Atomic structures of proteins, nucleic acids, and their complexes are determined using X-ray crystallography, NMR, or cryo-Electron Microscopy. These structures are essential to understand their structure-function relationships. However, the experimental conditions are totally different from the actual cellular environments and it is hard to understand how biomolecules behave in such cellular environments, just using the atomic structures. We have recently built protein crowding systems in computers and carried out all-atom molecular dynamics (MD) simulations of the systems to understand biomolecular dynamics in the crowded environments. The largest simulations we have ever performed were the all-atom MD simulations of a bacterial cytoplasm using K computer. By analyzing the simulation trajectories, we observed that non-specific protein-protein and protein-metabolite interactions play important roles in biomolecular dynamics and stability in a cell. The new insight from the simulations is useful not only for basic life science in molecular and cellular biology but also drug discovery in future for introducing the effect of non-specific protein-drug interactions.