Self-organizing Neural Networks for Modeling Robust 3D and 4D QSAR Application to Dihydrofolate Reductase Inhibitors 英文参考文献.docVIP

Molecules 2004, 9, 1148–1159 molecules ISSN 1420-3049 Self-organizing Neural Networks for Modeling Robust 3D and 4D QSAR: Application to Dihydrofolate Reductase Inhibitors Jaroslaw Polanski *, Andrzej Bak, Rafal Gieleciak and Tomasz Magdziarz Department of Organic Chemistry, Institute of Chemistry, University of Silesia, PL-40-006 Katowice Poland * Author to whom correspondence should be addressed; e-mail: polanski@.pl Received: 31 May 2004 / Accepted: 21 Oct 2004 / Published: 31 December 2004 Abstract: We have used SOM and grid 3D and 4D QSAR schemes for modeling the activity of a series of dihydrofolate reductase inhibitors. Careful analysis of the performance and external predictivities proves that this method can provide an efficient inhibition model. Keywords: Self-organizing neural network, 3D QSAR, 4D QSAR, SOM-4D QSAR, CoMSA. Introduction Drug discovery is a complex issue that lacks a general approach. Drugs are mainly synthetic products developed by chemists. However, in this context the most fundamental and lasting objective of synthesis is not a production of new compounds but the production of properties, as commented by Hammond and cited by Sharpless and co-workers [1]. This fact clearly makes QSAR (Quantitative Structure-Activity Relationships), in its broadest sense, an essential and irreplaceable method in this field. However, more and more sophisticated tools are needed for the transformation of the molecular structure into the compound property space. Generally, the drug-receptor interactions are complex phenomena, which cannot be easily described. Therefore, a QSAR strategy of the comparison of a series of drug ligands separately from the receptor structure has evidently limitations. In 3D or 4D QSAR molecular superimposition that should be performed for the compound series can be mentioned here as an illustrative example. By performing superimposition, intent