“Protein Structure Refinement with Ensemble-Based Scoring of Experimental Restraints”
Experimental data has been critical for protein structure determination with the Rosetta biomolecular modeling software. However, a key limitation in the way Rosetta handles that data is that a single structure must satisfy all data simultaneously. This in stark contrast to the physical reality that experimental data is derived from an ensemble of different protein conformations. The single structure approximation can therefore introduce artifacts and inaccuracies in the resulting protein structure models. The proposed work will develop computer code in Rosetta that enables scoring and optimization of multiple conformations simultaneously with experimental restraints or other score terms that are based on the entire ensemble and not a single structure alone.
We expect this will enable Rosetta to extract dynamics information from experimental data sources that explicitly capture structural heterogeneity. Such techniques include nuclear magnetic resonance (NMR), electron paramagnetic resonance (EPR), X-ray crystallography, electron microscopy (EM), X-ray scattering (SAXS), cross-linking mass spectroscopy (XLMS), etc. The accuracy of several of those techniques is limited by the lack of explicit atomic modeling of the ensemble of states the chemical labels or cross linkers can take, which could also be addressed with the proposed ensemble scoring methodology.