Here, we study if it is possible to reconstruct a protein's Cα trace solely from a CN vector or an HSE vector. These vectors are obtained from the protein's known native state and our goal is therefore to evaluate the information contents of these measures. This problem could become important for de novo structure prediction, for example if predicted HSE values are used as restraints. Preliminary results show that the HSE measure can be predicted with reasonably high accuracy[11]. Reconstruction of a protein structure from a predicted HSE vector might thus be an attractive way of approaching the sequence-to-structure problem. Predicted CN-/HSE vectors are expected to have errors compared to the exact vectors. The results presented in this paper are based on exact vectors and therefore provide an upper bound on the information contents of predicted CN-/HSE vectors. If protein structure prediction was carried out on a predicted HSE vector only, it is expected that the results would not be better than the results presented in this paper. It would therefore be natural to add other predictable information such as secondary structure, radius of gyration etc. to a structure prediction system using predicted HSE vectors. The problem of reconstructing protein structure from vectors of one-dimensional structural information has been studied before. Kinjo et al.[12] used exact vectors of secondary structure (SS), CN and residue-wise contact order (RWCO) together with refinement using the AMBER force field to reconstruct native like structures. Their results show that SS and CN information without the use of RWCO is not enough to reconstruct native like structures. Unfortunately, prediction methods for the RWCO measure only have moderate performance as compared to SS and CN[12].