The environmental fate of many functional molecules that are produced on a large scale as precursors or as additives to specialty goods (plastics, fibers, construction materials, etc.) is generally unknown.
Assessing their environmental fate is crucial when making decisions on the manufacturing, handling, usage, and release of these substances, as is the evaluation of their toxicity in humans and other higher organisms. A possibility is to use the experimental data already available on the biodegradability and toxicity of many unusual compounds to develop machine learning systems that will allow to predict these features.
In this work, a predictor of the “risk” associated with the use and release of any chemical has been generated by merging computational methods to predict biodegradability with others that assess biological toxicity. The combined platform is called BiodegPred, and it provides an informed prognosis of the chance that a given molecule can eventually be catabolized in the biosphere, as well as of its eventual toxicity, all available through a simple web interface.
Access the full paper here.