We developed a coarse-grained yet microscopic detailed model to study the statistical fluctuations of single-molecule protein conformational dynamics of adenylate kinase. We explored the underlying conformational energy landscape and found that the system has two basins of attractions, open and closed conformations connected by two separate pathways. The kinetics is found to be nonexponential, consistent with single-molecule conformational dynamics experiments. Furthermore, we found that the statistical distribution of the kinetic times for the conformational transition has a long power law tail, reflecting the exponential density of state of the underlying landscape. We also studied the joint distribution of the two pathways and found memory effects.