Nearest neighbor parameters for RNA from metadynamics simulations

Nearest neighbor rankings are used to predict the stability of nucleic acid double strands. Such data for RNA may explain ribosomal frameshifting, a feature programmed e.g. in HIV, which is a way to compact genetic material. Several sets of such parameters have been determined by experimental (SantaLucia et al. Biochemistry 1998) and quantum chem- ical studies (Svozil et. al. J.Phys.Chem.B. 2010). As far as we know, no attempt has been made to obtain these parameters from atomistic simulations. The reason for the lack of such works is that free energy calculations are computationally too demanding as they require the entire free energy surface of the system to be properly sampled. With state of the art techniques, such as metadynamics, it is possible to efficiently sample the part of the phase space which is relevant for our purposes. This work is aimed at reconstructing free energy landscapes describing the melting of a set of double stranded RNA molecules in aqueous salt solution by metadynamics using a tailored, non-discrete version of path collective variables (Branduardi et. al. J. Chem. Phys. 2007) and to obtain the nearest neighbor parameters for RNA. Besides free energies the hydrogen bonding patterns and the sugar-pucker states of the RNA molecule are also analyzed.