| Atomic structure determination of membrane proteins is an important problem. Because of the difficulties in crystallization and traditional NMR techniques using membrane proteins, other experimental methods are being developed and investigated. One such method, the solid-state NMR PISEMA (Polar Inversion Spin-Exchange at the Magic Angle) experiment, determines orientational contraints for the target membrane protein. These constraints can be used to build a high resolution atomic model.;This dissertation presents a detailed analysis of the PISEMA experimental data set and how it can be used to derive atomic structure. One of the aspects of the data set is that it is degenerate, that is, the orientational information measured by the data does not uniquely describe atomic locations. Thus, there are many possible structures that would provide the same data. An important goal of this work is to enumerate and characterize these degeneracies throughout each phase of model building and provide computational tools to manage them.;The process of building atomic models from PISEMA data involves three major steps: assignment, initial model building and atomic refinement. For each step we present new software that is intended to aid in the model building process. The tools are designed to be used consecutively with limited, though significant, human intervention.;The last section of the dissertation presents an application of our tools to a new PISEMA data set, from which we derive the first high-resolution atomic structure of the transmembrane portion of the M2 proton channel from the Influenza A virus in the presence of amantadine. For this case, we explain both the data processing and decision making that determined the final atomic model. It is likely that this semi-automated procedure will be applicable to other transmembrane protein PISEMA data sets. |
