Hamburg-based Soilytix has launched LOAM, a family of genomic AI models meant to help crop-protection companies decide which of the many potential biological products found in soil are worth sending into expensive laboratory, greenhouse and field trials, AgNavigator reported on 5 October.
"AI is producing very large numbers of potentially testable candidates. But if they all still have to pass through expensive laboratory, greenhouse and field experiments, simply generating more candidates can make the bottleneck worse," a Soilytix executive, Rajakumar, told AgNavigator. The company's answer is to narrow the pool step by step before that testing begins, not to replace it.
LOAM was trained on 15,640 microbial genomes reconstructed by long-read sequencing of soil, sediment and water, about 67.5 billion DNA bases in all. In the example Rajakumar gave, a company looking for antifungal peptides against a particular Fusarium species would first compare fields where the pathogen caused disease with fields where it was present but no disease developed. That "disease-suppression signal" points to the microbes that may be holding the fungus back; their genomes are searched with a version of LOAM tuned for antifungal discovery, candidates are filtered by computer, and a small set goes into cell-free protein synthesis screening before partners take the strongest into greenhouse and field tests.
The company also wants to explain why microbial inoculants, seed coatings and biostimulants work in one place and fail in another. Rajakumar and chief executive Bruno Steinkraus previously worked on biomarkers and patient stratification in cancer medicine, and they see the same question in farming: which biological product is likely to work in which soil, crop and environment. Soilytix already analyses soil biomarkers for companies running R&D trials.
There is a caveat: the approach has not yet produced a wet-lab validated candidate. Its pipeline from field data to computer-validated antifungal peptides is running, and Soilytix is starting its first in-vitro validation with the University of Hamburg. Discovery pilots with crop-protection companies are planned for 2027, and the company says it does not yet have the data to say how much time or money its enrichment would save.
For farmers, the stakes are in the price and reliability of biological alternatives to chemical pesticides: fewer dead ends in development and better matching of products to soils could make them more dependable.
Source: AgNavigator
Comments
(0)