What is the accuracy (compared to experiment) of the pK values computed by this server?
The answer may not be as straightforward as one may want it to be. Generally, the single-structure continuum solvent methodology used here is believed to have an error margin of an order of 1 pK unit, on average . However, the deviation may be larger in some cases (in particular, for CYS groups). In most cases you can be reasonably sure that the over-all trend (whether the pK of a given site is shifted up or down relative to the standard value in solution) is predicted correctly. As with all computational methods, computed differences are more accurate than the absolute values: for example, predicting an effect of a point mutation on the pK of a nearby site may be more accurate than calculating the absolute value of that pK. Likewise, quantitative predictions of the changes in a pK due to a well-defined local conformational change should be well within reach of the methodology. As with any pK-predicting method, lower resolution X-ray structures tend to provide worse accuracy; in partic
