S1: Computational nanotoxicology
Laboratory of Environmental Chemometrics,
University of Gdansk
In effect of conducting six EU FP7 projects (NanoPUZZLES, ModENPTox, PreNanoTox, MODERN, MembraneNanoPart, eNanoMapper) and COST MODENA action we have created solid basis for predictive modelling the relationships between structure and properties/activity of nanoparticles. Recently developed computational nanotoxicology methods include: (i) computational chemistry methods (Molecular Dynamics, Quantum Mechanics and Density Functional Theory calculations); (ii) chemoinformatic methods (Quantitative Structure-Activity Relationships modeling, similarity analysis and read-across) and (iii) bioinformatics methods.
Surprisingly, majority of the recently published conclusions on the structure-activity dependencies for nanoparticles have not been supported by the application of any formal, validated methodology (e.g. Nano-QSAR). Moreover, definitive conclusions have been often made based on studying trends in very small statistical samples (small number, usually 3-4 structures) selected without any reflection on their representativeness for the whole class (or subclass) of nanoparticles.
The reflection on very limited availability of data that can be eventually used for developing predictive models brings the reader to a deeper conclusion. The role of modeling groups in nowadays research projects has not been defined properly. This seems to be a classical “end-of-the-pipe” problem; the modelers usually have no real influence on the designing the experiment and selection of the number and structural variability of the empirically tested nanoparticles to ensure representativeness of the received data. At the end of the experimental work, they just receive from the experimentalists data of low value for modeling together with great expectations. The idea of this talk is to briefly summarize current achievements and provoke discussion on future directions of computational nanotoxicology in next years.