Methods for Building and Refining 3D Models of RNA

Interest in RNA has grown tremendously in recent years as we uncover more and more roles for RNA in the cell. Investigation of RNA function is often hampered by the absence of even a tentative 3D structure which can guide experiments. Experimental structure determination is difficult because of RNA’s large size, high charge and flexibility, propensity for kinetic trapping, and the lack of the distinctive surface features necessary for crystallization. Computational structure prediction is challenging for mostly the same reasons. In this work, we describe three methods which are used in different ways to predict the structure and dynamics of RNA. RNABuilder is an “erector set” for constructing RNA molecules based on experiments, hypotheses, or other information known to the user. NAST quickly produces ensembles of coarse-grained molecules based on the statistics of backbone conformation. Lastly, Zephyr uses the graphical processing unit rather than the CPU to speed up conventional molecular dynamics calculations.