Ocean Microbiome Signatures capturing the spatial organization of metabolic potential in the global sunlit ocean. Submitted to Nature Microbiology

Microbial communities are essential to marine ecosystem functioning, yet much of their metabolic potential remains hidden within unassigned sequence space, the ocean’s functional dark matter. Analyzing 1,379 sunlit-ocean metagenomes, we developed a protein-centric framework that clusters 147 million predicted proteins into Operational Protein Units (OPUs), sequence-based groups encompassing both known and uncharacterized proteins. This approach captures the full breadth of microbial functional diversity, extending beyond genome- or pathway-centric views. Using machine learning on OPU profiles, we identified ten Ocean Microbiome Signatures (OMSs), functional groups that represent dominant metabolic strategies of marine microbial communities. Each OMS displays distinct spatial patterns, taxonomic composition, and functional repertoires. By integrating functional dark matter into ecosystem-level analyses, this protein-centric framework expands current models of the ocean microbiome. It uncovers fundamental principles linking protein repertoires to environmental structure, revealing how conserved ecological strategies underpin the potential metabolic architecture of the global ocean.