Abstract
In the last several decades, the drug research has moved to involve various IT technologies in order to rationalize the design of novel bioactive chemical compounds. An important role among these computer-aided drug design (CADD) methods is played by a technique known as quantitative structure-activity relationship (QSAR).
The approach is utilized to find a statistically significant model correlating the biological activity with more or less extent data derived from the chemical structures. The present article deals with approaches for discriminating unimportant information in the data input within the three dimensional variant of QSAR - 3D QSAR.
Special attention is turned to uninformative and iterative variable elimination (UVE/IVE) methods applicable in connection with partial least square regression (PLS). Herein, we briefly introduce 3D QSAR approach by analyzing 30 antituberculotics.
The analysis is examined by four UVE/IVE-PLS based data-noise reduction methods.