The ongoing census of microbial life is hampered by disparate sampling across Earth’s habitats, challenges in isolating uncultivated organisms, limited resolution in taxonomic marker gene amplicons and incomplete recovery of metagenome-assembled genomes. Here we quantify discoverable Bacterial and Archaeal diversity in a comprehensive, curated cross-habitat dataset of 92,187 publicly available metagenomes. Clustering 502 million sequences of 130 marker genes, we predict ~705,000 Bacterial and ~27,000 Archaeal species-level clades, the vast majority of which were hidden among unbinned contigs. We estimate that ten and 145 previously undescribed Archaeal and Bacterial phyla, respectively, are discoverable in this dataset. We identify soils and aquatic environments as hotspots of discoverable lineages, but predict that undescribed taxa remain abundant across all habitats. Finally, we show that prokaryotic diversity appears to arise within common evolutionary patterns, as clade size distributions follow power laws, consistently across the Tree of Life. Re-analysis of over 92,000 metagenomes reveals hundreds of thousands of previously undescribed Bacterial and Archaeal clades hidden in plain sight.

Prokaryotic Argonaute proteins (pAgos) are programmable nucleases that always utilize DNA guides to cleave DNA targets. Recent studies show that some pAgos preferentially utilize DNA guides to cleave RNA targets rather than DNA targets. VbAgo, derived from a Verrucomicrobia bacterium, is a nuclease capable of specifically cleaving single-stranded RNA and highly structured RNA substrates at 37 °C, making it an ideal candidate for developing RNA manipulation toolkits. An in-depth investigation of its mechanism contributes to understanding the functional characteristics of gDtR-type Ago proteins. Here, we present cryo-electron microscopy structures of VbAgo, the VbAgo-guide DNA binary complex, multiple wild-type VbAgo-guide DNA-target RNA ternary complexes, and the catalytically inactive mutant (VbAgo-DM) guide DNA-target RNA ternary complex, with resolutions ranging from 2.5 to 3.2 Å. By integrating these cryo-EM structures with biochemical data, we elucidate the entire catalytic process of VbAgo, revealing its unique C-terminal regulatory mechanism. Specifically, in its apo state, VbAgo’s C-terminus occupies the nucleic acid binding channel, partially impeding its catalytic activity while enhancing its stability. The binding of guide DNA displaces the C-terminus, and subsequent binding of target RNA, along with conformational changes in the N-terminal and PAZ domains, facilitates VbAgo dimerization. Following this, the C-terminus transitions from a loop to a helix, enabling maturation of the catalytic center and inducing movements in the MID-PIWI’ interactions at the dimer interfaces, ultimately leading to dimer dissociation. Concurrently, cleavage of the target RNA and subsequent product release occur, after which VbAgo reverts to its binary state to initiate the next cleavage cycle. Moreover, we demonstrate that VbAgo exhibits guide DNA mediated RNA knockdown activity in mammalian cells. In summary, our study provides a comprehensive understanding of the molecular mechanisms governing self-inhibition, guide binding, target recognition, and product release in VbAgo. These findings offer valuable insights into the diverse mechanisms of pAgos, broadening their functional scope and enhancing the biotechnological potential of pAgo proteins. Prokaryotic Argonaute proteins are programmable nucleases. Here, the authors capturing structures of VbAgo at functional stages to show how its C terminus acts as a regulator and demonstrate its ability to cleave RNA in mammalian cells.

Bacterial biofilms are complex, spatially organized microbial communities that exhibit enhanced resistance to antibiotics and contribute to chronic infections. Understanding their structure, especially during early formation stages, is critical for developing effective intervention strategies. Here, we present a high-resolution computational framework that models biofilms as undirected interaction graphs, where individual bacterial cells are vertices and predicted intercellular interactions are edges. Combining microscopy visualization and deep learning, we developed a pipeline that integrates Mask R-CNN for cell segmentation and a custom neural network (BINet) for interaction prediction. The described graph-based representation enables quantitative analysis of biofilm growth, identification of recurrent structural motifs, and classification of substrate-specific colonization patterns. We demonstrate the utility of this approach in predicting both the developmental stage and material type from image-derived graph features. This study provides the tool for revealing nonobvious patterns of biofilm organization and describes a scalable, high-information-content approach for the automated analysis of microbial communities, opening new possibilities for systems-level microbiological research.

Crop productivity under climate stress remains constrained by conventional agricultural approaches that underuse plant–microbiome interactions. A microbiome-centred, climate-responsive framework was proposed to enhance crop resilience and agro-sustainability by prioritizing targeted manipulation of crop-associated microbiomes, offering a scalable and adaptive pathway to buffer climate stresses and stabilize crop performance.

Plant phenotyping increasingly relies on (semi-)automated image-based analysis workflows to improve its accuracy and scalability. However, many existing solutions remain overly complex, difficult to reimplement and maintain, and pose high barriers for users without substantial computational expertise. To address these challenges, we introduce PhenoAssistant: a pioneering AI-driven system that streamlines plant phenotyping via intuitive natural language interaction. PhenoAssistant leverages a large language model to orchestrate a curated toolkit supporting tasks including automated phenotype extraction, data visualisation and automated model training. We validate PhenoAssistant through several representative case studies and a set of evaluation tasks. By lowering technical hurdles, PhenoAssistant underscores the promise of AI-driven methodologies to democratising AI adoption in plant biology. Application of large language models (LLMs) for automating complex and data-intensive crop phenotype analysis remains unexplored. Here, the authors integrate LLM with curated toolkit for phenotype extraction, visualization, and model training and show its ability to reduce technical barriers and enhance accessibility.

CRISPR RNAs (crRNAs) guide recognition and targeting of intracellular invaders as part of adaptive immunity by CRISPR-Cas systems. crRNAs are transcribed from CRISPR arrays of conserved repeats interlaced with invader-derived spacers. While crRNA production is essential for immunity, its optimization for defense remains poorly understood. Here, we show that, in diverse RNA-targeting type VI CRISPR-Cas systems, the leader RNA encoded upstream of the CRISPR array prevents formation of an invader-independent extraneous crRNA (ecrRNA) by blocking processing of the first repeat. Using the VI-B2 system from Porphyromonas gingivalis as a model, we demonstrate that the leader RNA and first repeat form a conserved inhibitory hairpin that precludes binding and processing by the system’s Cas13b nuclease. Disrupting this hairpin enables ecrRNA production, which in turn can deplete invader-derived crRNAs and reduce Cas13b-mediated phage defense. Structure prediction indicates that these leader-repeat hairpins are widespread across diverse type VI subtypes, highlighting a conserved regulatory mechanism. Our findings reveal how a prevalent branch of CRISPR-Cas systems suppresses ecrRNA formation to promote RNA-guided immunity.