What do you think about the use of catalytic antibodies to design active sites for new enzymatic reactions. Then illucidating the structure and sequence of these sites and using the data as a template for the design of catalytic sites on proteins. These would obviously have pretty low rate enhancements initially (~10^4) or so. But they could probably be increased significantly through the use of guided evolution or even maybe some inginuitive fidling. It's an idea I've been fidling with. I know it's most likely being done now already, and i'll have a look around. But I'm curious on people oppinions on the idea.

Here's an article you may be interested in:

 

Dwyer, et al. (2004). Computational Design of a Biologically Active Enzyme. Science 304: 1967-1971.

 

Basically, they converted ribose binding protein into a triose phosphate isomerase (a glycolytic enzyme). The approach Dwyer, et al. used could possibly be extended to what you're thinking of since they first designed and obtained a protein which could bind their substrate, then designed the catalytic site afterward. Of course, designing a novel catalytic site from scratch is a daunting task. Modifying an existing catalytic site to work on a new substrate would be easier, but it's still a pretty difficult task.

Right but the bonus about desinging a catalytic antibody to the transition state of the reaction is that it comes with a substrate binding site that is already catalytic for the reaction.