A genomic catalog of Earth's bacterial and archaeal symbionts.

Machine learning predicts that 15-23% of uncultivated bacteria and archaea are likely symbionts across half of all known phyla, revealing widespread symbiosis.

• Why it matters: Understanding microbial symbiosis is crucial for grasping Earth's biological diversity and ecosystem functions, yet it remains underexplored due to culturing difficulties.
• What they did: Researchers developed symclatron, an ML framework, to analyze over 100,000 microbial genomes from metagenomic samples, creating the Symbiont Genomes (SymGs) catalog with predictions of symbiotic lifestyles.
• The result: The study uncovers genomic signatures of symbiosis, such as metabolic gene loss and host-dependent modules, enabling future mechanistic studies and engineering of microbial partnerships.