COMP 66 |
| Andreas Klamt, COSMOlogic GmbH&CoKG, Burscheider Str. 515, Leverkusen, 51381, Germany |
| Continuum solvation models like COSMO (Conductor-like Screening Model) have proven to be efficient and reliable tools for the approximate treatment of solvents in quantum chemical calculations. The combination of density functional theory with such continuum models gives good qualitative results at moderate computation time. A much more general and powerful access to the problem is provided by the COSMO-RS method. This uses the output of COSMO calculations, i.e. energies and screening charge densities on the molecular surface, as basis for a statistical thermodynamics treatment.This new method is much more efficient and probably more accurate than MD- or MC-simulations which are used for liquid systems by many computational chemists. In addition it provides much physical insight into the problem of solvation. On the other hand it is much more general applicable than group contribution methods like CLOGP or UNIFAC, which is state of the art in chemical engineering. Based on DFT/COSMO calculations COSMO-RS has been parameterized for all elements relevant in organic chemistry, yielding remarkable accuracy for vapor pressures, Ghydr and partition coefficients. Even more, solubilities, miscibilities, and phase diagrams are well described. For the use in life science modelling such direct property calculations may be of considerable value. All properties are calculated by surface integrals and hence logarithmic partition coefficients and hydration energies can be visualised on the molecular surface. The ability to calculate solubility of drug-like compounds is of special importance. In addition, COSMO-RS offers a set of linear descriptors for arbitrary partition behaviour, which may be of great use in any QSAR study which has to do with more complicated partition problems like blood-brain, intestinal absorption, BCF, etc. Very recently, even the pKa prediction based on COSMO-RS has been validated. An outlook will be given to application to large molecules and to docking questions, based on an semi-empirical COSMO-RS parameterisation, combined with linear scaling semi-empirical codes. Fast shortcuts of COSMO-RS, which may be useful for HTS-studies will be discussed as well. |
|
Computational Chemistry in Drug Discovery: Are High Information Content Calculations Better than Low Information Content Calculations?
1:30 PM-4:35 PM, Monday, September 8, 2003 Javits Convention Center -- 1E13, Oral
Division of Computers in Chemistry |