Antimony was originally developed by the systems biology community to solve a simple but persistent problem. SBML, the XML-based format used to exchange models between software tools, is precise and machine-friendly, but its nested tags make it cumbersome for researchers to write or inspect by hand. Antimony offered an alternative: a concise syntax in which a scientist can define species, compartments, reactions, and parameters in a few lines of text, then convert seamlessly back and forth between that text and standards-compliant SBML. The idea proved popular, and Antimony became embedded in widely used tools across the field, including the popular simulation environment Tellurium.

But software standards do not stand still. Since Antimony’s original release, SBML itself has evolved significantly, most notably with the introduction of SBML Level 3 and a growing family of extension packages that add capabilities the original specification lacked. Modeling practice has evolved too, with researchers increasingly combining kinetic reaction models with constraint-based approaches, attaching statistical uncertainty to parameters, and demanding publication-quality visualizations of their networks. Antimony’s developers recognized that the language needed a formal, comprehensive update to keep pace, and Antimony 3 is the result: a version that consolidates years of cumulative improvements and extends the language’s reach into the newest corners of the SBML ecosystem.

The headline technical advance in Antimony 3 is its expanded support for SBML Level 3 Core and several of its most important packages. The update covers the Flux Balance Constraints package, known as FBC, which underpins flux balance analysis, a technique that uses stoichiometric constraints to predict metabolic fluxes in genome-scale models. It also supports the Distributions package, which allows modelers to represent parameters as probability distributions rather than fixed numbers, capturing uncertainty in experimental measurements. The Layout and Render packages round out the list, enabling models to carry structured spatial diagrams and visual styling information alongside their mathematical content.

The practical significance of that package support is that Antimony 3 allows modelers to combine capabilities that previously required separate tools or manual file editing. A single plain-text model specification can now mix kinetic reactions with flux balance analysis, represent uncertain parameters using probability distributions, add biological context through annotations, and define publication-ready visualizations, all in one unified format. For researchers who build, share, and reuse models across laboratories and software platforms, this means fewer format conversions, fewer opportunities for error, and a single readable document that captures the full richness of a modern computational model.

Under the hood, Antimony 3 is delivered as a lightweight C/C++ library with a stable C API, a design choice that makes it easy to embed in other applications. The library ships with official bindings for three major scripting ecosystems: Python, Julia, and JavaScript compiled to WebAssembly. The WebAssembly support is particularly notable, because it means Antimony can run directly inside a web browser, opening the door to browser-based modeling tools that require no installation at all. The developers have also released a cross-platform desktop graphical interface, so researchers who prefer point-and-click workflows can use the language without touching a command line.

This multi-platform delivery reflects a broader trend in computational biology, where the same model may need to live in a Jupyter notebook, a desktop application, and a web service. By providing a stable C API at its core and consistent bindings above it, Antimony 3 ensures that a model written once behaves identically across all of these environments. The library is released as open-source software under the permissive BSD 3-Clause License, and the source code is freely available on GitHub, allowing anyone to inspect, modify, or integrate it into their own projects without licensing barriers.

For the systems biology community, the release addresses a growing pain point. As models have grown larger and more sophisticated, the burden of maintaining them in machine-oriented formats has grown with them. Genome-scale metabolic models, for example, can contain thousands of reactions, and constraint-based analyses increasingly sit alongside dynamic kinetic simulations in the same research workflow. A human-readable language that spans both worlds lowers the barrier for experimentalists who want to check what a model actually says, for students learning to build their first networks, and for reviewers who need to verify that a published model matches its description in a paper.

The emphasis on uncertainty representation also speaks to a maturing field. Point estimates for kinetic parameters are often optimistic, since experimental measurements carry noise and biological variability. By supporting the SBML Distributions package, Antimony 3 lets modelers encode that variability directly in the model definition, making uncertainty an explicit part of the shared artifact rather than a footnote in the accompanying text. Combined with annotation support for biological context, the language pushes models toward becoming self-describing documents that carry their evidence, their mathematics, and their visual presentation together.

Antimony 3 is available now at the project’s GitHub repository, and the developers describe it as a formalization of the language’s cumulative evolution rather than a departure from it. Existing users should find their models carry forward, while gaining access to the extended SBML Level 3 features. As computational models take on ever larger roles in drug discovery, metabolic engineering, and basic biological research, tools that make those models easier to write, read, and trust are likely to matter more with every passing year. With this release, the community’s answer to machine-only modeling formats has caught up with the state of the art.

Subject of Research: A human-readable modeling language for SBML Level 3 systems biology models

Article Title: Antimony 3: Extending human-readable model definitions for SBML Level 3 Core and Packages

Article References: Heydarabadipour, A., Smith, L. P., Hellerstein, J. L., & Sauro, H. M. (2026). Antimony 3: Extending human-readable model definitions for SBML Level 3 Core and Packages. PLOS Computational Biology, 22(10), e1014207. https://doi.org/10.1371/journal.pcbi.1014207

Image Credits: AI Generated

DOI: 10.1371/journal.pcbi.1014207

Keywords: Antimony, SBML Level 3, systems biology, flux balance analysis, FBC package, Distributions package, computational modeling, open-source software, biochemical networks, model interchange, WebAssembly, PLOS Computational Biology

Traditional approaches to metabolite importance ranking rely heavily on statistical significance testing or on measures of how much a compound contributes to a classification model’s decisions. The catch, as the Portuguese team points out, is that these approaches treat metabolites as independent variables. In reality, metabolism is anything but independent. Metabolites are linked through enzymatic reactions, and compounds that are close together in a biochemical network often change together when biology shifts. By ignoring this structure, conventional ranking methods can miss groups of interconnected metabolites that jointly drive the biological differences between sample classes, such as healthy versus diseased or treated versus untreated.

The new method begins with a data representation known as a Formula Difference Network, abbreviated FDiN. In this kind of network, each node represents a metabolite, and edges connect compounds whose chemical formulas differ in a meaningful and systematic way. This construction is closely related to mass-difference networks, which have become popular in the analysis of high-resolution mass spectrometry data, particularly the ultra-high-resolution measurements produced by Fourier-transform ion-cyclotron-resonance mass spectrometers. The idea is elegant in its simplicity: if two metabolites differ by a known chemical transformation, such as the addition of a specific functional group, that relationship is encoded as an edge, giving the machine learning model a scaffold of biochemical plausibility to work from.

Once the Formula Difference Network is built, it serves as the input to a predictive graph neural network, which the authors call a Formula Difference Graph Neural Network, or FDiGNN. Graph neural networks are a class of deep learning models designed to operate on graph-structured data. Instead of processing each data point in isolation, they pass information between neighbouring nodes, layer by layer, so that each metabolite’s internal representation comes to reflect not only its own measured abundance but also the abundances and properties of its biochemical neighbours. The architecture developed by the team includes a Global Attention Pooling layer, a component that weighs the contributions of different nodes when the network aggregates graph-level information to make its final class prediction, and the authors acknowledge Mário Nascimento for his help with that layer.

The genuinely novel step comes after the model has been fitted to the data. The researchers rank metabolites by their impact on the sample class prediction probabilities, in effect asking how much each node’s presence and behaviour influences the probability that the network assigns to a given class. This ranking strategy, which the authors refer to as a Probability-Impact based Network Node Importance method, abbreviated PINNI, turns the trained model itself into an interpretability tool. Rather than asking a separate statistical test which metabolites differ between groups, the approach interrogates the prediction machinery of the graph neural network and traces importance back to individual compounds within the network context.

To find out whether this strategy actually works, the team applied it to three benchmark datasets. The results showed a clear pattern: the ranking favoured subnetworks of metabolites, with connectivity acting as a driving factor in how importance was assigned. When the top-ranked compounds were examined, they turned out to be enriched in edges, meaning that the metabolites the method flagged as most important were more densely connected to one another than would be expected by chance. In practical terms, the method tends to highlight coherent biochemical neighbourhoods rather than scattered individual compounds, which is precisely the kind of output that biologists looking for altered pathways or reaction modules want to see.

The team also carried out a careful stress test of the method’s behaviour. Using datasets in which some metabolites had simulated significance artificially built into them, the researchers examined how the ranking distributed importance across the network. They found a clear bias toward assigning higher importance to nodes located in connected subgraphs. This is a double-edged finding. On one hand, it confirms that the method genuinely exploits network structure, which is its intended purpose. On the other hand, it means that isolated metabolites, however biologically relevant they might be, may be systematically under-ranked simply because they lack connections within the Formula Difference Network. Understanding this bias is essential for anyone hoping to apply the method to real experimental data, because it defines the kinds of biological signals the approach is best positioned to detect.

The authors position their strategy as an alternative to the common workflow in which important metabolites are mapped onto curated biological pathways and assessed through enrichment analysis. Pathway enrichment has served metabolomics well for years, but it depends on the completeness and quality of pathway databases, and it can struggle with compounds that are poorly annotated or absent from known pathways. A network-based ranking that emerges directly from a trained predictive model offers a complementary route: it does not require pre-existing pathway annotations, and it can, in principle, surface coordinated changes in metabolite connectivity that enrichment analysis might overlook. For researchers working with complex, high-resolution metabolomic datasets, particularly those generated in environmental, clinical or microbiome studies where many compounds remain unidentified, this independence from curated databases could prove valuable.

The work also speaks to a broader conversation in machine learning about the interpretability of graph neural networks. These models have become powerful tools across the life sciences, from drug discovery to protein structure prediction, but their decisions are notoriously difficult to explain. By building interpretability into the ranking of nodes according to their impact on prediction probabilities, the Lisbon team offers a template that could inspire similar approaches in other network-based biological applications, including gene regulatory networks, protein interaction networks and reaction networks in systems biology. The method demonstrates that a model’s internal prediction dynamics can be mined for biological insight rather than treated as an opaque black box.

The research was funded by the Fundação para a Ciência e a Tecnologia in Portugal through projects 2023.14744.PEX and 2024.06915.CPCA.A1, along with a PhD grant to Francisco Traquete, and was conducted within the European Network of Fourier-Transform Ion-Cyclotron-Resonance Mass Spectrometry Centers funded by the European Union’s Horizon 2020 programme, as well as the Portuguese Mass Spectrometry Network. Published open access under a Creative Commons Attribution licence, the study arrives at a moment when metabolomics is generating ever larger and more complex datasets, and when the demand for methods that can translate raw molecular measurements into biological understanding has never been higher. By teaching a neural network to respect the wiring diagram of metabolism and then asking it which wires matter most, the Portuguese team has added a genuinely new instrument to the metabolomics toolkit, one that promises rankings of metabolites that reflect not just statistical noise thresholds but the connected architecture of biochemistry itself.

Subject of Research: Graph neural network interpretability for ranking metabolites in metabolomics networks

Article Title: Enhancing interpretability in metabolomics: ranking metabolites by their impact on graph neural network predictions

Article References: Traquete, F., Cordeiro, C., Sousa Silva, M., & Ferreira, A. E. N. (2026). Enhancing interpretability in metabolomics: ranking metabolites by their impact on graph neural network predictions. BMC Bioinformatics. https://doi.org/10.1186/s12859-026-06633-7

Image Credits: AI Generated

DOI: 10.1186/s12859-026-06633-7

Keywords: metabolomics, graph neural networks, interpretability, Formula Difference Networks, biochemical networks, mass spectrometry, machine learning, metabolite ranking, BMC Bioinformatics, node importance, pathway analysis, computational biology