The scale of the problem is staggering. Regulatory agencies typically evaluate the parent active ingredients of pesticides and medicines, but environmental monitoring studies repeatedly show that degradation products can persist longer, travel further, and sometimes exceed the concentrations of the original compounds in groundwater and surface water. Classical examples include the transformation products of the herbicide atrazine, which can linger in aquifers for decades, and the metabolites of veterinary antibiotics that accumulate in soils and drive antimicrobial resistance. For pharmaceuticals, human metabolites excreted after drug treatment pass largely unchanged through wastewater treatment plants and re-enter drinking water sources. Traditionally, assessing each of these molecules would require years of laboratory animal testing, analytical method development, and environmental fate studies—an approach that is simply too slow and expensive for the thousands of metabolites already known and the many more still being discovered.

The study, authored by researchers working at the intersection of cheminformatics and environmental toxicology, argues that the moment has come to shift toxicity assessment from a purely experimental discipline to a hybrid one in which computational predictions guide and prioritize experimental work. The authors assemble and analyze the principal computational strategies now available for this task, ranging from classical quantitative structure–activity relationship models, which relate molecular descriptors such as lipophilicity, electronic properties, and topological indices to biological endpoints, to modern deep learning architectures capable of learning toxicological patterns directly from chemical structures. Graph neural networks, in particular, have emerged as a powerful tool because they represent molecules as networks of atoms and bonds, allowing algorithms to learn which substructures are associated with carcinogenicity, mutagenicity, endocrine disruption, or aquatic toxicity.

A central theme of the article is the diversity of toxicity endpoints that must be considered when evaluating agrochemical and pharmaceutical metabolites. Acute toxicity matters, but chronic effects are often more insidious and harder to detect. The authors discuss computational approaches for predicting genotoxicity, developmental and reproductive toxicity, endocrine receptor binding, skin sensitization, and ecotoxicological outcomes in organisms such as fish, daphnids, and algae. Machine learning classifiers trained on curated datasets from sources such as regulatory toxicity databases and public repositories can flag metabolites that share structural alerts with known toxins. At the same time, the article stresses that no single model suffices: predictions must be triangulated across multiple endpoints, and read-across approaches—inferring the properties of an unknown metabolite from structurally similar chemicals with experimental data—remain an essential complement when training data are sparse.

One of the most technically interesting contributions is the discussion of how metabolite-specific challenges strain conventional models. Machine learning models are typically trained on parent compounds, and metabolites often sit in chemical regions poorly covered by training data—a problem the community calls applicability domain violation. Degradation products may be smaller, more polar, or carry unusual functional groups introduced by photolysis or biotransformation, making naively applied predictions unreliable. The article describes strategies to address this, including data augmentation with transformation product libraries, uncertainty quantification methods that indicate when a model is extrapolating beyond its competence, and ensemble approaches that combine the votes of many models to produce calibrated confidence estimates. In silico metabolism simulators, which predict the likely transformation pathways of parent compounds, extend the framework further by generating lists of candidate metabolites that can then be prioritized for toxicity screening even before they are detected in the environment.

Molecular docking and other structure-based computational techniques also feature prominently. For endocrine disruption and receptor-mediated toxicity, the authors explain how metabolites can be docked into the binding pockets of estrogen, androgen, and thyroid hormone receptors, and how molecular dynamics simulations refine these snapshots into dynamic pictures of binding stability. Combining docking scores with machine learning-derived descriptors produces hybrid models that outperform either approach alone. Quantum chemical calculations, including density functional theory, add another layer of mechanistic insight by quantifying the reactivity of metabolites toward biological nucleophiles such as DNA bases and proteins, which is central to understanding genotoxic and sensitizing potential. This multi-layered computational pipeline mirrors how a toxicologist thinks, but executes in minutes what would otherwise take months of bench work.

The article does not shy away from the limitations of the field. Data quality remains the Achilles’ heel of machine learning toxicity prediction: experimental datasets are heterogeneous, drawn from different protocols, species, and laboratories, and labeling inconsistencies can propagate silently into models. Class imbalance is another persistent difficulty, since verified toxic molecules are far outnumbered by benign ones, biasing classifiers toward false negatives—arguably the most dangerous kind of error in safety assessment. Interpretability is equally critical, because a black-box prediction that a metabolite is carcinogenic carries little regulatory weight unless the model can explain which structural features drove the conclusion. The authors highlight techniques such as attention-based attribution, SHAP-style feature importance analysis, and substructure highlighting as ways to make deep models transparent enough for use in regulatory contexts, and they call for standardized benchmarks of metabolite toxicity data to allow fair comparison of competing methods.

Perhaps the most consequential part of the study is its forward-looking roadmap for regulatory integration. The authors envision a tiered workflow in which computational screening of transformation products happens early, identifying high-risk metabolites for targeted analytical monitoring and focused experimental testing. This mirrors emerging regulatory thinking in Europe and North America, where in silico evidence is increasingly accepted as supporting information in risk assessment frameworks for plant protection products and pharmaceuticals in the environment. Artificial intelligence could dramatically shorten the time between the introduction of a new pesticide or drug and the characterization of its full transformation footprint, enabling safer molecular design from the outset. The concept of benign-by-design chemistry—engineering molecules that degrade into genuinely harmless products—becomes tractable when toxicity predictions are fast, cheap, and accurate enough to guide synthesis decisions in real time.

For the public, the stakes are concrete: cleaner drinking water, safer food residues, and reduced chemical pressure on already stressed ecosystems. For industry, the framework offers a path to avoid costly late-stage failures and regulatory surprises. And for the scientific community, the article consolidates a rapidly moving field, connecting the dots between cheminformatics, deep learning, environmental analytical chemistry, and toxicology in a way that few previous reviews have attempted. As artificial intelligence continues to penetrate every corner of chemistry, this work makes a persuasive case that the transformation products of our agrochemicals and medicines—long the invisible residues of modern life—can finally be brought into the light of systematic, predictive safety science, before the next hidden toxin makes headlines for the wrong reasons.

Subject of Research: AI-driven computational prediction of the toxicity of agrochemical and pharmaceutical metabolites

Article Title: Toxicity assessment of agrochemical and pharmaceutical metabolites using artificial intelligence machine learning and computational approaches

Article References: Jana, S., Matore, B. W., Murmu, A., Gawande, P., Shukla, D., Kumar, A., Singh, J., & Roy, P. P. (2026). Toxicity assessment of agrochemical and pharmaceutical metabolites using artificial intelligence machine learning and computational approaches. Discover Chemistry, 3(1), Article 533. https://doi.org/10.1007/s44371-026-00993-y

Image Credits: AI Generated

DOI: 10.1007/s44371-026-00993-y

Keywords: artificial intelligence, machine learning, toxicity prediction, agrochemicals, pharmaceuticals, metabolites, computational chemistry, QSAR, deep learning, environmental toxicology, cheminformatics, risk assessment

The system, called HD-SKRG, short for hierarchical dual-view and structure-knowledge relation graph enhancement network, was developed by Luyi Jia, Mingyang Wang, Zeming Wang of Northeast Forestry University in Harbin, China, together with Xianjie Wang of the Harbin Institute of Technology. Their work addresses a problem known as few-shot molecular property prediction: the challenge of adapting a model to a brand-new property task using only a handful of labeled molecules. In drug discovery, where a promising compound may be tested against just a few biological targets before resources run out, this is not an academic concern but a practical constraint on how quickly new medicines can be identified.

The researchers identified two fundamental weaknesses in existing approaches. First, the molecular representations themselves are often insufficient. Most graph neural networks treat molecules as collections of atoms connected by bonds, but this flat view misses the hierarchical reality of chemistry, where functional groups such as hydroxyls, amines, or aromatic rings carry semantic meaning that individual atoms do not capture alone. Second, the way models relate molecules to one another within a prediction task tends to be biased. When relations between molecules are built purely on structural similarity, the model can be misled, because two compounds may look alike on a two-dimensional scaffold yet behave very differently in a biological context, particularly when labeled examples are too scarce to correct such errors.

HD-SKRG attacks the first problem with a dual-view representation strategy. The model builds two complementary graphs for every molecule: an atom-level graph that captures fine-grained connectivity, and a functional-group-level graph that groups atoms into chemically meaningful motifs. Crucially, the two views are not built in isolation. The architecture injects elemental knowledge, information about the intrinsic properties of chemical elements, directly into the atom representations, and then transfers local atomic information upward into the functional-group representations. This hierarchical flow means that what a functional group knows is grounded in what its constituent atoms encode, while the group-level view provides context that a single atom cannot supply.

To distill these two views into a single molecular fingerprint, the researchers introduced a frequency-aware aggregation module. Rather than treating all structural patterns equally, the module weighs information according to how frequently particular substructures appear, producing what the authors describe as molecular-level knowledge representations. The intuition is that rare structural features may be highly informative for unusual properties, while common motifs provide a stable backbone of chemical meaning, and the aggregation process balances these contributions automatically rather than by hand-tuned rules.

The second problem, biased relation construction, is addressed through a pair of relation graphs that govern how information flows between molecules during a prediction task. The structure relation graph, built from molecular similarity, serves as the main pathway for feature propagation, allowing labeled molecules to inform unlabeled ones through learned message passing. The knowledge relation graph plays a complementary role: it supplies semantically related neighbors that structural similarity alone would miss, and it refines the weights on the relation edges. By letting semantic knowledge modulate a purely structural graph, the design reduces the graph-construction bias that plagues methods relying on structural similarity as their only signal of molecular relatedness.

Training proceeds in two stages that mirror how the model is ultimately used. The dual-view encoders are first pretrained with cross-view contrastive learning, a technique in which the model learns by aligning the atom-level and functional-group-level views of the same molecule while distinguishing them from views of different molecules. This pretraining draws on the large ZINC15 chemical database, giving the encoders a broad foundation in molecular structure before they ever see a specific prediction task. The full model is then meta-trained under the model-agnostic meta-learning framework, or MAML, which optimizes the network’s parameters so that they can rapidly adapt to new tasks from very few examples, a strategy borrowed from the broader few-shot learning literature.

The empirical evaluation covered four widely used benchmarks drawn from the MoleculeNet repository: Tox21, which tests prediction of nuclear receptor and stress response pathways; SIDER, a database of drug side effects; MUV, a virtual screening benchmark designed to be maximally unbiased; and ToxCast, a large toxicology dataset. The authors tested the model under both 1-shot and 10-shot conditions, meaning the model had access to either one or ten labeled examples per class. Across the eight resulting settings, HD-SKRG achieved the best results in five and the second-best in the remaining three, a consistent performance profile that the authors argue reflects the robustness of the dual-view representation and the debiased relation graphs rather than luck on any single benchmark.

Ablation studies, in which individual components of the architecture are removed one at a time, confirmed that each module contributes measurably. Removing the elemental knowledge injection, the frequency-aware aggregation, or the knowledge relation graph each degraded performance, indicating that the gains do not come from a single clever trick but from the interplay of hierarchical representation, knowledge enrichment, and relation refinement. The datasets themselves are publicly available, and the pretraining data can be downloaded from an existing motif-based pretraining repository, which should make the approach reproducible and testable by other groups.

The broader significance of the work lies in what it says about the future of computational chemistry under data scarcity. Large language models and foundation models have dominated headlines by leveraging enormous corpora, but in molecular science the labeled data that matters most, confirmed toxicity, verified side effects, measured bioactivity, remains stubbornly scarce. Architectures like HD-SKRG suggest a different path: rather than waiting for bigger datasets, encode more chemistry into the model itself, through hierarchical structure, elemental knowledge, and semantically informed relations, and let meta-learning handle the adaptation to new problems. If such methods continue to mature, the early stages of drug discovery could become dramatically cheaper, allowing researchers to triage candidate compounds with confidence even when experimental data is a luxury. For a field where a single failed late-stage trial can cost hundreds of millions of dollars, teaching machines to reason from a single example may prove one of the most consequential bets in modern AI-driven chemistry.

Subject of Research: Few-shot molecular property prediction using a hierarchical dual-view and structure-knowledge relation graph neural network

Article Title: HD-SKRG: a hierarchical dual-view and structure-knowledge relation graph enhancement network for few-shot molecular property prediction

Article References: Jia, L., Wang, M., Wang, Z., & Wang, X. (2026). HD-SKRG: a hierarchical dual-view and structure-knowledge relation graph enhancement network for few-shot molecular property prediction. Molecular Diversity. https://doi.org/10.1007/s11030-026-11719-8

Image Credits: AI Generated

DOI: 10.1007/s11030-026-11719-8

Keywords: few-shot learning, molecular property prediction, graph neural networks, drug discovery, meta-learning, contrastive learning, molecular representation, toxicity prediction, relation graphs, MAML, chemical knowledge, MoleculeNet

The research, led by Peiling Han, Jingwen Chen, Yongle Zhu, Jingyuan Yang, and Xuehua Li of Dalian University of Technology in China, together with Willie J. G. M. Peijnenburg of the Dutch National Institute for Public Health and the Environment and Leiden University, introduces the intelligent high-throughput multi-species physiologically based toxicokinetic model, abbreviated HM-PBTK. At its core, the model addresses a stubborn technical problem. Physiologically based toxicokinetic models have long been a cornerstone of toxicology because they simulate how a chemical enters an organism, distributes through its tissues, is metabolized, and is ultimately eliminated—the so-called ADME processes. But these models depend on dozens of species-specific parameters, such as blood flow rates, tissue volumes, tissue composition, and metabolic clearance rates, that are known for only a handful of laboratory species like zebrafish and rainbow trout. For the vast majority of the more than 30,000 fish species on Earth, such data simply do not exist.

The Chinese-led team’s solution was to build machine-learning models capable of predicting these physiological and biochemical parameters across species that have never been tested. Drawing on a multimodal dataset that integrates biological traits, phylogenetic information, environmental context, and chemical properties, the researchers trained AI systems to estimate the parameters that a toxicokinetic model needs, from cardiac output and oxygen consumption to in vitro intrinsic clearance rates. The resulting framework covers 151 freshwater and marine fish species, spanning a phylogenetic and ecological range that conventional modeling approaches could never approach. In effect, the team taught an algorithm to infer the internal plumbing and biochemistry of fish it has never seen, using patterns extracted from species that have been studied.

Once the AI-predicted parameters are plugged into the toxicokinetic model, the system can quantify how much of a given chemical accumulates in specific tissues—the liver, the gills, the blood, the gonads—under realistic exposure scenarios. This tissue-specific internal dose is the quantity that matters for toxicology, because the concentration of a chemical at its site of action, not merely its concentration in the surrounding water, determines whether harm occurs. The researchers validated the model against an extensive literature-derived dataset of 703 internal exposure measurements spanning multiple species and chemicals, providing an unusually rigorous test of the framework’s predictive power.

The performance results are striking. In a case study focused on oestrogenic effects—the induction of vitellogenin, an egg-yolk precursor protein that serves as a classic biomarker of endocrine disruption in fish—the model’s quantitative in vitro to in vivo extrapolation, or QIVIVE, placed 85 percent of toxicity predictions within fivefold of the corresponding experimental observations. In a field where predictions spanning orders of magnitude are common, and where animal testing for every species-chemical combination is impossible, a fivefold window across such a diverse species set represents a substantial gain in reliability. The case study is also ecologically pointed: synthetic oestrogens from wastewater treatment effluent have been shown in earlier work, including a landmark 2007 study in the Proceedings of the National Academy of Sciences, to collapse entire fish populations in experimental lakes.

The implications extend well beyond endocrine disruption. The researchers demonstrated the model’s application to chemicals that are frequently detected in the environment, simulating absorption, distribution, metabolism, and excretion in both freshwater and marine species under real exposure conditions. The framework handles both neutral and ionizable chemicals, a critical distinction because many pharmaceuticals and emerging contaminants carry electrical charges that dramatically alter how they move through biological membranes and how they partition into tissues. Earlier multispecies toxicokinetic efforts, including those by Brinkmann and colleagues and Mangold-Döring and colleagues in Environmental Science & Technology, laid important groundwork but were limited in species coverage and chemical scope; the new AI-driven approach scales the concept by orders of magnitude.

Recognizing that a powerful model is only as useful as it is accessible, the team also built a user-friendly web platform designed to make the technology available to risk assessors, regulators, and researchers who are not modeling specialists. The platform allows users to conduct comprehensive exposure-toxicity predictions for chemicals simply by setting up an exposure scenario—specifying the chemical, the environmental concentrations, and the species or conditions of interest—without writing code or manually parameterizing differential equations. This kind of operational tooling matters because regulatory ecological risk assessment, governed by frameworks such as the European Union’s chemicals legislation and the United Nations’ post-2020 global biodiversity framework, is under mounting pressure to evaluate thousands of substances for which experimental data are sparse.

The timing of the work is significant for reasons that go beyond computational novelty. Chemical pollution is now recognized alongside climate change and habitat destruction as a primary driver of global biodiversity decline. The landmark tire-rubber-derived chemical 6PPD-quinone, which was shown in 2021 in Science to kill coho salmon within hours of stormwater runoff exposure, illustrated how a single ubiquitous contaminant can devastate a wild fish population before anyone knew it was toxic. Meanwhile, the demand for animal testing is under ethical and practical strain: European statistics show millions of fish used in regulatory toxicity testing, and the scientific community has embraced replacement, reduction, and refinement principles. A validated computational framework that predicts internal exposure and toxicity without live animals directly serves those goals, offering regulators a route to screening that is faster, cheaper, and humane.

The study’s technical architecture reflects broader trends in computational toxicology, where machine learning has begun to infuse every layer of physiologically based pharmacokinetic and toxicokinetic modeling. Prior work had demonstrated multimodal deep learning for predicting drug clearance in humans and machine-learning models for tissue-to-blood partition coefficients, but the translation of these techniques to ecological species—where data are scarcer, species diversity is vastly greater, and environmental variables such as temperature and salinity complicate parameterization—required the kind of systematic data assembly and model integration this team undertook. By combining AI-predicted physiology with established toxicokinetic equations, the approach retains the mechanistic interpretability that regulators demand while gaining the coverage that pure data-driven models lack.

The researchers have made their work openly available to accelerate adoption: the source code for the HM-PBTK model is hosted on GitHub, and the underlying data have been deposited on figshare, alongside extensive supplementary information detailing the species datasets, model construction, and validation results. The study was supported by the National Key Research and Development Program of China, the National Natural Science Foundation of China, and the Programme of Introducing Talents of Discipline to Universities. Whether the framework becomes a standard tool in regulatory risk assessment will depend on further independent validation and integration into formal assessment guidelines, but the direction is clear. As chemical inventories continue to expand and monitoring budgets remain constrained, the ability to predict which chemicals will reach which tissues of which fish—and at what internal concentrations—may prove one of the most consequential applications of artificial intelligence in the service of aquatic biodiversity protection.

Subject of Research: AI-driven prediction of internal chemical exposure and toxicity in freshwater and marine fish for aquatic ecological risk assessment.

Article Title: Advancing aquatic ecological risk assessment through AI-driven prediction of chemical exposure and toxicity in freshwater and marine fish

Article References: Han, P., Chen, J., Zhu, Y., Yang, J., Peijnenburg, W. J. G. M., & Li, X. (2026). Advancing aquatic ecological risk assessment through AI-driven prediction of chemical exposure and toxicity in freshwater and marine fish. Nature Water. https://doi.org/10.1038/s44221-026-00709-7

Image Credits: AI Generated

DOI: 10.1038/s44221-026-00709-7

Keywords: aquatic toxicology, ecological risk assessment, machine learning, PBTK model, chemical pollution, fish biodiversity, internal exposure, toxicity prediction, freshwater ecosystems, marine ecosystems, endocrine disruption, ADME simulation