| Homepage | Imprint |
|
||
 |
Quantitative structure-activity relationshipQSAR (quantitative structure-activity relationship, sometimes the A stands also for affinity= reactivity) is the quantitative correlationof the biological(ecological, toxicologicalor pharmacological) activity to the structure of chemical compounds, which allows the predictionof the so-called "drug efficacy" of a structurallyrelated compound. It is thus closely related to the more general field of QSPRand employs many of the latter's methodology. The biological activity of molecules is usually measured in assaysto establish the level of inhibition of particular signal transductionor metabolic pathways. Chemicals can also be biologically active by being toxic. Drug discoveryoften involves the use of QSAR to identify chemical structures that could have good inhibitory effects on specific targetsand have low toxicity(non-specific activity). Of special interest is the prediction of LogP, which is an important measure used in identifying "drug-likeness" according to Lipinski's Rule of Five. While many Quantitative Structure Activity Relationship analyses involve the interactions of a family of molecules with an enzymeor receptorbinding site, QSAR can also be used to study the interactions between the structural domainsof proteins. As in the article Structural modeling extends QSAR analysis of antibody-lysozyme interactions to 3D-QSAR, protein-protein interactions can be quantitatively analyzed for structural variations resulted from site-directed mutagenesis. In this study, a wild-typeantibodyspecific for lysozymeand 17 single and double mutantsof the antibody were investigated. Quantitative models for the affinity of the antibody-antigeninteraction were developed. 3D-QSAR is a specialization concerned with three-dimensional quantitative structure-activity relationships . This involves the analysis of the three-dimensional properties of molecules (chemical conformation). Note: QSAR is pronounced "q-sar" or "quasar". Post-QSAR technologies- Cheminformatics and Computer-assisted drug designis mostly based on docking technologies. There are a number of QSAR based molecular dockingtechniques-- some of them have been successfully applied for finding new drug candidates. However, there is a widespread opinion that these programs have certain shortcomings. The new paradigm in cheminformatics and molecular modelling? applying quantum and molecular physics instead of statistical scoring-function -like and QSAR-like methods, e.g. Quantum 3.1. These novel techniques demonstrate accurate affinity calculations due to fast quantum calculations, which take into account full flexibility of molecules, solvationeffects, and entropycontribution. The first striking discrepancy between Quantum ? like and the QSAR based techniques is the method of calculation of Free Binding Energy. Quantum applies ?pure? physics models whereas other docking programs use scoring functions for energy evaluation. - Scoring functions evaluate free energy by substituting the exact physical model with simplified statistical methods. Such methods take into account the data of a training set? different parameters of intermolecular interaction for a set of protein-ligand complexes. As a result of such a statistical approach, the scoring function approximates only those ligands, which are similar to those in the training set with any level of accuracy and is not appropriate for analyzing novel molecules or for de novo drug design. Because of this shortcoming, docking programs based on scoring functions are typically used for docking molecules that are similar to the training set molecules; or as rough filters before wet chemistry screening to get rid of molecules with very low affinity. Moreover, and this is critically important for pharma and other industries, scoring functions can not predict inhibitorsof novel classes, because the accuracy of all scoring functions depends greatly on the quality of training sets of known ligands belonging to known classes. That impacts patentabilityand increases the danger of infringementon the rights of third parties. The accuracy of binding calculations and docking performed using ?pure? physics models is not effected by this and Quantum's-like approaches can generate truly novel inhibitors that are both strong and patentable. - As a rule, scoring functions are very simple mathematical functions, which require rather small computer resources and therefore calculate very quickly. Post-QSAR?s algorithms employ far more sophisticated mathematical models, which require a great deal of computer recourses. Quantum?s optimization algorithms are sophisticated tools for analyzing huge numbers of local minima of energy for protein-ligandcomplexes. The latest achievements of mathematics ? modern Multigridmethods and so called Minima Hopping Methods - are implemented in Quantum?s algorithms. Free energy calculations take from minutes to half an hour per complex depending on size of the ligand and the active site of the protein at the one node. - Scoring functions perform calculations quickly, but the quality is very poor. The configurational space in these calculations has fewer minima than a real complex would have. This means that scoring functions are not selective towards the ligand and all ligands have approximately the same predicted affinities. - QSAR-based technologies apply mostly very simplified solvationmodels like the Surface Area/Generalized Born. Quantum also uses this model, but only during the initial phases of screening. Some technologies ignore this subject entirely pretending that these effects are already taken into account in the scoring function. Obviously, this omission creates errors in predicting affinities. Quantum uses the Poisson-Boltzmann Solvation Model, which requires solving partial differential equations. This takes time, but dramatically improves the results. - Protein Flexibility. Almost all chemoinformatics technologies treat ligandsas flexible, but only few, has moved further and considered protein flexibility. Flexibility is one of the most important features because small changes in geometry of molecules lead to significant alterations in the free binding energy. There are two kinds of flexibility of proteins in Quantum. The first accounts for small changes in the geometry of the protein by calculating corrections to entropy. In the second type, overall movements of the protein are taken into account; these movements are modeled by Normal ModeAnalysis. - The result of technological advances is a dramatic difference in the accuracy of calculation between the post-QSAR and QSAR and scoring functions based programs. Post-QSAR techniques show an accuracy of only 15% error in free energy calculation. The majority of QSAR applications either try not to mention accuracy at all or indicate figures in the range of 40-50% error. Even those figures are doubtful because they can be obtained on complexes that are similar to those in the training sets and do not represent the general pattern. A fair estimate of errors of methods based on scoring functions should be near 100%. So these methods will have very limited application in future drug design. The application of Quantum -like techniques does not demand the use of a training set. The comparison of the perfomance of the new technologies with the QSAR based and experimental tests can be found or made independently, e.g. here here.
See also
External links
ja:定量的構造活性相関 zh:?????? Categories: Medicinal chemistry| Pharmacology
|
