PreADMET program resides entirely on a Web server, and can be accessed by browsers such as Netscape or Internet Explorer. The application is written mainly in PHP, a scripting language commonly used for Web application which communicates with the browser through a CGI interface. The PHP code in turn uses a set of C program that implement much of the functionality of PreADMET. PreADMET consists of four main parts as following:
- Molecular Descriptor Calculation: The ADME/Tox properties are closely related to physico-chemical descriptors such as lipophilicity (logP), molecular weight, polar surface area, and water solubility. The TOPOMOL module calculates more than 2500 molecular descriptors including constitutional, topological, electrostatic, physico-chemical, and geometrical descriptors for ADME/Tox prediction from 2D and 3D chemical structure: TOPOMOL reads MDL mol or sd files and provides a rapid means to calculate all 2D descriptors (except BCUT descriptors) of 1,000,000 compounds in less than 1 hour using Pentium IV 3.4GHz PC.
- Drug-likeness Prediction: The most well known rule relating the chemical structures to their biological activities is Lipinski’s rule , it is called the ‘rule of five’. Another well known rule is the Lead-like rule , i.e. starting from a quantitative survey based upon 18 lead and drug pairs of structure. PreADMET contains drug-likeness prediction module based on these rules. Also, it is possible to use several drug-like rules that several researchers defined drug-like characteristics of drug DB such as CMC, WDI, and MDDR DB.
- ADME Prediction: Numerous in vitro methods have been used in the drug selection process for assessing the intestinal absorption of drug candidates. Among them, Caco2-cell model and MDCK(Madin-Darby canine kidney) cell model has been recommended as a reliable in vitro model for the prediction of oral drug absorption. In absorption, this module provides prediction models for in vitro Caco2-cell  and MDCK cell  assay. Additionally, in silico HIA (human intestinal absorption) model and skin permeability model can predict and identify potential drug for oral delivery and transdermal delivery. In distribution, BBB (blood brain barrier) penetration can give information of therapeutic drug in the central nervous system (CNS), plasma protein binding model in its disposition and efficacy. In order to build these QSAR models, genetic functional approximation is used to select relevant descriptors from all 2D descriptors that calculated by Topomol module, followed by Resilient back-propagation(Rprop) neural network to develop successful nonlinear model.
- Toxicity prediction: In silico toxicity prediction will have more and more importance in early drug discovery since 30% of drug candidates fail owing to these issues. We tried to develop reliable model which can readily classify mutagenicity of Ames Salmonealla TA100, TA98, TA1535 species and rodent carcinogenicity 2year assay of rat and mouse by backward elimination and Rprop neural net method.