Jaqpot 5: How to access and use the Naive Bayes model for the prediction of cytotoxicity of metal oxide nanoparticles

Standard experimental methods for toxicity testing in biological systems (cells, model organisms) can be overwhelmingly costly in resources as well as ethically inappropriate. Given the contemporary popularity of the nanomaterials, new in-silico approaches have been developed in order to limit the traditional experimental testing, while providing accurate results. Simeone and Costa (2019) recently developed and reported a simple, yet accurate, model to calculate the probability that nano-oxides will activate biological pathways which are causally linked to in vitro cytotoxicity. This probability is inferred from physicochemical parameters that do not need to be experimentally determined, but are available to everyone: oxidation number, ionic potential, surface reducibility, and redox reactivity. These parameters were chosen as good candidates for a Naive Bayesian classifier model, because it has been found that they determine the degree of dissolution, surface charge, surface reactivity and bulk reactivity that have been experimentally linked to toxicity. The choice of the Naive Bayesian classifier offers an important opportunity for the future, since it can be easily extended to incorporate additional sources of information as soon as they become available. The classifier predicts the most probable level of toxicity of a nanoparticle given its structural formation, based solely on the assumption that the variables used contribute independently to the probability of each toxicity level. A web application of the classifier has been implemented through the Jaqpot platform that offers all the necessary infrastructure and functionalities to develop the model, deploy it as a web service and share it with the scientific community as well as industry stakeholders. This document provides a tutorial on accessing and using the Jaqpot implementation of the Naïve Bayes model . The resource has been made available at https://app.jaqpot.org/model/2XyMtChRiSfZ6UmKirhi