AplicaÃÃes de metodologias teÃricas em estudos de QSAR-3D de pirimidinas substituÃdas

AUTOR(ES)
DATA DE PUBLICAÇÃO

2004

RESUMO

The aim of this work is determining a 3D QSAR model for a homologous series of substituted pyrimidines with known anti-inflammatory activities. Previous results from classic QSAR (which include AM1 calculated parameters) have shown the importance of electronic parameters in the successful prediction of a novel and more active compound. The first step was to make an evaluation of the calculation of level and the basis set due to the dependence between electronic parameters of QSAR models and the geometry and the base function used. The non substituted molecule (5h) was selected as a model compound and, initially, molecular orbital calculations at AM1, HF, DFT and MP2 levels were used within and/or without base function that contain diffuse and/or polarization functions using the G98 program. A chemometrics analyses and a comparison of x-ray crystallographic data of related systems suggested the DFT-BP86/6-31G** method to be used for the study oh the other molecules of this series. After determinations of which method and basis set should be used, a conformation analysis and the evaluation of the effect of a partial and a full geometry optimization procedure over the atomic ChelpG charges was performed using the chemometric method of PCA. Then an EPS study of this series was performed with the complete by optimized molecule. At this step AM1 and BP86/6-31G** levels were used to assess the effect of the method on the EPS of the same equilibrium geometry. The ALINHAMOL program was used to align the molecules to the EPS calculation with the G98 program. After that, the CAMOL program was used to generate a data matrix of PE with which was performed the 3D QSAR procedure. Then, a systematic study of the influence of the variables of this program (cutoff values, etc) on the quality of the chemometric models (PCA and PLS) was carried out. Our results had indicated a high relationship between method/base function and the EPS, because the pattern of PC1 versus PC2 scores plots will depend in a significant way on the choice of the calculation methodology. On the other hand the regression analysis showed that CamolÂs variables (VCDS and VCIS) do not have a great influence on the prediction quality of the obtained models.

ASSUNTO(S)

qsar-3d quimica pirimidinas substituÃdas qsar-3d metodologia teÃrica

ACESSO AO ARTIGO

Documentos Relacionados