Predicting ADME properties and side effects: the BioPrint approach.

摘要: Computational methods are increasingly used to streamline and enhance the lead discovery optimization process. However, accurate prediction of absorption, distribution, metabolism excretion (ADME) adverse drug reactions (ADR) is often difficult, due complexity underlying physiological mechanisms. Modeling approaches have been hampered by lack large, robust standardized training datasets. In an extensive effort build such a dataset, BioPrint database was constructed systematic profiling nearly all drugs available on market, as well numerous reference compounds. The composed several large datasets: compound structures molecular descriptors, in vitro ADME pharmacology profiles, complementary clinical data including therapeutic use information, pharmacokinetics profiles ADR profiles. These allowed development computational tools designed integrate program chemistry into library design development. Models based chemical structure strengthened results that can be additional descriptors predict complex vivo endpoints. pharmacoinformatics platform represents accelerate process discovery, improve quantitative structure-activity relationships develop vitro/in associations. this review, we will discuss importance set size diversity model development, implementation linear neighborhood modeling approaches, silico potential liabilities.