The influence of different modes of search-model preparation on success. (a) Successful search models are found covering between 4.2 and 100% of the target sequence, with the highest success rate in the range 21–40%. The trace indicates the mean coordinate error, expressed as Cα r.m.s.d. compared with the deposited structure, of ensembles in each bin. (b) Successful search models are found covering between three and 116 residues of the target sequence, with the highest success rate in the 15–40-residue range. (c) Successful search models include both longer and less accurate and short but very accurate ensembles. Accuracy is estimated from the r.m.s.d. over all Cα atoms of the first member of the search-model ensemble. This is an underestimation of the accuracy, since it does not include other models in the ensemble which may better represent certain regions of the native structure.