Protein folding in contact map space

We represent a protein's structure by its contact map. Our aim is to identify the unknown fold of a known sequence by minimizing a (free) energy defined in the space of contact maps. To this end, we developed an efficient method to search this space and to generate low energy maps that are also physical. We proved that the standard pairwise approximation to the free energy is unable to stabilize the native fold of a single protein against a set of carefully generated decoys.