summary : calculation of pka valuse by ab initio methods

Dear CCLers,
 Recently, I've asked a question about calculation pKa values using ab initio
 methods (see below). For the moment I've received 3 replies:
 COSMO-RS is the only code I know that will deal with real solvents, beyond
 the zeroth order dielectric continuum...
 John McKelvey
 I am very skeptical of the ability of any current ab initio method to
 accurately calculate pKa's.  Consider acetic acid.  Its pKa is dominated by
 solvent effects and is largely determined by deltas and not deltaH.
 Wayne E. Steinmetz
 we recently published a calculation scheme for pKa (pKb will follow soon,
 prel. results can be obtained on request). We just calculated dG_diss based
 on DFT/COSMO and added the statistical thermodynamics contribution from
 COSMO-RS. We got a single very good linear correlation of dG_diss with pKa
 for a very broad range of organic and inorganic acids. It is  probably the
 most accurate "ab initio" prediction of pKa. Nevertheless there is a
 serious
 problem with the slope of the regeression line which is discussed in detail
 in the paper:
 A. Klamt, F. Eckert, M. Diedenhofen, and M.E. Beck, "First Principles
 Calculations of Aqueous pKa Values for Organic and Inorganic Acids Using
 COSMO-RS Reveal an Inconsistency in the Slope of the pKa Scale", J. Phys.
 Chem. A., 107, 9380 - 9386 (2003)
 Andreas Klamt
 Orginal question:
 Dear CCLers,
 According to you, what is the best method to calculate absolute pKa values,
 i.e. without first deriving some linear semi-empirical equation for a
 training set, but to calculate pKa directly from the Delta G of the
 dissociation reaction. I would like to calculate pKa values for some medium
 sized molecules, NOT just in water, but in mixtures of water with alcohols
 for which experimental data are very sparse.
 I am aware that the calculated solvation energy for charged species can
 contain relatively big errors, affecting Delta G and finally resulting in
 errors of several units in the pKa value.
 According to some people, this problem can partially be overcome by applying
 continuum models in which explicitly some solvent molecules are taken up.
 Is DFT accurate enough, which continuum model is best, etc...
 --------
 Theodorus de Bruin
 Institut Frangais du Pitrole
 Direction Chimie et Physico-Chimie Appliquies
 Dipartement Thermodynamique et Modilisation Moliculaire
 1 & 4 Avenue de bois Preau
 92852 Rueil-Malmaison Cedex
 Tel: +33 (0)1.47.52.54.38
 Fax: +33 (0)1.47.52.70.58