

To engineer these climate-resilient plants, scientists at the Salk Harnessing Plants Initiative are using a new and advanced research tool called SLEAP, an easy-to-use artificial intelligence (AI) software that tracks many aspects of root growth. Created by Salk employee Talmo Pereira, SLEAP was initially designed to track the movements of animals in the laboratory. Pereira has now worked with plant researcher and Salk colleague Professor Wolfgang Busch to apply SLEAP to plants.

Advanced research with SLEAP

In a published study Plant Phenomics, Bush and Pereira introduced a new protocol for using SLEAP to analyze plant root phenotypes—how deep and wide they grow, how large their root systems become, and other physical characteristics that were boring before SLEAP. for measurement. Application of SLEAP to plants has already allowed researchers to create the most comprehensive catalog of plant root phenotypes to date.

At best, tracking these physical properties of the root system helps scientists find genes associated with these traits and find out whether multiple root traits are determined by the same genes or independently. This allows Salk’s team to determine which genes are most useful for designing their plants.

“This collaboration is a true testament to what makes Salk science so special and impactful,” says Pereira. “We don’t just ‘borrow’ from different disciplines; we truly put them on equal footing to create something greater than the sum of their parts.”

Before SLEAP, monitoring the physical characteristics of both plants and animals was labor intensive and slowed down the scientific process. If researchers wanted to analyze an image of a plant, they would have to manually label plant and non-plant parts of the image (frame by frame, piece by piece, pixel by pixel). Only then can legacy AI models be applied to image processing and facility structure data collection.

What sets SLEAP apart is its unique use of both computer vision (the ability of computers to understand images) and deep learning (an artificial intelligence approach to teaching a computer to learn and work like the human brain). This combination allows researchers to process images without going pixel by pixel, instead going directly to plant features identified from the input image, bypassing this time-consuming intermediate step.

“We have created a robust protocol that has been tested in many types of setups, reducing analysis time and human error, while emphasizing accessibility and ease of use. And does not require any changes to the actual SLEAP software,” says lead author Elizabeth Berrigan, a bioinformatics analyst in the Bush lab.

Effect of SLEAP on plant breeding

Without changing the underlying SLEAP technology, the researchers developed a downloadable SLEAP toolkit. sleep roots. Using sleep roots SLEAP can manipulate biological properties of root systems such as depth, mass, and growth angle.
Salk’s team tested the package “sleeping roots” on a variety of plants, including crop plants such as soybeans, rice, and canola, as well as model plant species Arabidopsis thaliana – a flowering herb from the mustard family. At a variety of facilities tested, the new SLEAP-based method was able to generate descriptions 1.5 times faster, train an AI model 10 times faster, and predict facility structure based on new data 10 times faster than existing applications, all with equal or better accuracy. They found that it performed better. . from before

With large-scale genome sequencing efforts to elucidate genotype data of a large number of crop varieties, these phenotypic data, such as the root system of a plant growing particularly deep in the soil, can be extrapolated to understand the genes responsible for the formation of plant species. This is a particularly deep root system.

This step, which links phenotype to genotype, is critical to Salk’s mission to create plants that sequester more carbon for longer because these plants will need deeper, stronger root systems. Implementation of this precise and efficient software will enable the Harnessing Plants Initiative to link desired phenotypes to target genes with groundbreaking ease and speed.

“We have been able to create the largest catalog of plant root phenotypes to date, which really accelerates our research to create carbon-sequestering plants that combat climate change,” says Busch, chair of the Hess Department of Plant Science at Salk. “Thanks to professional software development, Talmo SLEAP is very easy to implement and use and will become an indispensable tool in my laboratory.”

When creating SLEAP and Sleap-roots Pereira was thinking primarily about accessibility and repeatability. Since the software and toolkit sleep roots It’s free, researchers are eagerly awaiting how to do this Sleep Roots will be used all over the world. They began discussions with NASA scientists, hoping to use the vehicle not only to help plants absorb carbon on Earth, but also to study plants in space.

At Salk, the collaboration team isn’t ready to disband just yet; They have already begun to tackle the new challenge of 3D data analysis with SLEAP. Efforts to improve, expand and share SLEAP and sleep roots It will continue for years to come, but their use in the Salk Conditioning Facilities Initiative accelerates facility design and helps the Institute address climate change.