Research Projects

The recipe for perpetual ignorance is: be satisfied with your opinions and content with your knowledge.

Elbert Hubbard

Computational Protein Design

My graduate thesis project deals with the problem of rational protein design. Over the last decade, techniques have evolved to either improve current catalytic activity or produce novel activity in proteins. Directed evolution and catalytic antibodies are the first examples that come to mind. While both of these techniques harness the same power of evolution, they both lack physical insight into the reasons behind the mutations that occur.

"Rational" design is distinct from these techniques in that it attempts to choose mutations based on an underlying physical understanding of the system. Such design is useful because success or failure indicates the accuracy of our knowledge of the system. This is especially important for proteins, where the basic physical principles that govern their structure are still under debate.

Computational protein design takes rational design a step further, attempting to automate the process. By developing an accurate energy function that is able to quickly approximate the relevant physics of a protein system, we can highlight mutations that may have the desired effect. With search algorithms such as Monte Carlo and Dead-End-Elimination, we can find reasonable minima in the design space.

The field of protein design may not be young, but its potential is still mostly untapped. Computational design allows us to expand the applications of protein design to systems that were previously intractable by directed evolution or catalytic antibodies. While it may not be as effective as simple evolution in all cases, it is more widely applicable and teaches us more about the real physics.

Relevant Papers

Enzyme Design

Specifically, I wish to design an enzyme active site into an otherwise inactive protein. Many reactions used in organic synthesis of popular drugs are necessarily carried out under harsh conditions of high temperature and pressure because efficient enzymes have not been discovered. This does not mean such enzymes cannot exist, but simply that nature has had no need to evolve them.

This is not unreasonable. Even with the natural battery of 20 amino acids, non-natural catalytic activities can be produced by evolving antibodies against unusual transition-state analogs. Nature also chose proteins as their primary enzyme scaffold because of their immense versatility and activity. It would be foolish to ignore hundreds of thousands of years of evolutionary choice.

The Hellinga lab at Duke University was the first to ever publish a computational design of a novel active site into an inactive scaffold protein, Ribose Binding Protein. These periplasmic binding proteins are a heavily studied family that has many properties that lend themselves towards computational design. While this breakthrough is exciting, there are still many areas of possible improvement, and many more reactions to catalyze. However, for those interested, this accomplishment is a great place to start learning about what computational protein design can offer!

Relevant Papers

The Gene Responsible for ODDD

During my undergraduate education at Johns Hopkins University, I worked for Dr. Simeon Boyadjiev trying to clone the gene responsible for a disorder called Oculodentodigital dysplasia (ODDD). My work primarily involved mutation-screening and computational analysis of the human genome in order to identify candidate genes. While I did not find the gene myself, I am happy to report that the gene was found soon after I left for graduate school.

Relevant Papers