STRUCTURE-QUANTITATIVE PROPERTIES (QSPR)

• This subject introduces the student to the characteristic methodology of the QSPR approach in the following steps: (a) bibliographic research on the experimental values of the property to be studied; (b) drawing of the structures of the compound using the program Chemlite; (c) optimization of the geometry using the routines of molecular mechanics in the program Chemlite and quantum methods (MOPAC); (d) calculation of the molecular descriptors using the program CODESSA (which obtains more than 600 different descriptors); (e) selection of those descriptors most closely related with the variable dependent (experimental property object of study), using different statistical techniques; (f) statistical analysis of the results. The correctness of the correlation is decided in function of the values of the usual statistical parameters, R2, coefficient of the test F, standard deviation, and especially for the physicochemical meaning of the descriptors. The models proposed for the estimation of the property studied can be linear, deduced by methods of multiple linear regression, or non-linear derived by neural nets.
  
  RECOMMENDED READING
  A. R. Katritzky, V. S. Lobanov, M. Karelson. 1995. "QSPR: The correlation and Quantitative Prediction of Chemical and Physical Properties from Structure". Chem. Rev., 95, 279.
  P. C. Jurs. 1998. "Quantitative Structure-Property Relationships (QSPR)" Encyclopedia of computational chemistry (Paul von Ragué Schleyer, ed.).
  M. Karelson. 2000. "Molecular descriptors in QSAR/QSPR" Wiley.
  A. R. Katritzky, V. S. Lobanov, M. Karelson. 1995. "QSPR: The correlation and Quantitative Prediction of Chemical and Physical Properties from Structure" Chem. Rev., 95, 279.