Protein structure determination via computational methods remains a significant challenge. Modeling based on templates from related proteins has been successful for a long time in generating approximate models of protein structures but the refinement of such models to higher resolution has been very difficult. The typical problem is the refinement of an initial model with a few Å RMSD deviation to below 1 Å RMSD from what would be the correct native structure determined experimentally. Molecular dynamics simulations have had partial success in achieving such refinement but it has been unclear whether force field or sampling issues have prevented more significant refinement. Here, a recent study of refinement targets from CASP is presented where extensive MD simulations on time scales of 10s of microseconds were analyzed via Markov state modeling. The results describe the energy landscapes between the initial homology models and the native states and allow the extraction of transition pathways including energetic and kinetic information. The overall conclusion is that refinement of homology models is hindered by rough energy landscapes with significant kinetic barriers but current force fields are accurate enough for the native state to be found at the global free energy minimum.<br/>

Protein structure refinement requires sampling on rough energy landscapes

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.