Not entirely sure what you are looking for, but the ROSETTA design method is one of the most successful so far for designing new active sites in enzymes:

1. Create a theozyme, a collection of the active site residues in space that stabilize the transition state (for enzymes) or ground state (for antibodies, for example)

2. Check to see if the theozyme will match in anything within a scaffold library of stable protein structures

3. If a match is found, mutate all surrounding residues in the active site to alanine that are not part of the theozyme and move the backbone slightly to maximize interactions

4. Mutate the the alanine residues made in the active site using a Monte Carlo method

This method has been used to successfully redesign several enzymes using a modification of the standard ROSETTA forcefield with geometrical constraints from the theozyme defined in step 1. It is likely this approach may have problems when the active site includes a loop where the structure may vary for different substrates. I have some ideas for this problem using our new program EvoDesign http://nar.oxfordjournals.org/content/41/W1/W273.long

you can send me a confidential message if you want if you want to discuss more details.

http://www.plosone.org/article/info:doi/10.1371/journal.pone.0019230