Structure prediction techniques potential energy models

In protein structure prediction, potentials are used to assign an energy-like quantity to a conformation of a protein molecule. If this quantity enables us to distinguish the native state of a protein, the potential is regarded as a reasonable model for a protein-solvent system. The rationale behind this relies on two assumptions (a) a solved protein in its native state can be described by an ensemble of closely related conformations, and (b) in this state the system is in the global minimum of free energy. Virtually all techniques designed for structure prediction are based on these principles [3,4]. 

---