BMRB 6470 does not show which alignment is more accurate for 2K39 and 6V5D.

For ubiquitin coordinate ensembles 2K39 and 6V5D, generalized Procrustes fits used C-alpha residues 1–70 or 1–76, and N-H coordinate order was compared with the two experimental S² lists in BMRB 6470. Both lists contained 61 shared residues for the primary comparison. RMSE and Pearson correlation selected different alignments for both ensembles; removing any one residue did not change these directions. Fixed scale sensitivity checks for 2K39 retained the RMSE direction. However, BMRB 6470 lacks list-to-model identities and experimental error values, and its recorded conditions (300 K, pH 4.7) differ from the PDB headers (308 K, pH 6.5). The analysis has no statistical confidence intervals and does not establish physical alignment accuracy, motion rates, populations, or biological causes. It reports numerical agreement for these files and proposes a future check using independently obtained, condition-matched N-H data.

ubiquitin · protein structure · coordinate alignment · NMR · order parameter · prediction error