Bacteria and phages have coexisted for billions of years engaging in continuous evolutionary arms races that drive reciprocal adaptations and resistance mechanisms. Among the diverse antiviral strategies developed by bacteria, modification or masking phage receptors as well as their physical removal via extracellular vesicles are the first line of defense. These vesicles play a pivotal role in bacterial survival by mitigating the effects of various environmental threats, including predation by bacteriophages. The secretion of extracellular vesicles represents a highly conserved evolutionary trait observed across all domains of life. Bacterial extracellular vesicles (BEVs) are generated by a wide variety of Gram (+), Gram (−), and atypical bacteria, occurring under both natural and stress conditions, including phage infection. This review addresses the multifaceted role of BEVs in modulating bacteria–phage interactions, considering the interplay from both bacterial and phage perspectives. We focus on the dual function of BEVs as both defensive agents that inhibit phage infection and as potential facilitators that may inadvertently enhance bacterial susceptibility to phages. Furthermore, we discuss how bacteriophages can influence BEV production, affecting both the quantity and molecular composition of vesicles. Finally, we provide an overview of the ecological relevance and efficacy of BEV–phage interplay across diverse environments and microbial ecosystems. omv

Antimicrobial resistance and the emergence of multi-drug-resistant pathogens necessitate holistic care. Novel antimicrobial drug discovery involves an in-depth assessment of quorum sensing (QS) signaling and cell-cell communication. Bacteria regulate their metabolism to cope with complex host-environmental changes through QS signaling. Previous studies suggest that the social behaviors of bacteria include biofilm formation, virulence, and drug resistance mediated by QS. Over several decades, autoinducer receptors, signaling pathways, and the regulatory networks that control gene expression have corroborated QS signaling molecules. Multiple QS systems and chemical structural diversity signaling molecules have been sporadically reported, but there is no comprehensive review of these findings. This review systematically and comprehensively summarizes the paper, addressing bacterial QS-based cell-cell communication and the potential mechanisms of QS systems in bacterial drug resistance. Furthermore, a status quo update has been established for novel QS systems in the realms of pathogenicity, poly-microbial interactions, and/or antimicrobial resistance patterns. Nevertheless, the applications of QS systems would invoke newer incorporations for futuristic research values. Thus, the complexity and evolutionary insights of QS systems, quorum sensing signaling molecules (QSSMs), and regulatory mechanisms need a “cloud” based repurposing for the design of novel agents in antimicrobial resistance (AMR) and communication.

Artificial intelligence models are transforming protein research, enabling advances in areas ranging from structure prediction to the design of functional enzymes. However, these models operate as black boxes, and their underlying working principles remain unclear. Here we survey emerging applications of explainable artificial intelligence (XAI) to protein language models and describe the potential of XAI in protein research. We organize existing work around four points in a typical modelling pipeline: the data used for training; the user-provided inputs; the internal model architecture; and input–output relationships. Across these contexts, we highlight methods and applications of XAI. In addition, from published studies we distil five potential roles for XAI in protein research: Evaluator, Multitasker, Engineer, Coach and Teacher, with the evaluator role being the only one widely adopted so far. While our analysis focuses on protein language models, our categorization is broadly applicable to any other architecture. We conclude by highlighting critical areas of application for the future and outlining a path to advance the interpretability of protein artificial intelligence. Hunklinger and Ferruz provide an overview of explainable artificial intelligence methods for protein language models.

Nuclear localization signals (NLSs) and nuclear export signals (NESs) mediate nucleocytoplasmic transport of proteins through the nuclear pore complex and are essential determinants of protein function. However, their short and degenerate sequence patterns frequently lead to high false-positive rates in sequence-based prediction methods, as similar motifs occur widely in proteins without mediating nuclear transport. Here, we present SPSignal, a webserver for improved identification of NLS and NES motifs by integrating sequence-based predictions with structural features. SPSignal combines curated datasets of experimentally validated signals with analyses of solvent accessibility, intrinsic disorder, and structural context derived from experimental or predicted protein structures. Using these features, interpretable machine-learning models based on the RuleFit algorithm prioritize candidate motifs that are structurally exposed and therefore more likely to be functional. The web server integrates sequence predictors with structure-informed analyses in a unified workflow that accepts protein sequences or structures as input and provides interactive visualization of predicted signals within their three-dimensional context. SPSignal assigns confidence scores to candidate motifs and allows users to explore their spatial distribution along protein sequences and structures. Application to proteins with validated localization signals shows that SPSignal improves prediction accuracy by reducing false positives without compromising sensitivity. SPSignal is available at https://sps.cragenomica.es.

Bacterial genomes contain thousands of biosynthetic gene clusters (BGCs) responsible for the production of structurally diverse natural products with applications in medicine, agriculture, and biotechnology. Expression of these BGCs is tightly regulated by transcription factors (TFs) responding to environmental cues, yet predicting which TFs regulate specific BGCs remains challenging. In particular, TF binding sites (TFBSs) within BGCs often diverge from canonical motifs, limiting the effectiveness of standard motif-scanning approaches and hindering systematic exploration of BGC regulation. Here, we present COMMBAT (COnditions for Microbial Metabolite Biosynthesis Activated Transcription), a framework for large-scale prediction of TF–BGC regulatory interactions across bacterial genomes. COMMBAT integrates motif matching with genomic context and gene function information to predict functional TFBSs. The COMMBAT web platform (https://www.commbat.uliege.be) enables users to (i) identify BGCs potentially regulated by a given TF, and (ii) predict candidate TFs that control a specific BGC. With over 4000 TF position weight matrices from four public repositories and more than 400 000 BGCs from MIBiG and antiSMASH DB, COMMBAT provides a scalable resource to predict regulatory inputs and guide/prioritize culture conditions and genetic engineering strategies for natural product discovery.

Antimicrobial resistance (AMR) poses an escalating threat to global health, as multidrug-resistant pathogens undermine therapeutic efficacy and surveillance systems. Although whole-genome sequencing and phenotypic drug susceptibility testing have strengthened resistome profiling, translating multi-omics data into reliable, clinically deployable intelligence remains computationally fragmented. Following PRISMA 2020 guidelines, we systematically reviewed 156 records published between 2016 and 2025, of which 93 studies were included in the final synthesis. We organize AMR modeling into three methodological strata: (i) classical and interpretable machine-learning frameworks, (ii) structural and deep genomic architectures, and (iii) transformer-based and applied large language model systems that integrate genomic, clinical, and epidemiological signals. Across studies, we identify four converging integrative directions: embedding-level multimodal fusion, knowledge-graph-guided causal reasoning, evolutionary and temporal forecasting, and agentic artificial intelligence systems enabling autonomous, evidence-grounded workflows. Comparative analysis reveals substantial heterogeneity in dataset scale, frequent reliance on internal validation, limited assessment of cross-site robustness, and vulnerability to distribution shift, particularly for minority resistance phenotypes. We argue that future AMR intelligence must integrate uncertainty-aware modeling, standardized validation protocols, and FAIR-compliant infrastructures to transition from static genomic classification toward interpretable, temporally adaptive, and clinically actionable decision systems within One Health surveillance ecosystems.

The microbial valorization of glycerol, a major biodiesel byproduct, into high-value chemicals remains challenging at industrially competitive titers. Here we engineered Corynebacterium glutamicum for high-level 1,3-propanediol (1,3-PDO) production, validating each step via fed-batch fermentation. First, C. glutamicum ATCC 13032 was metabolically engineered to produce 138 g l−1 1,3-PDO from glucose–glycerol and 100.9 g l−1 from glycerol alone. Key engineering strategies included establishing glycerol uptake and 1,3-PDO biosynthetic pathways, minimizing byproducts and optimizing fed-batch fermentation. We then transferred these strategies to a newly isolated strain, C. glutamicum SC97. Further engineering, including antibiotic-free plasmid addiction system and sucCD overexpression, enabled 141.5 g l−1 1,3-PDO at 2.95 g l−1 h−1 without antibiotics. Scalability was demonstrated at 30-l and 300-l pilot-scale fermentations, reaching 120.2 g l−1 and 127.8 g l−1 of 1,3-PDO, respectively. Techno-economic and life-cycle assessments support industrial feasibility and environmental impact, providing a robust blueprint for sustainable microbial 1,3-PDO production at scale. This study reports on engineered Corynebacterium glutamicum capable of converting glycerol, a major biodiesel byproduct, into 1,3-propanediol at industrially relevant titers. Through integrated strain design, optimization of fed-batch conditions, implementation of antibiotic-free systems and validation in pilot-scale fermentation, this study establishes a scalable blueprint for sustainable 1,3-propanediol biomanufacturing.

The rising rate of drug-resistant fungal infections and the emergence of intrinsically resistant pathogens pose growing clinical challenges. Because fungi are closely related to mammals, developing antifungals without toxic off-target effects is difficult. Targeted gene repression can model drug-mediated inhibition and reveal gene dosage sensitivity, but traditional approaches in the fungal pathogen Candida albicans are labor intensive and low throughput. Here we adapt pooled CRISPR interference (CRISPRi) screening in C. albicans to enable large-scale functional genomic analysis. We assess repression sensitivity of 130 essential genes conserved in fungi without close homologues in humans and identify highly dosage-sensitive genes across multiple pathways. Screening across ten environmental conditions reveals environment-dependent effects on gene sensitivity. Extending these experiments to two drug-resistant clinical isolates shows that many fitness defects are conserved across genetic backgrounds. Thus, CRISPRi pooled screening enables rapid, large-scale functional genomics across diverse genetic backgrounds in C. albicans. This study presents a pooled CRISPRi screening approach in Candida albicans to investigate essential genes at scale, revealing environment- and strain-dependent gene sensitivities and conserved vulnerabilities relevant to antifungal drug discovery.

Dysregulated extracellular proteolytic activity is a prominent hallmark of cancer and can thus be exploited for tumor detection and therapeutic development. However, the discovery of tumor-responsive probes has been hindered by the lack of methods to directly screen proteolytic events in specific tissue samples. Here we report PSurf, a platform that enables the identification of tissue-specific protease sensors with tissue specimens. Through differential selection of tumor-specific sequences over healthy tissue, PSurf identifies context-specific tumor-activated probes that precisely distinguish metastatic lesions in lung tissue slices. Using these substrates, we engineered nanobody-targeted biosensors that release urinary reporters upon tumor-specific cleavage in vivo, enabling precise non-invasive tumor detection in a mouse lung metastasis model. PSurf provides a foundation for developing conditionally activated agents through tissue-specific activity mapping and probe discovery. Dysregulated extracellular proteolytic activity is a hallmark of cancer that can be exploited for tumor detection and therapeutic development. The authors developed a platform that screens proteolytic events in tissue samples and differentially selects tumor-specific sequences, enabling the design of biosensors detecting tumor-specific cleavage of peptide substrates in tissue samples and in vivo.

Many routine genomics tasks in molecular biology still depend on heterogeneous and proprietary software tools that hinder accessibility, reproducibility, and seamless laboratory use. We present GEAR (https://www.gear-genomics.com/), a unified, web-based genomics framework that provides a collection of lightweight, interactive applications for common molecular biology and genomics analyses directly in the browser. GEAR requires no software installation, user registration, or licensing and is designed for rapid, intuitive use without prior bioinformatics expertise. The platform integrates robust, well-established backend algorithms with modern web technologies to support a diverse set of tasks, including Sanger chromatogram visualization, alignment and variant detection, primer and padlock probe design, in-silico PCR, qPCR analysis, barcode generation and inspection, sequencing quality control, DNA manipulation, and sequence alignment visualization. In summary, GEAR serves as an integrated, open, extendible, and user-friendly genomics web server that consolidates a diverse set of tools within a single coherent framework with all code free and open-source (https://github.com/gear-genomics). By emphasizing interactivity, reproducibility, and ease of use, GEAR aims to support both routine laboratory tasks and exploratory genomic analyses across a broad range of research applications.

The increasing growth in the volume of biomolecular data has introduced significant challenges for extracting meaningful molecular-level insights, particularly in predicting interactions between biological sequences such as DNA, RNA, and proteins. These interactions are fundamental to complex biological processes, including gene regulation and immune response. Artificial Intelligence (AI) has played a major role in advancing discoveries in this field, enabling the identification of novel interactions, as demonstrated by various predictive modeling studies. Despite the growing number of scientific publications in this domain, accessibility to practical computational tools has not progressed at the same pace. Existing studies differ substantially in availability: some provide only methodological descriptions, others release source code exclusively for experimental reproducibility, and only a limited number deliver fully automated solutions ready for broad use. Given this context, this paper investigates state-of-the-art studies in biological sequence interaction prediction, emphasizing the public accessibility and usability of available tools, especially for researchers who are not experts in AI or computational methods. We compile and discuss the input requirements of current tools, along with the types of outputs they generate, enabling users to better understand the scenarios in which each solution can be effectively applied. Furthermore, we analyze accessibility-related aspects to support informed selection of tools according to user expertise, ranging from web-based servers with pretrained models that require minimal computational skills to fully end-to-end frameworks capable of training new models on user-defined datasets, though often lacking user-friendly interfaces.