| Three-dimensional structures of proteins determined in solution by NMR spectroscopy have the unique advantage of revealing details of molecular structure and dynamics in a physiologically relevant state; however, the many tedious steps needed to solve and validate a structure make this method challenging. The barriers to NMR structure determination become higher for larger proteins whose spectra are harder to resolve. It is clear that advances need to be made in automating protein structure determination by NMR spectroscopy. The goal of my research has been to use computational methods to advance the development of high-throughput NMR spectroscopy. Accelerating and streamlining the structure determination process will enable investigators to spend less time solving structures and more time investigating challenging biomolecular systems. My goals have been to develop an automation protocol that integrates multiple steps, ensures the robustness of each step, incorporates iterative corrections, and includes visualization tools to validate and extend the results. I developed PINE-SPARKY as a graphical interface for checking and extending automated assignments made by the PINE-NMR server. ADAPT-NMR directs fast data collection by reduced dimensionality on the basis of ongoing NMR assignments. I helped develop a version of ADAPT-NMR (originally only for Varian spectrometers) for Bruker spectrometers, and I created ADAPT-NMR Enhancer as a visualization tool for validating and extending assignments made by ADAPT-NMR on either spectrometer system. I developed the PONDEROSA package to automate the next steps. PONDEROSA carries out automatic picking of 3D-NOESY peaks and iterative structure determinations with the protein sequence and the assignments as inputs. These automation and visualization tools cover almost all of the steps involved in protein structure determination by NMR spectroscopy. As a practical test of this technology, I solved the structure of the 2A proteinase from the human rhinovirus. As a side project, I built a relational database (PACSY DB) that combines information from the Protein Data Bank (PDB) and the Biological Magnetic Resonance data Bank (BMRB) and incorporates tools for structure analysis. PACSY DB can carry out complex queries that combine atomic coordinates, NMR parameters, and structural features of proteins. |
