Computational Toxicology of Nanomaterials

Computational toxicology of nanomaterials integrates data-driven modelling, machine learning and mechanistic simulations to predict adverse effects of engineered nanomaterials. Central to this discipline is the transformation of physicochemical characterisations into quantitative nanodescriptors and the coupling of these descriptors with algorithms ranging from quantitative structure–activity relationships (QSAR) to physiologically based pharmacokinetic (PBPK) models. In silico approaches enable high-throughput screening and hazard ranking without exhaustive animal testing, accelerating both regulatory assessment and the design of safer-by-design materials. Key challenges include the curation of harmonised, high-quality datasets, the definition of applicability domains for predictive models and the reconciliation of in vitro findings with in vivo outcomes. Advances in cloud-based platforms and publicly accessible databases further empower researchers to share integrative tools for read-across, dosimetry modelling and multi-endpoint risk assessment. By combining data curation, descriptor generation and robust validation protocols, computational toxicology is establishing a foundational framework for global nanosafety and sustainable nanotechnology development.

Research from Nature Portfolio

Recent studies have delivered foundational resources and predictive frameworks for nanotoxicology modelling. A comprehensive database comprising over 700 unique nanomaterials has been curated with up to six physicochemical and bioactivity endpoints per entry, each annotated into machine-readable nanostructure files. This platform generates more than 2,000 nanodescriptors to support machine learning and rational design of novel materials. In parallel, a generalised toxicity classification model for seven oxide nanomaterials was developed using literature-sourced datasets screened by quality scores. Preprocessing techniques such as oversampling addressed class imbalance, and neural network classifiers outperformed other algorithms. Defining the applicability domain via k-nearest neighbours and identifying critical features—dose, formation enthalpy, exposure time and hydrodynamic size—provided a robust tool for predicting toxicity across diverse experimental contexts.

Computational Toxicology of Nanomaterials publication trend

The graph below shows the total number of articles in computational toxicology of nanomaterials across all publications each year (not limited to Nature Index journals).

Technical terms

In silico modelling: Use of computer simulations and algorithms to predict biological or environmental effects of substances without physical experiments.

Quantitative structure–activity relationship (QSAR): Statistical models correlating chemical structures or descriptors with biological activity or toxicity.

Physiologically based pharmacokinetic (PBPK) modelling: Mechanistic simulation of absorption, distribution, metabolism and excretion of substances in living organisms.

Dosimetry: Calculation or measurement of the internal or external dose of a substance received by an organism.

Read-across: Extrapolation of toxicity data from one substance to another based on structural or property similarity.

Nanodescriptor: Quantitative representation of nanoscale features (size, shape, surface chemistry) used as inputs for predictive models.

Safe-by-design: Strategy to incorporate safety considerations early in material development by avoiding hazardous properties through design choices.

References

1. NANO.PTML model for read-across prediction of nanosystems in neurosciences. computational model and experimental case of study. Journal of Nanobiotechnology (2024).
2. Review of emerging concepts in nanotoxicology: opportunities and challenges for safer nanomaterial design. Toxicology Methods (2019).
3. NanoSolveIT Project: Driving nanoinformatics research to develop innovative and integrated tools for in silico nanosafety assessment. Computational and Structural Biotechnology Journal (2020).
4. Construction of a web-based nanomaterial database by big data curation and modeling friendly nanostructure annotations. Nature Communications (2020).
5. Towards a generalized toxicity prediction model for oxide nanomaterials using integrated data from different sources. Scientific Reports (2018).