Allosteric regulation involves collective conformational transitions or fluctuations between few closely related states, caused by the binding of effector molecules . The states involved into the function and its regulation are inherent to the protein, in the sense that they are visited by the protein also in the absence of effector ligands. Binding leverage  makes it possible to find sites where ligand binding can shift the conformational equilibrium of a protein. It is calculated for a set of motion vectors representing independent conformational degrees of freedom, and we show that it can be used for predicting both catalytic and allosteric sites. To analyze allosteric communication between them, the concept of leverage coupling  was introduced. This measure is based on the assumption that only pairs of sites that couple to the same conformational degrees of freedom can be allosterically connected. We show that leverage coupling can be used to analyze allosteric communication in a range of enzymes and huge molecular machines such as chaperones. Because binding leverage and leverage coupling can be calculated from a single crystal structure, it was possible to develop a computational framework for predicting latent allosteric sites and determining allosteric communication in any protein .