Our researchers, Filipe Marques (MSc student), Ana Margarida Fortes (BioISI), and project coordinator and PI Francisco Couto, developed MicroGenAi, an artificial intelligence framework integrated into LASIGE’s Data and Systems Intelligence (DSI) and Health and Biological Intelligence (HBI) research lines. This project awarded funding through the competitive LASIGE/FCT Seed Projects Call 2025.

The study focused on the automated extraction of biological relationships from scientific text to support sustainable agriculture, with particular attention to interactions between microorganisms, plants, and environmental stresses such as salinity, drought, and extreme temperatures. The results show that even base open-source large language models outperform traditional machine learning approaches without fine-tuning. The team also found that few-shot prompting provides an additional improvement in performance.

Overall, the project highlights the potential of combining open-source AI with strategic information retrieval to help researchers identify relevant biological relationships more efficiently, supporting the discovery of effective biofertilizers.

Photo credits: wikipedia