| Title: | A QSAR for the prediction of rate constants for the reaction of VOCs with nitrate radicals |
| Address: | "Bayer CropScience AG, Building 6692, Alfred Nobel Str. 50, D-40789 Monheim, Germany. Electronic address: michael.schindler@bayer.com" |
| DOI: | 10.1016/j.chemosphere.2016.03.096 |
| ISSN/ISBN: | 1879-1298 (Electronic) 0045-6535 (Linking) |
| Abstract: | "A QSAR for the prediction of rate constants for the degradation of volatile organic compounds by nitrate radicals is developed using the Partial Least Squares technique. The QSAR is based on experimental data published in the literature for 260 compounds. They are modeled by a set of calculated descriptors from standard descriptor generation tools and from quantum chemistry. Out of several diversity-based partitionings of the data set a diverse set of 99 compounds turned out to be the optimum choice with regard to simplicity and performance. The final QSAR model is characterized by r(2) = 0.831 (fit) and q(2) = 0.823 (prediction), and by an r(2)pred = 0.862 for the n = 155 external validation set. The QSAR needs 3 latent variables. The most important descriptors for the QSAR are the ionization potential, obtained from density functional theory, and the energy of the highest occupied molecular orbital, which are modulated by fingerprints indicating the presence of specific molecular fragments like functional groups or ring systems. The applicability domain of the new QSAR was studied for some compound classes which are important for the crop protection industry, including (di)hydroxbenzenes and heterocyclic compounds" |
| Keywords: | "Benzene Derivatives/chemistry Heterocyclic Compounds/chemistry Kinetics Least-Squares Analysis Models, Theoretical Nitrates/*chemistry *Quantitative Structure-Activity Relationship Volatile Organic Compounds/*chemistry Applicability domain Diversity selec;" |
| Notes: | "MedlineSchindler, Michael eng England 2016/04/03 Chemosphere. 2016 Jul; 154:23-33. doi: 10.1016/j.chemosphere.2016.03.096. Epub 2016 Mar 31" |
