Abstract

In this study, we present Apellis, a web application developed for training grouping/read-across models for the prediction of engineered nanomaterials (ENMs) toxicity-related endpoints. The application applies a generic and novel computational workflow for estimating the endpoint of interest, which can be either categorical or numerical. During the training procedure, the application selects the most important ENM properties of concern that affect their toxic behaviour. In the process of grouping ENMs for performing read-across predictions, the multi-perspective characterization of ENMs can be taken into account, by defining more than one similarity criteria. The workflow converges to the grouping hypothesis that leads to the most accurate read-across estimations. Visualisation tools are included in the application, which offer better and more clear understanding of grouping and similarities among ENMs. The trained models can be saved in an electronic format, so that they can be easily retrieved, for calculating new predictions. In addition, this allows model developers to disseminate and share the produced models with the community. Apellis is free to use and accessible at apellis.jaqpot.org.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call