Jean-Loup Faulon

Although metabolomics has shown considerable promise for biomarker discovery, and the development of diagnostic and prognostic applications, its translation into routine clinical practice remains limited by analytical complexity, cost, throughput, and standardization challenges. These limitations underscore the need for complementary tools, particularly in resource-limited settings. In this study, we developed a workflow for the engineering and characterization of growth-coupled metabolic sensors capable of disease detection (healthy vs. infected) and outcome prediction (mild vs. severe), which we illustrated using COVID-19 as a proof-of-concept application. We first generated a biomarker-guided library of 34 candidate sensors leveraging both auxotrophic phenotypes and less stringent metabolic dependencies. We then screened the library against patient plasma pools, identifying 19 sensor candidates with diagnostic and/or prognostic potential, including 14 with prognostic potential. Lastly, a selected subset of candidates was further evaluated on a patient cohort using two newly developed analytical frameworks designed to extract additional information from bacterial growth curves. The best-performing sensors achieved a balanced accuracy of 0.88 ± 0.06 for prognostic prediction (outer-test AUC = 0.89, 5-fold cross-validation, n = 37) and 1.00 for diagnostic classification (outer-test AUC = 1.00, 5-fold cross-validation, n = 56). Collectively, these findings establish a proof of concept for translating disease-associated plasmatic metabolic signatures into low-cost, growth-coupled biosensors with diagnostic and prognostic capabilities.

Molecular and reaction fingerprints are widely used as static descriptors, but their relationship to graph-level reaction application remains unclear. We address this question in counted Extended-Connectivity Fingerprint (ECFP) space, representing reactions as signed differences between product and substrate fingerprints. We define ECFP-compatible templates, for which graph-level reaction application induces a constant fingerprint displacement, yielding a functorial mapping from graph transformations to affine translations in fingerprint space. We also introduce reaction-center ECFPs to encode the local environments required for template application and to provide a fast coordinate-wise prefilter for graph-level reaction applicability. We evaluate this framework on USPTO-50k and MetaNetX reactions across a range of ECFP radii. Both representations carry predictive information for reaction-class and enzyme-class classification, with reaction-center ECFPs contributing information not captured by reaction ECFPs alone. This framework connects reaction templates and reaction fingerprints to molecular pathway search and reaction-property prediction in fingerprint space.

Understanding genome regulation is limited by the complexity of molecular interactions in living cells. Cell-free systems provide a simplified platform for studying gene expression, but low mRNA levels have prevented RNA-seq. To address this, we develop an active learning workflow combining Bayesian optimization with automated high-throughput experimentation to systematically explore over 1.6 million buffer compositions, experimentally testing 653. We identify a “mRNA-optimized” buffer (20-fold increase in mRNA yield) and a “trade-off” buffer (13-fold increase while maintaining protein production). Using direct RNA-seq, we profile the T7 phage transcriptome in cell-free systems and compare it with a purified T7-RNAP transcription system and phage-infected bacteria. This comparative analysis reveals distinct regulatory layers: the T7-RNAP system captures promoter-strength hierarchies but lacks RNA degradation, whereas cell-free systems provide an accurate estimation of in vivo expression and reveal mRNA maturation sites. This work establishes cell-free transcriptomics as a controlled approach to study genome regulation.

Membrane transport is a fundamental biological process with profound implications for pharmacology, biotechnology, and microbiology. While computational approaches have largely adopted a protein-centric perspective to annotate transportomes, inferring transport function directly from the intrinsic properties of substrates remains a major challenge. Addressing transport at the compound level enables the systematic evaluation of whether molecules undergo active transport and by which mechanisms, independent of prior transporter annotation. Here, we introduce ChemProFlow, a comprehensive computational framework that redefines transport analysis from a substrate-centric perspective. By integrating geometric deep learning with orthology-based genomic mapping, ChemProFlow predicts molecular transportability, assigns transport mechanisms according to the Transporter Classification Database, and identifies the microorganisms encoding the corresponding transport systems. We show that this integrated pipeline enables scalable, end-to-end mapping of substrate-transporter-organism relationships, with broad applications in pharmacology for anticipating drug transport, in biotechnology for guiding strain engineering, and in microbiology for dissecting substrate utilization across diverse taxa. By capturing the chemical derminants of transportabiliy, ChemProFlow generalizes to previously unseen substrates and provides a high-throughput framework for systematic exploration of molecular transport across diverse biological contexts.

Summary: This study presents dAMN, a genome-scale neural–mechanistic hybrid model that combines neural networks with dynamic flux balance analysis to predict bacterial growth dynamics across diverse nutrient environments. Using a residual network architecture, dAMN predicts reaction fluxes and lag-phase parameters from initial medium composition, then integrates these predictions under stoichiometric constraints derived from genome-scale metabolic models. Trained on Escherichia coli and Pseudomonas putida growth datasets across combinatorial media, dAMN accurately forecasts temporal growth dynamics and generalizes to unseen media conditions, with mean R² ≥ 0.9. The model also reproduces biologically relevant behaviors including substrate depletion, acetate overflow, and diauxic shifts, while explicitly modeling lag phases usually absent from standard dFBA. Availability and implementation: The dAMN software, associated models, and datasets are available at https://github.com/brsynth/dAMN-main-release and via Zenodo DOI: 10.5281/zenodo.17908125.

Summary: This study presents dAMN, a hybrid neural–mechanistic model that integrates neural networks with genome-scale dynamic flux balance analysis (dFBA) to predict bacterial growth curves across diverse nutrient environments. dAMN uses neural networks to infer dynamic behavior from initial metabolite concentrations, while mechanistic constraints ensure stoichiometric and thermodynamic consistency based on genome scale metabolic models. dAMN is trained on E. coli and P. putida experimental growth data from media containing various combinations of sugars, amino acids, and nucleobases, and evaluated on two test sets: one for forecasting over time and another for predicting growth dynamics on unseen media. dAMN achieved high predictive power (R ² ≥ 0.9), successfully reproducing growth and substrate depletion dynamics including acetate overflow and glucose-acetate consumption shift for E. coli . An interesting innovation of dAMN is the treatment of the lag phase, enabling realistic adaptation dynamics absent from standard dFBA models. dAMN stands out for its ability to generalize across combinatorial nutrient inputs and produce full growth-curve predictions from minimal input data. Availability and implementation: The dAMN software, along with the associated models and data, is available at https://github.com/brsynth/dAMN-main-release and via DOI 10.5281/zenodo.17908125

RetroRules (https://retrorules.org) is an open resource of reaction templates, which are generic reaction representations that describe the atomic transformations underlying biochemical reactions. These templates are key to supporting metabolic pathway discovery, reaction prediction, and enzyme engineering. The 2026 release updates biochemical sources (MetaNetX, Rhea) and newly integrates organic chemistry reactions (USPTO), extending the scope of the database beyond enzymatic systems. The template encoding has been simplified by using implicit hydrogens and minimal atomic descriptors, resulting in faster and more compact representations. Radius range now spans 0–10, allowing finer control of reaction specificity. In addition, mass-imbalanced reactions are included, expanding the coverage of biochemically relevant transformations. Reaction mapping now relies on the transformer-based tool RXNMapper, improving accuracy. RetroRules 2026 comprises 1 174 216 templates derived from 92 698 reactions, covering 5796 fourth-level EC numbers. A redesigned website, updated Online Template Generator, and OpenAPI-defined API enable multi-criteria exploration (dataset, radius, and EC number), visualization, and data export in multiple formats. Sequence annotations from UniProt were refreshed and summarized as a normalized sequence-support score for ranking. Together, these updates establish RetroRules as a cross-domain resource bridging biochemistry and organic chemistry, offering broader coverage, controllable specificity, and enhanced usability for high-throughput pathway design, reaction prediction, and enzyme engineering.

Reverse engineering in molecular design aims to identify optimal structures based on activities, or properties, computed through molecular descriptors like fingerprints. This task is known to be particularly difficult for the widely used Extended-Connectivity Fingerprints (ECFPs), due to significant loss of structural information during vectorization. While recent artificial intelligence-based works have raised awareness about the privacy risks associated with ECFP-based data sharing, we contribute a more conclusive demonstration by introducing a deterministic algorithm that reconstructs molecular structures from ECFPs. Using MetaNetX and eMolecules as databases of natural compounds and commercially available chemicals, the deterministic algorithm benchmarks a Transformer-based generative model trained to predict SMILES from ECFPs. The generative model achieves a top-ranked retrieval accuracy of 95.64% but struggles with exhaustive enumeration. Additionally, applying the deterministic method to a drug dataset reveals its potential for de novo drug design, as many of the reverse-engineered structures are found to be patented or supported by bioassay data. Graphical Abstract