Applications of Molecular Connectivity Indexes and Multivariate Analysis in Environmental Chemistry

摘要

The authors have developed a data matrix of 90 variables calculated from molecular connectivity indices for 19,972 chemicals in the Toxic Substance Control Act (TSCA) inventory of industrial chemicals. The first three principal components convey generalized information on chemical structure: size, degree of branching, and number of cycles. The other components contained more specific information on branching, bonding, cyclicness, valency, and combinations of these structural attributes. Here the authors explored the use of the connectivity indices and their calculated principal components for their potential in predicting biodegradation as measured by biochemical oxygen demand (BOD) and the octanol/water partition coefficient. This approach showed promise in the prediction of biodegradation, but was of limited use in the prediction of the partition coefficient. Because it is possible to calculate the connectivity indices at a nominal cost for nearly all chemicals, the approach will prove especially useful for the identification of chemicals with similar structures and for systematically exploring where data are lacking on biological endpoints for chemicals in TSCA.