NanoSAR : in silico modelling of nanomaterial toxicity

• Main Author: Oksel, Ceyda
• Other Authors: Wang, Xue Z.
• Published: University of Leeds 2016
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.693076

The number of engineered nanomaterials (ENMs) being exploited commercially is growing rapidly, due to the novel properties of ENMs. Clearly, it is important to understand and ameliorate any risks to health or the environment posed by the presence of ENMs. However, there still exists a critical gap in the literature on the (eco)toxicological properties of ENMs and the particular characteristics that influence their toxic effects. Given their increasing industrial and technological use, it is important to assess their potential health and environmental impacts in a time and cost effective manner. One strategy to alleviate the problem of a large number and variety of ENMs is through the development of data-driven models that decode the relationships between the biological activities of ENMs and their physicochemical characteristics. Although such structure-activity relationship (SAR) methods have proven to be effective in predicting the toxicity of substances in bulk form, their practical application to ENMs requires more research and further development. This study aimed to address this research need by investigating the application of data-driven toxicity modelling approaches (e.g. SAR) that are beneficial over animal testing from a cost, time and ethical perspective to ENMs. A large amount of data on ENM toxicity and properties was collected and analysed using quantitative methods to explore and explain the relationship between ENM properties and their toxic outcomes, as a part of this study. More specifically, multi-dimensional data visualisation techniques including heat maps combined with hierarchical clustering and parallel co-ordinate plots, were used for data exploration purposes while classification and regression based modelling tools, a genetic algorithm based decision tree construction algorithm and partial least squares, were successfully applied to explain and predict ENMs’ toxicity based on physicochemical characteristics. As a next step, the implementation of risk reduction measures for risks that are outside the range of tolerable limits was investigated. Overall, the results showed that computational methods hold considerable promise in their ability to identify and model the relationship between physicochemical properties and biological effects of ENMs, to make it possible to reach a decision more quickly and hence, to provide practical solutions for the risk assessment problems caused by the diversity of ENMs.