Abstract:

logP is an important parameter for evaluating the hydrophobicity of organic compounds and other species. It is widely used in 2D-QSAR and recently in 3D-QSAR studies. On the basis of the current fragment addition method, we proposed a new algorithm called XLOGP by introducing atomic parameters like atom accessible surfaces, van der Waals volumes and partial charges. Step-wise multiple regression on a test set of 400 organic compounds were carried out. The model was then tested by cross-validation. The final correlation coefficient for the whole data set fitting and for cross -validation is 0.982 and 0.971, respectively. This shows that the model obtained is valid both at data fitting and at prediction. Compared to other algorithm, this method can discriminate a common atom or fragment in different environment.

Key words: Patition coefficient, Fragment addition, Atom parameters, Organic compound