I'm guessing you should try it experimentally (I mean, predictions aside, there is no better way than developing a small experimental design testing antibody-antigen binding with and without coomassie-dyed antibodies...

Regarding the theory behind it... a few considerations:

-The coomassie interaction with proteins is non-covalent, so it covers pretty much the entire antibody... Nonetheless, an antigen specificity will cause it to still bind to the recognition site, and you are supposed to be succesful in that imunoreaction...

-Because of what I just said, coomassie stained antibodys will probably bind, and the dye will interfere, rather than inhibit... so, it won't be a matter of "all-or nothing", but more of  "a percentage of the no-comassie affinity"...

-Besides those questions, assuming you successfully bing your antibody-antigen, either fully or partially, you will then have the complex to be blue and you won't be able to detect any color development, so you must assure that detection is by chemiluminescence or fluorescence... 

Best of lucks...

Thanks Rafael, 

I am well aware of the binding mechanism, and these are the same considerations I made and represent my concerns, I was wondering if anyone tried it before, because it's so simple that I can't belive it has not been tried yet, and if so, I wonder why.  

Cheers, 

Max

 I havent checked it but I found you a reference that might shed some light on the theme...

Velvizhi Ranganathan, Prabir K. De. Western Blot of Proteins from Coomassie-Stained Polyacrylamide Gels. Analytical Biochemistry V 234 (1), 1996, pp. 102–104

cheers