Log in or Sign Up

Statistics

In the 3-D QSAR server are available now 30 datasets,
more than 230 3-D QSAR models and over 1600 unique molecules.

Select a dataset from the box to the left and start to inspect one of the available model.

AI

Description

The AI dataset was taken from J. Chem. Inf. Comput. Sci. 1997, 37, 1162-1170. Paper abstract: The all-grid probe Lennard-Jones 6-12 potential, typically used as steric descriptor in CoMFA, has been replaced by the volumes of van der Waals envelopes intersections between the probe-atom and the ligand molecules under investigation. The intersection volumes present a smoother distance dependence that overcomes the problems arising from formulation of a precise alignment and docking into the 3D lattice. A CoMFA-type application on a set of 78 steroid aromatase inhibitors with different grid and probe-atom characteristics suggests an improved model robustness for the new steric field in comparison with the classical 6-12 potential. Predefined cut-off and “minimum sigma” values are not required. Systematic variation of the 3D lattice position leads to lower variations of cross-validated r-squared (q2 LOO) at all levels of model complexity, which can be reduced with exclusion of some “interior” points. The relative simplicity and inexpensive computational demands make this steric field a promising alternative for routine CoMFAbased 3D-QSAR analyses.