Toward molecular design for hazard reduction-fundamental relationships between chemical properties and toxicity

Abstract

The relationship of in-silico predicted physical/chemical properties and human toxicity is analyzed for a statistically significant sample size of chemical compounds. Results for compounds with known toxicity endpoints, as designated by EPA's Toxic Release Inventory (TRI), are compared to a series of commercial chemicals that are not regulated under TRI. Physical properties for all compounds are predicted using Schrodinger's QikProp, an established tool for predicting adsorption, distribution, metabolism, and excretion (ADME) characteristics. The results of this analysis indicate that the physical/chemical property distributions of TRI chemicals are statistically significantly different from those of bulk commercial chemicals, particularly related to properties associated with bioavailability. Using a partitioning analysis, several key physical/chemical properties and ranges are identified that can be used to readily differentiate TRI chemical characteristics from those of bulk commercial chemicals. © 2009.