Figure 5

Evaluating two clusters of loop models by RDC geometry. Helix-helix loop 36–40 in the 1Q2N Z-domain ensemble [43] shows two quite different conformational clusters of models, in spite of the fact that all 10 models match the experimental RDC data that measures orientation. a) Pro38 CαH bond vectors for the two model clusters point in nearly opposite directions, matching opposite limbs of the paired RDC target curves. b) Quality comparison of the two model clusters, from the MolProbity structure-validation site [34]: the 4 bad models have Ramachandran outliers (green), many steric clashes (hot pink spikes) and sidechain rotamer outliers (gold), while the 6 good models have no clashes or Ramachandran outliers, few bad rotamers, and several hydrogen bonds (lenses of green dots). One can conclude that the RDC match for the 4-model conformation is a coincidence, and those models are incorrect.