(145a) Prediction of EHS Properties Using T.E.S.T | AIChE

(145a) Prediction of EHS Properties Using T.E.S.T


Martin, T. - Presenter, National Risk Management Research Laboratory, US EPA
Young, D. - Presenter, National Risk Management Research Laboratory, US EPA

The US EPA has developed tools (such as the WAste Reduction Algorithm) which aid in the development of sustainable chemical process designs. In order to utilize these tools one must be able to characterize chemicals in terms of their EHS (environmental health and safety) properties. Since experimental values are not available for all industrially relevant chemicals, one needs to be able to estimate these properties using in silico based methods.  T.E.S.T. (Toxicity Estimation Software Tool) was developed to allow users to easily estimate toxicity values using quantitative structure activity relationship (QSAR) models.  The hierarchical clustering QSAR methodology was selected because it can develop models for large, diverse datasets.  This methodology utilizes Ward’s method to divide a training set into a series of structurally similar clusters.  The structural similarity is defined in terms of about 800 2-D physicochemical descriptors (such as connectivity and E-state indices).  A genetic algorithm based technique is used to generate statistically valid QSAR models for each cluster.  The toxicity for a given query compound is estimated using the weighted average of several different cluster models (assuming that the compound is within the domain of applicability of the cluster model).  T.E.S.T. can estimate toxicity values for a variety of toxicity endpoints including acute aquatic toxicity, acute mammalian toxicity, fish bioaccumulation factor, developmental toxicity, and mutagenicity.  TEST can also estimate physical properties including normal boiling point, density, water solubility, and thermal conductivity. These properties are important for green solvent design.  The predictive ability for each endpoint was validated using an external validation set.