Modelling the in vitro cytotoxicity of metal/metal oxide and silica nanomaterials under diverse experimental conditions

In spite of the increasing potential applications of nanoscale materials, concern regarding their possible adverse effects on human health has arisen. In light of this, several scientific studies and international research initiatives have focused on the assessment of the in vitro cytotoxicity of nanomaterials under different experimental conditions. In particular, the European MODENA COST Action [1] has allowed for the coordination of nationally-funded research on a European level. In this work, experimental data obtained from the different partners involved in the MODENA COST action were used as input for the development of in silico models which are sensitive to variation in both the chemical composition of the nanomaterials considered and different sets of experimental conditions. Descriptors derived from experimental conditions included assay type, exposure duration, cell type, serum concentration and the dispersion protocol that was adopted to perform the test; descriptors from measurements of the physico-chemical properties of nanomaterials such as size, coating, shape, aspect ratio, surface area and zeta potential were also included in the modelling. Different machine learning and modelling techniques, including the random forest and CORAL approaches [2], were used to develop models. The leave-one-out method was applied to check the predictive performance of the developed models. This research has received funding from the European Union Seventh Framework Programme [FP7/2007-2013] under grant agreement n° 309837 (NanoPUZZLES project).