Premium
From Crystal Structures and Their Analysis to the in silico Prediction of Toxic Phenomena
Author(s) -
Dobler Max,
Lill Markus A.,
Vedani Angelo
Publication year - 2003
Publication title -
helvetica chimica acta
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.74
H-Index - 82
eISSN - 1522-2675
pISSN - 0018-019X
DOI - 10.1002/hlca.200390134
Subject(s) - in silico , chemistry , virtual screening , quantitative structure–activity relationship , computational biology , cheminformatics , small molecule , solvation , biochemical engineering , ligand (biochemistry) , pharmacophore , chemical space , molecule , drug discovery , combinatorial chemistry , computational chemistry , stereochemistry , receptor , organic chemistry , biochemistry , biology , engineering , gene
Abstract While the development of potential drug molecules based on the known three‐dimensional structure of the macromolecular target is doubtless one of the more‐potent approaches to rational drug design, the estimation of associated changes in the free energy of ligand binding is all but trivial. Major obstacles include the treatment of long‐range electrostatic effects and charge transfer, the calculation of solvation energies, the treatment of entropic effects, and the quantification of induced fit. In the last decade, a number of computational concepts have nonetheless matured into powerful tools for the development of drug‐candidate molecules. These concepts have mainly focussed on the binding of the small molecule to a bioregulator. More recently, need has arisen to develop tools for a safe prediction of more‐complex phenomena such as metabolism, toxicity, and bioavailability. We describe the ongoing development of a virtual laboratory on the Internet to allow for a reliable in silico estimation of harmful effects triggered by drugs, chemicals, and their metabolites. For this, we used our recently developed underlying technology (5D‐QSAR, based on five‐dimensional quantitative structure‐activity relationships) and compiled a pilot project, including the models of five receptor systems known to mediate adverse effects (the aryl hydrocarbon ( Ah ), 5 HT 2A , cannabinoid, GABA A , and estrogen receptor, resp.) which were already validated against 280 compounds (drugs, chemicals, toxins). Within this setup, we could demonstrate that our virtual laboratory is able both to recognize toxic compounds substantially different from those used in the training set as well as to classify harmless compounds as being nontoxic. The results suggest that our approach can be used for the prediction of adverse effects of drug molecules and chemicals and, thus, bears a significant potential to recognize hazardous compounds early in the development process hence improving resource and waste management and reducing animal testing. It is the aim to provide free access to this technology – particularly to universities, hospitals, and regulatory bodies.