Biomolecular simulations are only as good as their force fields reliably agree with experimental observables. In assessing and validating force field accuracy, it is critical to completely sample all accessible conformational space, decoupling errors in the potential function from errors due to incomplete sampling. This is particularly challenging for nucleic acid force fields, because their high charge and high conformational variability makes convergence difficult to achieve for large structures. Here, we use an RNA tetranucleotide, GACC, which shows significant conformational variability. We generate the complete conformational ensemble using multidimensional replica exchange methods (M-REMD), implemented in AMBER's accelerated GPU code. Using large scale GPU resources such as NCSA Blue Waters and XSEDE Keeneland and Stampede, we are able to perform 50-80 �s aggregate sampling, and show convergence within and between independent runs. This method allows us to converge multiple runs with different force fields and compare resulting conformational ensembles to experimental data. The results of this work advance the level of simulation required to validate force fields, as well as provide complete conformational ensembles which will be used to propose modifications to bring current force fields in better agreement with experiment.

Evaluating force field accuracy of RNA dynamics using enhanced sampling methods

248th National Meeting (2014)

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