The Center for Data Innovation spoke with Diane Wu, co-founder of Trace Genomics, an agricultural genomics startup based in Burlingame, California. Wu discussed why it can be useful for farmers to know about different organisms in their soil as well as the potential to fully understand soil health.
Joshua New: I’ve seen Trace Genomics described as the “23andMe of farms.” Is that accurate?
Diane Wu: “23andMe for Farms” is definitely a popular analogy! It’s a great one in some sense but it can also be a bit misleading. Like 23andMe, we’re working to make genetic testing readily available, and we believe that by doing so, we’re actually building a database to power innovation and research at a scale that will completely change the industry. However, we differ from 23andMe in that, instead of testing for the human or plant genome, we’re actually sequencing the soil microbiome. And unlike the human genome, which is difficult to change or act upon the information, the soil microbiome is much easier to change—and in fact, changes with every planting season! We’re making it easy for farmers to access this critical information on the ecosystem that supports the health and yield of their crops. Our customers have been able to use this information to direct real decisions on the farm and to see significant returns.
New: What kind things are the most useful for a farmer to know about in their soil? Is it just the presence of certain pathogens?
Wu: Each tablespoon of soil contains over a billion organisms, and it certainly isn’t just pathogens that matter for plant health. There are thousands of organisms that work together in communities to provide nutrients to the plant, boosting the yield potential for the crop. There are also hundreds more that cause stress to the plant but aren’t often recognized as the main pathogen that kills the plant. There are even organisms that compete with these pathogens for nutrients, which actually helps to suppress the disease. The soil is full of beneficial organisms as well as potentially harmful ones; measuring and understanding the balance of the two is critical to predicting and improving plant health. Our reports show farmers discrete, actionable interventions they can make from the results of their test.
New: Microbiomes can contain millions of different organisms. Can you describe what’s involved in sequencing such a huge amount of genetic information? How do you extract meaningful information from this?
Wu: The microbiome can contain millions of different organisms, but our public knowledge of what these organisms are and what they do is still quite limited. Microbiome research traditionally focuses on bacterial populations because they’re easier to analyze and cheaper to sequence. However, the soil in particular is rich with fungi that play a significant role in regulating plant health and nutrition, so capturing this population is important too. We’ve developed a process to digitize the soil microbial population for both fungi and bacteria by breaking the organisms up into lots of pieces, sequencing these pieces, and then using software to figure out what organisms they come from. This allows us to use machine learning and genomics to deliver diagnostic capabilities far above what was previously available to farmers or to the scientific community. For example, we developed the first test on the market for a major lettuce disease that is decimating the Salinas area. Previously, farmers were unable to test for this disease and watched their fields melt away after an entire season of hard work. In order to deliver the diagnostic capability for this disease, we had to write software that is able to distinguish between very similar organisms that share very similar genomes, some of which are pathogenic and some of which are not. We then use machine learning and AI tools to fill the gaps of knowledge between what organisms exist in the soil and what that means for crop yield.
New: Is this kind of analysis just useful for industrial-scale farms, or could the average gardener benefit from knowing more about their soil’s microbiome?
Wu: We’re working to democratize microbiome testing so that any farmer, large or small, can use the test. We’re an artificial intelligence company, so there is of course a lot of powerful analytics we can deliver on the microbiome when we work with a larger farmer across fields with varying levels of productivity. However, a big focus for our work is to provide actionable insights from just a few samples by applying the power of big data to our soil database. We now deliver reports on over a dozen different types crops, including berries, greens, orchards and nut trees, grapevines, vegetables, corn, and soybeans.
New: The more people use Trace Genomics’ services, the more data you have at your disposal. What kind of things do you hope to learn about soil health?
Wu: Soil health is so poorly understood but its importance has been recognized by farmers for centuries, so we’re working to transform an art into a science. Soil microbiomes are so complex that farmers can’t make sense of it alone, so we’re helping to build a community around this problem and using big data tools to advance our understanding of what a “healthy soil” and a “diseased soil” looks like. So far, we’ve already identified over a dozen disease susceptibility factors and that has enabled us to increase our disease prediction accuracy dramatically. And we do more than just predict disease. We also point our customers to why they’re experiencing yield losses and what they can do to prevent it. We’ve learned a lot about organisms that provide nutrients for the plant and boost yields, so farmers can use these specific insights on the balance of soil health and disease to identify remediations that increase their yield potential even further.