You are here:
DEVELOPMENT OF QUANTITATIVE STRUCTURE ACTIVITY RELATIONSHIPS TO PREDICT TOXICITY FOR A VARIETY OF HUMAN AND ECOLOGICAL ENDPOINTS
Citation:
MARTIN, T. M., P. F. HARTEN, D. M. YOUNG, AND R. VENKATAPATHY. DEVELOPMENT OF QUANTITATIVE STRUCTURE ACTIVITY RELATIONSHIPS TO PREDICT TOXICITY FOR A VARIETY OF HUMAN AND ECOLOGICAL ENDPOINTS. Presented at 2006 Toxicology and Risk Assessment Conference, West Chester, OH, April 24 - 27, 2006.
Description:
The number of chemicals released into the environment has significantly increased over the past few years, leading to increased risk of human exposure through inhalation, ingestion, or dermal uptake. In addition, the risk also increases with increasing toxicity of the chemical. A number of ranking schemes that are based on exposure and toxicity have been developed to aid in the prioritization of research funds for identifying chemicals of regulatory concern. However, there are significant gaps in the availability of experimental toxicity data for most health endpoints. Predictive toxicological approaches such as Quantitative Structure-Activity Relationships (QSARs) provide a means to estimate the toxicities for chemicals that lack experimental data. For the purposes of this study, QSARs are mathematical equations that relate the toxicity of a chemical to its physicochemical properties calculated using Kier and Hall type indices (2-dimensional descriptors) and computational quantum chemistry (3-dimensional descriptors).
The objective of this study is to construct a software tool to predict the oral rat lethal doses (LD50s) of a wide variety of chemicals. The software tool will be coded in Java and accessible through the Web. The user will simply input a chemical to be evaluated by drawing it in a 2-D chemical sketcher window, entering a SMILES string, or entering a CAS number. To build the tool, a database containing LD50s of approximately 3000 chemicals was gathered from the literature. Algorithms for calculating 2-dimensional descriptors such as connectivity, shape, E-State, and other information indices are being developed in Java, and the descriptors calculated by the program are validated against those caluclated by the descriptor generator programs ADAPT, Dragon, MDL QSAR and Molconn-Z. The 3-dimensional descriptors such as partial charges and surface areas are calculated using freely available platform-specific programs such as MOPAC and TINKER. Algorithms for generating these descriptors are being developed in Java and validated against AMPAC, MOPAC, NWChem and TINKER. C/C++ and Java based algorithms are also being developed that group the chemicals in the database into clusters of chemicals with similar physicochemical properties (based on Ward's method), and generate the QSAR equation to predict the LD50 of a chemical from its physicochemical properties for each cluster using parallelized genetic algorithms. The chemicals in each cluster and their predicted LD50s will be validated against those generated by MDL QSAR and SAS.