We briefly review our studies on the folding/unfolding mechanisms of proteins. In biological self-assembly processes such as protein folding, the number of accessible translational configurations of water in the system increases greatly, leading to a large gain in the water entropy. The usual view looking at only the water in the close vicinity of the protein surface is capable of elucidating neither the large entropic gain upon apoplastocyanin folding, which has recently been found in a novel experimental study, nor the pressure and cold denaturation. With the emphasis on the translational entropy of water, we are presently constructing a reliable method for predicting the native structure of a protein from its amino-acid sequence.