Large language models (LLMs) are increasingly used by plant biologists to summarize literature, generate hypotheses, and interpret experimental results. However, LLMs are unreliable sources of exhaustive, source-attributed facts, a critical limitation for the list-style queries that pervade plant biology (e.g., "list all transcription factors regulating secondary cell wall (SCW) biosynthesis in Arabidopsis"). Here, we query ChatGPT, Claude, and Gemini with such queries and demonstrate that none return complete gene lists with reliable citations. We trace these failures to how LLMs store knowledge: as statistical patterns distributed across billions of internal parameters, with no mechanism to guarantee completeness, provenance, or reproducibility. We also review fine-tuning mitigation strategies, including multi-task instruction tuning, parameter-efficient methods, and context engineering, that alleviate but do not resolve these limitations. We then discuss retrieval-augmented generation (RAG), which feeds relevant documents to the LLM at query time, and argue that while it improves source attribution, it remains impractical when answers require synthesizing information scattered across hundreds of papers. As an alternative, we advocate graph retrieval-augmented generation (GraphRAG), in which the LLM serves as a reasoning and language interface over a structured, provenance-linked knowledge graph (KG) that returns complete result sets reproducibly. We outline a practical GraphRAG architecture and survey existing plant KG resources. Finally, we discuss open challenges, including entity disambiguation, relation normalization and evidence grading, and propose a roadmap for building open, continuously updated plant KGs that can turn "read 1,000 papers" into a single reproducible query.

Mapping protein-protein interactions (PPI) at structural resolution is essential for understanding cellular machinery. While tools like AlphaFold enable proteome-wide structural predictions, the field lacks high-throughput experimental methods to verify these predictions at scale and establish empirical thresholds for confidence metrics, such as the widely used interface predicted Template Modeling (ipTM) score. We developed LUCIA (LUminescent Cell-free Interaction Assay), a rapid biochemical screening platform bypassing traditional cloning and protein expression to validate direct binary interactions within days. Applying this to herpesviruses, clinically relevant human pathogens with large proteomes, we generated an exhaustive computational interactome of 23,215 AlphaFold-predicted dimer models across three species (HSV-1, HCMV, and KSHV), accessible via our HerpesPPIs database. Using the HSV-1 interactome as a benchmark, testing 83 high-confidence predicted dimers with LUCIA yielded 23 novel, experimentally validated interactions. By calibrating AlphaFold metrics against this direct binding data, we demonstrated that an ipTM score ≥ 0.80 identifies bona fide interactions with a positive predictive value of 77%. To demonstrate the functional power of this pipeline, we characterized a previously unknown interaction between HSV-1 UL42 and UL8, linking the viral DNA polymerase and helicase-primase complexes. Structure-guided mutations at the predicted and LUCIA-verified UL42-UL8 interface strongly reduced binding in vitro and completely abolished viral replication in cell culture. Combining the scalable LUCIA platform with computational predictions enables researchers to rapidly translate atomic-level models into validated biological mechanisms without the need to solve experimental structures.

Modern molecular analyses have revolutionized the study of microbial communities, yet DNA extraction and sequencing remain critical sources of bias. This study investigated the impact of seven different DNA extraction protocols and two 16S rRNA hypervariable regions (V1–V3 and V3–V4) on the profiling of a complex anaerobic fermentative biomass selected for medium-chain fatty acids production. Microscopic analysis established a baseline community dominated by Actinobacteria (53% ± 2%) and Firmicutes (47% ± 3%). The results demonstrate that Kit1 and Kit5 provided the highest DNA yields (up to 603 ng/μL) and the most effective recovery of these hard-to-lyse phyla, although they introduced a slight taxonomic bias toward Actinobacteria. In contrast, protocols relying on intensive chemical lysis without robust mechanical disruption (Kit4) significantly underestimated total bacterial abundance and showed the lowest purity. 16S rRNA gene sequencing revealed that the V3–V4 region provided higher alpha-diversity and a more balanced representation of the community core compared to V1–V3, which was more susceptible to extraction-related variability and overrepresented the genus Olsenella. Our multi methodological approach reveals significant biases introduced by both extraction technique and 16S rRNA gene region. This evidence highlights that protocol optimization is mandatory for achieving an accurate and comprehensive characterization of microbial ecosystems.

Precise regulation of gene expression in batch bacterial cultures is challenging because the underlying dynamics vary with cellular physiological state over time. Although cell-silicon systems enable rapid, real-time optogenetic control, disturbance rejection remains difficult in batch culture because the plant dynamics shift across growth phases, limiting the effectiveness of fixed-gain controllers designed under constant-growth assumptions. Here, we present a multiscale model-guided feedback control framework for disturbance rejection in batch E. coli cultures. Frequency-response analysis shows that the input-output dynamics of gene expression depend strongly on growth phase, revealing operating-point-dependent limits on the disturbance rejection performance of a fixed-gain PID controller. To address this limitation, we develop two growth-aware control strategies: a gain-scheduled PID (PID-GS) controller that adapts to cellular physiological state, and a gain-scheduled feedback-feedforward controller (PID-GS-FF) that further compensates for growth perturbations. We also introduce a controller evaluation framework that identifies three distinct operating regimes for targeted experimental validation. Together, these results show that accounting for growth-state-dependent dynamics is necessary for robust disturbance rejection in batch culture and provide a control-oriented framework for regulating living systems with shifting operating conditions.

Metabolic engineering often treats microbial metabolism as an inventory of metabolites, reactions, and the enzymes that catalyze them. This Perspective argues that function emerges from metabolic architecture, the connectivities that bind reactions into stable regimes shaped, among other factors, by space and time. The Japanese Metabolism movement motivates an architectural view in which the same metabolites could lead to rather different phenotypes when cells reconfigure metabolic routing subjected to environmental constraints. Natural examples, including the native cyclic glycolytic wiring of Pseudomonas putida, show how redox supply and carbon flow depend on regime-level organization and space-influenced state changes. The same principles explain why microbial engineering often fails when intermediates leak, cofactors are misallocated, or timing breaks productive hand-offs. Serine-based synthetic cycles for one-carbon assimilation expose these limits as they must couple carbon entry, redox demand, and amino acid pool control around a chiral metabolite linked to translation. The emerging picture is that future designs should make routing, insulation, compartmentalization, and metabolic segregation explicit engineering targets.

During evolution, bacteria have developed the ability to interact intimately with eukaryotic hosts. These interactions span a dynamic continuum ranging from pathogenicity to mutualism, along which bacteria can rapidly evolve and shift their lifestyles. However, the molecular mechanisms that enable bacteria to adapt to new hosts and to transition between distinct interaction modes remain poorly understood. Here, using a unique combination of two independent evolution experiments, we identified and characterized parallel adaptive mutations in spoT, which encodes the bifunctional (p)ppGpp synthetase-hydrolase. These mutations promote the adaptation of the plant pathogen Ralstonia pseudosolanacearum to two distinct plant-associated environments and two distinct lifestyles, the xylem of both susceptible and tolerant host plants as a pathogen and the root nodules of a legume as a symbiont, without compromising virulence on susceptible hosts. These mutations enhance the utilization of multiple carbon and nitrogen sources, including substrates known to be abundant in xylem sap, and increase bacterial exponential growth rate in minimal medium, suggesting reduced basal (p)ppGpp levels. Assessment of a strain deficient in SpoT synthetase activity confirmed that lowering basal (p)ppGpp levels is adaptive in both plant environments. Together, our findings reveal that fine-tuning intracellular (p)ppGpp concentrations represents an efficient strategy for optimizing bacterial adaptation to complex host-associated environments.

Despite >50 years of methods development, specific antibodies are still generated at low throughput and remain in high demand across biotechnology. Most biologics and immunoprobes are monoclonal antibodies, developed using a combination of inoculating animals with a target antigen, engineered candidate libraries, and multiple rounds of selection using phage or yeast display. Here we introduce a synthetic biology scheme to eliminate the need for nearly all of these steps, by combining Surface display on E. coli and Phage display with the microvirus ΦX174, Assisting Continuous Evolution (SurPhACE). Instead of building libraries for screening, SurPhACE runs a closed evolutionary program. A typical experiment can have 1011 mutant candidates under active selection, with complete turnover of the mutant population every 30min, or >5x1012 unique mutants per day, using less than 100mL of bacterial culture media. We demonstrate SurPhACE for optimizing a nanobody to a related epitope, and develop novel nanobodies for an arbitrary target using a minimal starting library to establish a proof of concept and identify best practices for this scalable method for generating protein binders.

Integrating semiconductors with microorganisms is attracting significant attention as a sustainable platform for solar-to-chemical conversion. This semibiological design combines the excellent light harvesting ability of semiconductor materials with intracellular biocatalytic pathways to enable efficient solar energy conversion into complex products with high selectivity. However, the effectiveness of this interdisciplinary biohybrid approach relies on a complex interfacial biotic–abiotic interaction, and it remains challenging to construct efficient and stable microbe–semiconductor systems for practical applications. In this review, we provide a systematic overview of the fundamental mechanisms behind microbe–semiconductor systems with an emphasis on interfacial electron transfer and highlight recent advancements in the assembly of biohybrids for solar-driven biosynthesis using nonphotosynthetic bacteria. First, we provide a comprehensive introduction of semibiological photosynthesis with an emphasis on extracellular electron transfer at the biotic–abiotic interfaces. Then, we discuss the engineering of biohybrid interfaces, the characterization of microbe–semiconductor interfacial electron transfer, and their deployment in solar-to-chemical conversion. We conclude by exploring the challenges in developing and optimizing biotic–abiotic interfaces as well as providing an outlook for potential future innovations. This review therefore presents the basic principles and provides guidance for the development of semibiological photosynthetic systems.

Sliding-window phylogenetic analyses of multiple sequence alignments (MSAs) generate sequences of phylogenetic trees that can reveal recombination and other sources of phylogenetic conflict, yet comparing trees across genomic windows remains challenging. Phylo-Movies is a browser-based tool, also available as a standalone desktop application, that decomposes topological differences between consecutive phylogenetic trees into interpretable subtree migrations and animates these transformations. We demonstrate its utility in two contexts: identifying recombination breakpoints in norovirus genomes, where lineages shift from polymerase-based to capsid-based clustering at the ORF1/ORF2 junction, and detecting rogue taxa that change position across bootstrap replicates. Phylo-Movies complements summary statistics such as Robinson-Foulds distances by showing which lineages move, where they move from, and which new groupings they form. Phylo-Movies is freely available at https://github.com/enesberksakalli/phylo-movies, with a norovirus demonstration video at https://vimeo.com/1162400544, the first rogue taxon example at https://vimeo.com/1162561152, and the second example at https://vimeo.com/1162563101.