Structure Prediction with Experimental Constraints

Predicting the three dimensional structure of proteins given their amino acid sequence is one of the primary challenges of biochemistry. The functions of natural proteins are often dependent on their ability to adopt precise and ordered structures, and the effects of amino acid sequence on protein structure are among the fundamental selective forces defining evolution. The Rosetta structure prediction algorithm provides a generalized model of the relationships between sequence and structure, and can be used to provide physics based modeling for datasets that are insufficient for protein structure determination. In this dissertation I present methods for modeling structures from sparse experimental constraints using the Rosetta ab initio method. I introduce new protocols for incorporating disulfide bond connectivity as a constraint for use in ab initio structure prediction and in structure prediction from chemical shift data, and I demonstrate a new method for modeling structures based on the assignment of backbone chemical shifts in the presence of paramagnetic lanthanides.