Since the first molecular dynamics (MD) simulation of proteins in 1977, the technique has achieved tremendous progress. With the recent investment in specialized architecture and algorithm, MD simulations are now approaching the millisecond time scale, and are much more relevant to biology than ever before. In light of this exciting technological advancement, I believe that the time has come for MD simulations to tackle a central problem in molecular biophysics - mapping the protein conformational changes at atomic level. In this talk I will describe my recent effort to study the conformational change of an ion channel during its gating transition, in which I adopted a multiscale approach that combines coarse-grained calculations based on the mixed elastic network models and MD simulations with the string method. I intend to further improve these methods and apply them to another protein system - the ATP-binding cassette (ABC) transporters. These proteins use the energy of ATP to actively pump substrates, such as drug molecules, across the cell membrane, and pose a major challenge in the battle against cancer, malaria, and other diseases. The transition between two alternate conformations of the ABC transporter is believed to be the key step in the transport process. To study the mechanism of this molecular machine, I will combine MD simulations and theories in statistical mechanics, aiming to obtain the major experimental observables, such as the transition rates, from the simulations.