Computational Design of Protein-Protein Interactions
Here is my prediction for the future... Researchers will find a gene responsible for an interesting phenotype, say from a standard genetic screen, and they will ask, "Hmmm, what does this gene do?" They will send their gene sequence to a computer server, which will i) accurately predict the structure, ii) identify possible functional sites based on structural features or conservation, and iii) design new proteins that will bind and block the site/s of interest. A few days later, you will receive DNA in the mail encoding your newly-designed binding proteins. (Sadly, I predict that mail service will not be quicker, and this will be the slowest part of the whole operation.) Forget making proteins to inject in mice, forget screening hybridomas or evolving antibodies... computational prediction and design will dramatically reduce effort and time towards doing functional studies with protein inhibitors. Sounds fanciful? That's what everyone thought about Michael Phelps winning eight golds at a single Olympics. There is no theoretical reason why it can't be done, and eventually it will happen!
However, clearly we are a long way from this dream, and we are still developing the methods. I work on the computational design of new protein-protein interactions, including the de novo design of new proteins with backbone geometries purpose-built to maximize complementarity to the target interaction surface. I have designed a protein that binds the active site of lysozyme to inhibit catalytic activity, and lately I have been designing proteins to bind Bcl-2 family members to initiate apoptosis selectively in cancer cells.
Check out my publications on PubMed, which in general all share the theme of investigating protein interactions and conformational changes:
Also check out my profiles on Google Scholar, ResearchGate and LinkedIn: