| Molecular Dynamics (MD) simulations of DNA provide time-resolved atomic-level description of molecular motions not readily available in experiments. Exponentially growing performance of modern computers allows for longer MD trajectories and bigger molecular systems. As simulations approach the microsecond time scale it is extremely important that theoretical tools at hand are also kept up-to-date with the increasing supercomputer power.; Cornell et al. (parm94) (Cornell and others 1995) and Cheatham et al. (parm99) (Cheatham and others 1999) AMBER force fields are widely used for the simulation of nucleic acids and proteins. These force fields show good overall performance, however, longer simulations (10 ns and higher) are often subject to irreversible backbone transitions which cause severe distortions of the DNA structure (Varnai and Zakrzewska 2004).; A refinement of the AMBER force field, parmbsc0, has been recently introduced by Modesto Orozco's group in Barcelona (Perez and others 2007) in order to compensate for stability issues. The preliminary results suggest that simulations using parmbsc0 are stable on the 10-200 ns time scale. These results, however, require further validation and this is in part the subject of my thesis research (Chapter 4). To test stability of the long (∼100 ns) parm94 trajectories and the capabilities of the new force field I provide a detailed comparison of parmbsc0's performance to the old Cornell et al. version. I give a full analysis of the 100 ns parm94 DNA simulation and relate my results to experimental data.; My participation in the following collaborative projects allows me to discuss: (1) A new method of probability analysis for the comparison of MD trajectories (Chapter 5); (2) The sequential mechanism of A->B transitions (Chapter 6). |