MicrobeDiscover: A Knowledge Graph-Enabled AI Framework for Identifying Microbes for Inorganic Nanomaterial...

As an environmentally friendly, mild, sustainable preparation method for nanomaterials (NMs) and the foundation of microbe-material hybrid systems, microbial synthesis of NMs hold great promise in the sustainable future. However, reported instances cover only approximately 400 microbes and 90 NMs, merely scratching the surface of the theoretical potential of enormous microbe-NMs combinations. Research methods predominantly based on empirical approaches and trial-and-error are significantly challenged in terms of screening efficiency. Here, we introduce an AI-based framework, MicrobeDiscover, which identifies potential microbes for NM synthesis within a vast search space by integrating and representing microorganisms, NMs, and their interactions. A central component of MicrobeDiscover is a knowledge graph guided by expert insights into microbial synthesis, bridging microbiological and materials science domains to provide the AI model with robust data for screening and predictive modeling. Among the top 20 microorganisms predicted by MicrobeDiscover, the recommendation success efficiency reached 80.77%. Leveraging the framework's predictions, we successfully synthesized several kinds of trimetallic NMs using Shewanella oneidensis MR-1 as suggested by MicrobeDiscover, in the context of non-trimetallic NMs were reported producing in a biosynthetic way. This approach is anticipated to significantly advance the development of effective microorganisms and enhance the controllable synthesis of NMs.