RMH

Motivation Accurate detection and quantification of bacterial strains in clinical samples is necessary to measure their colonization and persistence. Past methods to achieve this relied either on strain-specific qPCR assays, or shotgun metagenomic read mapping approaches. The resident microbial community is a major source of interference in both assays because it can contain conspecific strains bearing similarity to the focal strain(s). Results We present kSanity, a k-mer based application for the detection and quantification of targeted bacterial strains in shotgun metagenomic data. Because kSanity uses exact string matches between the reads and reference, it is less sensitive to inter- ference by conspecific strains. We test the performance of kSanity using a combination of in silico spike-in experiments, and in vivo observational data. Our results demonstrate that kSanity provides precise and accurate quantification of targeted bacterial strains, even when they are present at low sequence coverage in the metagenome. Availability and implementation kSanity is available at: https://github.com/ravel-lab/kSanity

Scientific discovery in the life sciences remains hindered by fragmented workflows, narrow-scope computational models, and inefficient links between in silico prediction and wet-lab validation. We present BioLab, a multi-agent system that integrates domain-specialized foundation models to automate end-to-end biological research. BioLab comprises eight collaborating agents, including a Planner, Reasoner, and Critic, orchestrated through a Memory Agent that enables iterative refinement via retrieval-augmented generation and a suite of 219 computational xBio-Tools spanning five biological scales (DNA, RNA, protein, cell, and chemical). These tools are built on the xTrimo Universe, a collection of 104 models derived from six foundation models (xTrimoChem, Protein, RNA, DNA, Cell, and Text), the majority of which achieve state-of-the-art (91-100% SOTA ratios) on domain benchmarks. Across standard reasoning tasks (PubMedQA, MMLU-Pro/Biology, GPQA-diamond), BioLab consistently outperformed leading large language models, including GPT-4o, Gemini-2.5, and DeepSeek-R1. Beyond benchmarks, BioLab autonomously executed a fully computational pipeline for de novo macrophage-targeting antibody design, progressing from target mining to multi-objective antibody optimization, where molecular dynamics simulations revealed structural mechanisms underlying enhanced affinity of optimized variants. Closing the computational-experimental loop, BioLab designed optimized antibodies (Pem-MOO-1, Pem-MOO-2) that achieved IC50 values of 0.01-0.016 nM, markedly surpassing the parental Pembrolizumab ( 0.027 nM) for PD-1. Functional assays confirmed enhanced pathway blockade and improved multi-parameter performance profiles. Together, these results establish BioLab as a generalizable framework for AI-native scientific discovery, demonstrating how multi-agent systems coupled with foundation models can autonomously generate, execute, and experimentally validate novel biological hypotheses.

Synthetic biology often employs heterologous enzymatic reactions to reprogram cell metabolism or otherwise introduce novel functions. However, precise control of a particular metabolic pathway can be difficult to achieve because cofactors are shared with endogenous enzymes from a common pool. Recently, the use of noncanonical cofactors (NCCs) has emerged as a promising approach to bypass this problem by isolating desired reactions without the need for a physical barrier. Metabolic pathways that exclusively utilize NCCs can be insulated from the native machinery of the host cell, allowing them to function independently of the thermodynamic constraints imposed by sharing cofactors. This perspective explores the different types of NCCs and their synthesis methods, advancements in engineering NCC-dependent enzymes, and the potential applications of NCC-utilizing cells across various areas of synthetic biology.

Prophages, viruses integrated into bacterial or archaeal genomes, can carry cargo that confers beneficial phenotypes to the host. The porcine gut microbiota constitutes a complex, dynamic, and interconnected ecosystem, yet the distribution of prophages and their unique functional characteristics within this microbial community remains poorly understood. In this study, we identified 10,742 prophage genomes through systematic screening of 7,524 prokaryotic genomes from porcine gut sources, representing both bacterial and archaeal lineages, with the distribution of integrated prophages exhibiting pronounced heterogeneity across host species. Additionally, 1.70% (183/10,742) of prophages exhibited a broad host range infectivity, while 5.07% (545/10,742) of integrated prophages enhanced prokaryotic adaptive immune capabilities by augmenting or directly providing host defense mechanisms. Notably, within tripartite phage-phage-host interactions network analysis, we observed that these prophages (n = 15) exhibit preferential acquisition of exogenous invasive phage sequences through CRISPR spacer integration mechanisms. Functional annotation revealed that prophage-encoded integrases and tail tube proteins may be critical determinants of phage host specificity. In addition, key auxiliary metabolic genes are encoded in the prophage of the pig intestinal tract, such as those promoting the synthesis of host microbiota-derived vitamin B12, encoded antibiotic resistance genes, and virulence factors that provide the host with a survival advantage. Furthermore, comparative analysis with existing viral and phage sequences uncovered a substantial reservoir of high-quality novel prophage sequences. Our findings systematically investigated the diversity of prophages in the pig gut, further characterizing their host range, functional attributes, and interactions with both host bacteria and other phages, through large-scale analysis of porcine gut microbiota genomes. This work offers new insights into the ecological roles of prophages and provides valuable genomic resources for studying prophages in this ecosystem.

Plastic pollution is a major environmental challenge, with millions of tonnes produced annually and accumulating in ecosystems, causing long-term harm. Conventional disposal methods, such as landfilling and incineration, are often inadequate, emphasising the need for sustainable solutions like bioremediation. However, the bacterial biodiversity involved in plastic biodegradation remains poorly understood. To address this gap, we present the Plastic-Microbial BioRemediation (Plastic-MBR) database, a curated multi-omics resource that integrates publicly available genetic and enzymatic data related to putative plastic-degrading microorganisms. This database supports in silico analyses of metagenomic data from plastic-contaminated environments and comparative genomics, aiming to identify microbial taxa with potential plastic-degrading functions. We validated the functionality of the Plastic-MBR database by applying it to metagenomic datasets from plastic-contaminated soil and river water, successfully identifying numerous putative plastic-degrading genes across diverse microbial taxa. These results support the use of the Plastic-MBR database as a tool to identify candidate bacteria for future experimental validation, strain isolation, and functional studies, ultimately contributing to a deeper understanding of microbial potential in plastic bioremediation. While this study focuses on database development and computational validation, future studies will be essential to confirm and translate these genomic predictions into effective bioremediation strategies.

Human Bone Morphogenetic Protein-2 (hBMP-2) serves as a critical regulator in bone and cartilage formation; however, its industrial application is hindered by its inherent tendency to form inclusion bodies in prokaryotic expression systems. To address this issue, we established a recombinant hBMP-2 (rhBMP-2) expression system using the pCold II plasmid and the SHuffle T7 strain. We explored several strategies to enhance the solubility of rhBMP-2, including coexpression with molecular chaperones, vesicle-mediated secretory expression, fusion expression with synthetic intrinsically disordered proteins (SynIDPs), and fusion expression with small-molecule peptide tags. Our results showed that coexpression with the molecular chaperone pGro7 significantly improved the solubility of rhBMP-2. Fusion with SynIDPs led to complete solubility of rhBMP-2; however, the protein was expressed exclusively in the monomeric form. Among the tested small-molecule peptide tags, GB1 was the most effective, achieving fully soluble rhBMP-2 expression. Western blot analysis confirmed the coexistence of monomeric and dimeric forms of rhBMP-2. Subsequent purification of rhBMP-2 through metal chelate chromatography resulted in an expression level of 109.7 ± 5.0 mg·L–1. In summary, we successfully demonstrated fully soluble expression of rhBMP-2 in Escherichia coli, providing a valuable foundation for its industrial-scale production.

Predicting small molecule binding sites on proteins remains a key challenge in structure-based drug discovery. While AlphaFold 3 has transformed protein structure prediction, accurate identification of functional sites such as ligand binding pockets remains a distinct and unresolved problem. Graph neural networks have emerged as promising tools for this task, but most current approaches focus on local structural features and are trained on relatively small datasets, limiting their ability to model long-range protein-ligand interactions. Here, we develop YuelPocket, a new graph neural network that addresses these limitations through an innovative global graph design featuring a virtual joint node connecting all protein residues and small molecule atoms to capture long-range interactions while maintaining linear computational complexity. Trained on the comprehensive MOAD dataset containing over 38,000 protein-small molecule complexes, YuelPocket achieves exceptional performance with AUC-ROC values of 0.85 and 0.89 on COACH420 and Holo4k datasets, respectively. docking

Metabolic engineering is rapidly evolving as a result of new advances in synthetic biology tools and automation platforms that enable high throughput strain construction, as well as the development of machine learning tools (ML) for biology. However, selecting genetic engineering targets that effectively guide the metabolic engineering process is still challenging. ML can provide predictive power for synthetic biology, but current technical limitations prevent the independent use of ML approaches without previous biological knowledge. Here, we present FluxRETAP, a simple and computationally inexpensive method that leverages the prior mechanistic knowledge embedded in genome-scale models for suggesting targets for genetic overexpression, downregulation or deletion, with the final goal of increasing the production of a desired metabolite. This method can provide a list of desirable engineering targets that can be combined with current ML pipelines. FluxRETAP captured 100% of reaction targets experimentally verified to improve Escherichia coli isoprenol production, 50% of targets that experimentally improved taxadiene production in E. coli and ∼60% of genetic targets from a verified minimal constrained cut-set in Pseudomonas putida, while providing additional high priority targets that could be tested. Overall, FluxRETAP is an efficient algorithm for identifying a prioritized list of testable genetic and reaction targets.

Genome-scale metabolic models (GEMs) are widely used in systems biology to investigate metabolism and predict perturbation responses. Automatic GEM reconstruction tools generate GEMs with different properties and predictive capacities for the same organism. Since different models can excel at different tasks, combining them can increase metabolic network certainty and enhance model performance. Here, we introduce GEMsembler, a Python package designed to compare cross-tool GEMs, track the origin of model features, and build consensus models containing any subset of the input models. GEMsembler provides comprehensive analysis functionality, including identification and visualization of biosynthesis pathways, growth assessment, and an agreement-based curation workflow. GEMsembler-curated consensus models built from four Lactiplantibacillus plantarum and Escherichia coli automatically reconstructed models outperform the gold-standard models in auxotrophy and gene essentiality predictions. Optimizing gene-protein-reaction (GPR) combinations from consensus models improves gene essentiality predictions, even in the manually curated gold-standard models. GEMsembler explains model performance by highlighting relevant metabolic pathways and GPR alternatives, informing experiments to resolve model uncertainty. Thus, GEMsembler facilitates building more accurate and biologically informed metabolic models for systems biology applications.