Can Protein Chains Lock?

Smith College, Northampton MA

Investigators

Abstract

This research explores a theoretical question that has potentially important implications for protein folding simulations. Drug design (for example, blocking the action of a deadly virus) would be significantly faster if researchers could predict computationally the folding of a protein from its sequence of amino acids, rather than having to explore options *in vitro*. However, progress on this "protein folding problem" has been slow because of the vast number of different ways a long protein can fold. These different ways are abstractly collected into a "space" of possible folding paths. The investigator and his collaborators proved a geometric theorem that, under some simplifying assumptions, implies that this space of possibilities is significantly smaller than previously thought. The goal now is to show that this reduction holds without those simplifying assumptions. In particular, the reduction in the dimensions of the space depends on the existence of "locked" polygonal chains (which model protein backbones). It is known that there do exist such locked chains with arbitrary link lengths, but it is an unsolved problem whether locked chains exist when all links have about the same length, which tracks the situation for proteins. If this question can be answered positively, the next step will be to exploit the dimension reduction to accelerate protein folding simulations, and thereby reduce the drug-design cycle time. The research is conducted at a women's college, and all five of the investigator's student collaborators are female undergraduates in areas with significant national under-representation of women: they are computer science, mathematics, and engineering majors. Several are headed to graduate school, and their training on this grant serves as invaluable preparatory experience.