| |
Abstract:
The nanomaterials posed risks to human health. There is
a need for simple and effective methods to predict the
toxicity of unknown nanoparticles. Quantitative
structure-activity relationships were used to research
the toxicity of metal oxide nanoparticles. 25
descriptors of 21 metal oxide nanoparticles were
obtained. It is found that the toxicity is related to 5
descriptors including xc, Shift, Eg, ZPVE, and Softness
by using the multiple linear regression method with an R2
of 0.893. Structures of two metal oxide (silver oxide
and lead oxide) nanoparticles were designed based on
multiple linear regression equation. And we found the
structural models of them with the lowest toxicity. The
multiple nonlinear regression model was used to further
improve the prediction accuracy. The R2 were
0.943 in training set and 0.931 in test set, which
indicated that the prediction accuracy has been improved
effectively. Based on the results, we explained the
effects of the 5 descriptors on toxicity, discussed the
mechanism of toxicity of metal oxide nanoparticles, and
proved that reactive oxygen species is an important
cause of the toxicity. Finally, it is concluded that the
quantitative structure-activity relationships model is a
reasonable method to design the structures of metal
oxide nanoparticles with lower toxicity.
|
|
|
