The acid activity of enzymes, characterized by the minimum pH at which enzymes remain active (pHmin), is crucial for industrial applications in acidic environments. However, the rational design of acid-active enzymes remains challenging due to limited understanding of sequence-structure-activity relationships under acidic conditions. Here, we propose ACENet, a graph neural network that predicts enzyme pHmin by integrating surface features of protein structures with evolutionary representations derived from the large-scale protein language model ESM-2. ACENet achieved a Pearson correlation coefficient of 0.85 on the test dataset, significantly outperforming other deep learning baseline models and maintains stable pHmin predictions under various conditions. Even on a subset of the dataset with less than 20% homology, the PCC remains above 0.5, with an RMSE (Root mean square error) less than 1.4. ACENet also present excellent performance in the annotation of pHmin for homologous proteins and the predictive screening of minimal active pH in protein mutants. Remarkably, ACENet could identify the catalytic region as key determinants of acid activity through residue-level interpretability analysis. Overall, ACENet accelerates the development of highly efficient biocatalysts for diverse applications where acidic conditions predominate.

Infertility affects ~1 in 6 people of reproductive age and remains difficult to treat because causes are heterogeneous and diagnostics are incomplete. Recent evidence reframes the female reproductive tract as a low-biomass but biologically active microbial ecosystem. Dysbiosis, typically loss of protective Lactobacillus species (notably L. crispatus) with overgrowth of anaerobic pathobionts, is associated with implantation failure and recurrent pregnancy loss. Framing conditions such as chronic endometritis and reproducible low-Lactobacillus endometrial profiles as dysbiosis-related disorders clarifies opportunities for prevention, companion diagnostics and microbiome-directed therapies. This narrative review contrasts receptive (Lactobacillus-dominant) versus dysbiotic states and summarizes mechanisms linking microbiota to fertility: microbial metabolites (lactic acid, short-chain fatty acids) support epithelial barrier function and immune tolerance, whereas dysbiosis provokes inflammation that impairs implantation. Although observational data consistently associate Lactobacillus dominance with better outcomes, evidence quality is low-to-moderate due to retrospective designs, methodological heterogeneity, and a lack of adequately powered, diagnostic-stratified randomised trials. The review highlights precision microbial therapeutics under development, single-strain next-generation probiotics, synthetic consortia, engineered live biotherapeutics, postbiotics, targeted phage/endolysins and vaginal microbiota transplantation, and proposes a diagnostic-driven roadmap that matches microbiome endotypes and clinical contexts (e.g., preconception vs. immediate embryo transfer) to specific interventions. Regulatory and safety issues for reproductive biologics are also considered. The reproductive microbiome is a promising translational frontier but currently offers a consistent signal rather than definitive proof of benefit. To translate promise into practice requires standardised low-biomass sampling/reporting, mechanistic validation in human-relevant models and diagnostic-stratified randomised trials with staged endpoints, alongside strategies to address engraftment, formulation and regulatory pathways.

Diverse bacterial pathogens have evolved complex regulatory mechanisms to adapt to various environmental stresses during infection. The uncertainty in mRNA-protein levels in response to environmental stressors complicates our understanding of bacterial physiology and their adaptation to stressful environments. To examine this issue, we have integrated transcriptomics and proteomics data on three human bacterial pathogens Salmonella enterica Typhimurium[CE4.1][KA4.2], Yersinia pseudotuberculosis, and Staphylococcus aureus under ten infection-relevant stress conditions. We observed positive correlations between mRNA and protein levels, which were decreased under different stress conditions. Essential genes exhibited higher expression levels with lower variation across the conditions and stronger mRNA-protein correlations compared to non-essential genes, highlighting their critical role in bacterial adaptability and survival. Moreover, we identified a substantial number of genes with stress-induced non-correlating mRNA-protein levels, particularly under conditions triggering strong stress responses. Particularly this level was dramatically lowered for osmotic stress specific genes affected by impaired translational activity under osmotic stress. Our findings highlight the prevalence of non-correlating mRNA-protein levels and the potential role of post-translational modifications in modulating protein levels in response to environmental stressors during infection. This study provides a comprehensive framework for integrating transcriptomics and proteomics data and identifies potential gene products that might significantly impact the ability of diverse bacterial pathogens to adapt to hostile infection environments.

Computational protein design is often constrained by slow, complex, inaccessible, and highly sophiscated and expert-dependent workflows that hinder its transferrability and generalization power for broader applications. We present ProteinMCP, an agentic AI framework designed to accelerate and democratize protein engineering. ProteinMCP automates end-to-end scientific tasks, delivering dramatic gains in efficiency; for instance, a comprehensive protein fitness modeling workflow was completed in just 11 minutes. This performance is achieved by an AI agent that intelligently orchestrates a unified ecosystem of 38 specialized tools, made accessible through a Model-Context-Protocol (MCP). A cornerstone of the framework is an automated pipeline that converts existing software into MCP-compliant servers, ensuring the platform is both powerful and perpetually extensible. We further demonstrate its capabilities through the successful autonomous design and selection of high-affinity de novo binders and therapeutic nanobodies. By removing technical barriers, ProteinMCP has the potential to shorten the design-build-test cycle and make advanced computational protein design accessible to the broader scientific community.

Genetic code expansion (GCE) enables the site-specific incorporation of noncanonical amino acids (ncAAs) into proteins but is constrained by reliance on exogenously supplied chiral ncAAs. Achieving intracellular ncAA biosynthesis would enable more scalable and cost-effective GCE. Here, we report the continuous hypermutation and evolution of amino acid synthases that produce high levels of ncAAs inside yeast, thus supporting GCE from simple ncAA precursors. We encoded an engineered 'tyrosine synthase' (TmTyrS) on an error-prone orthogonal DNA replication system (OrthoRep) and selected variants based on ncAA biosynthesis from readily available phenol analogs and intracellular L-serine. Our selection employed orthogonal ncAA-specific aminoacyl-tRNA synthetases (aaRSs) as biosensors whereby target ncAA production leads to aminoacylation of an amber suppressor tRNA and the translation of a selectable reporter containing an amber stop codon. Our evolution successfully yielded TmTyrS variants that efficiently produced 3-iodo-, 3-bromo-, 3-chloro-, and 3-methyl-L-tyrosine, enabling amber codon-specified ncAA-dependent translation, in some cases at levels comparable to sense codon-specified natural amino acid translation. This work reduces barriers for expressing proteins containing substituted tyrosines. Moreover, because aaRSs can themselves be evolved (including with OrthoRep) for a flexible range of ncAA specificities, these results establish an end-to-end framework for evolving ncAA biosynthetic enzymes in vivo.

With an increasing global cancer burden, the regulatory function of the human microbiome and its metabolites in tumor epigenetics has garnered significant interest. Microbial metabolites are not merely passive byproducts but serve as signaling molecules and epigenetic modulators, contributing to tumor progression through multiple overlapping pathways. Short-chain fatty acids (SCFAs) such as butyrate directly inhibit histone deacetylases to reactivate tumor suppressor genes, while secondary bile acids (BAs) induce gene silencing via DNA methylation remodeling by altering the FXR/TGR5 signaling pathway. Folate and vitamin B12 serve as substrates for DNA and histone methylation through one-carbon metabolism. A complex bidirectional feedback loop exists between microbial metabolism and tumor epigenetics: reprogramming driven by hypoxia or oncogenes alters metabolite flux, generating molecules such as lactate and succinate that not only remodel chromatin and the tumor microenvironment (TME) but also selectively promote the growth of metabolically adapted microbial species, thereby reinforcing epigenetic dysregulation. Despite growing mechanistic insights, establishing causality and correlating spatiotemporal dynamics and dose responses within the highly heterogeneous TME remain major challenges. Data integration across multi-omics remains limited by methodological and computational constraints. Resolving these issues will be critical for understanding the microbe–metabolite–epigenetic axis and advancing personalized precision oncology.

Protein phosphorylation is one of the most common and versatile regulatory mechanisms in cells. Most human proteins are phosphorylated at multiple sites, giving rise to large numbers of possible phosphorylation patterns. Each phosphorylation pattern can lead to a different functional or pathological outcome. Yet, linking defined phosphorylation patterns to specific biological functions remains a major experimental challenge. In this review we describe the main strategies to study phosphorylation patterns at the protein and domain levels and highlight how they complement each other. We first discuss cellular approaches, including phosphomimetics, kinase-based assays, and genetic code expansion, which allow working in a native environment but have their significant drawbacks. We then describe in vitro methods, such as enzymatic phosphorylation and semi-synthetic phosphoproteins generated by ligation, which afford mechanistic insights but result in low yields and are difficult to scale for producing libraries. We focus on synthetic phosphopeptide libraries as tools that offer precise control over the number and position of phosphosites and are uniquely suited for systematic mapping of phosphorylation patterns. This comes at a price of not working at the protein level, but rather at the domain level. Peptide libraries are often used for preliminary identification of key phosphorylations, later studied in detail at the protein level. We conclude that ideally more than one method should be used and that these methods should not be viewed as competing but rather as complementary. A combined use of several of these approaches provides a practical toolbox for dissecting how phosphorylation patterns regulate protein behavior.

Physiologically relevant biosensors are in increasingly high demand, yet existing ones are severely limited in the number and type of biomarkers that are detected. The lack of biorecognition elements for most medically relevant biomarkers restricts the development of next generation single and continuous use monitors. Over billions of years, microbes have evolved a vast array of proteins to sense and metabolize small molecules, including those pertinent to human health. Of particular interest to us is the identification and subsequent integration of new microbial redox enzymes into electronic biosensors building off the established electrochemical technology of the continuous glucose monitor. Here we deploy genomic screening to identify analyte specific redox enzymes for biosensor development. As a proof of concept, we report the first electrochemical enzyme-based nicotine biosensor from a novel microbial enzyme, and use a variant with improved catalytic performance to enhance sensor performance. The biosensor detects nicotine over 0.4-100 μM, a range relevant to nicotine concentrations present in active smoker sweat, saliva, gastric juice, and urine. This microbial mining approach for discovering redox enzymes expands the sensing parts toolbox available over conventional antibodies and aptamers.

Deep learning methods for protein structure generation, sequence design, and structure and property prediction have created unprecedented opportunities for protein engineering and drug discovery. However, using these tools often requires navigating incompatible software environments, diverse input/output formats, and high-performance computing infrastructure, any of which may hinder adoption by primarily experimental chemical biology laboratories. Here we present BioPipelines, an open-source Python framework that allows researchers to define multi-step computational design workflows in a few lines of code. Additionally, its robust yet modular architecture provides a straightforward way to expand the toolkit with different functionalities, particularly by leveraging coding agents, with little effort. The framework currently integrates over 30 tools encompassing structure generation, sequence design, structure prediction, compound screening, and analysis. The same workflow code can be prototyped interactively in a Jupyter notebook and then submitted for production-scale runs without modification. We demonstrate applications in inverse folding, gene synthesis, de novo protein design, compound library screening, iterative binding site optimization, and fusion-protein linker optimization. We hope this framework will empower researchers, allowing them to focus on the scientific question rather than computational logistics.  https://github.com/locbp-uzh/biopipelines 

Metaribosome profiling (metaRibo-Seq) enables genome-wide measurement of translation across complex microbial communities by sequencing ribosome-protected mRNA fragments, but the short length of these footprints creates substantial nonspecific mapping against large reference genome collections, leading to spurious taxonomic and functional assignments. Here we present MOPP (Multi-Omics Processing Pipeline), a modular reference-based workflow that denoises meta-Ribo-Seq data by leveraging matched metagenomic coverage breadth to identify genomes likely to be truly present in a sample before aligning metatranslatomic and optional metatranscriptomic reads. MOPP generates taxon-by-gene count tables across genomic, transcriptional and translational layers, enabling integrated downstream analyses of microbial function. We evaluated MOPP using a defined 79-member synthetic human gut community profiled by metagenomics and metaRibo-Seq. Coverage breadth filtering markedly improved detection accuracy relative to a standard baseline workflow, with performance remaining robust across a broad intermediate threshold range and peaking at 92-95% coverage breadth. At a 92% threshold, MOPP reduced the number of distinct detected operational genomic units by 99.4% while retaining 87.8% of aligned metaRibo-Seq reads on average, and increased the F1 score from 0.02 to 0.61. Residual false positives were predominantly attributable to genomes with extremely high nucleotide similarity to true community members, whereas false negatives were enriched among low-abundance taxa, indicating that remaining errors are driven primarily by biological similarity and detection limits rather than widespread nonspecific mapping. Together, these results establish MOPP as a high-throughput workflow for robust processing of metaRibo-Seq in the context of matched metagenomics and position it as a scalable framework for integrated taxonomic and functional analysis of microbial communities across genomic, transcriptional and translational layers.