Hi Cathy,

Can you advise the best way to determine or estimate error in parameters such as slide, twist, roll, x-displacement for a given estimated coordinate error? (in our case refinement programs indicate the coordinate error is ~0.3 Angstroms).

All the nucleic acid analysis programs I know of, 3DNA included, calculates a set of structure parameters (slide, roll etc) based on given x-, y- and z-coordinates in PDB format and the coordinate uncertainty (B-factor) is not taken into account.

In NMR derived structures, one has an ensemble of models that fit the constraints, and analyzing all of them would give a mean/std of the structural parameters. It seems there is only one model in x-ray determined structures, and I do not know how the coordinate error in x-ray crystal structures can be directly applied to estimate errors in parameters. I would imagine that the errors in structure parameters should be relatively insensitive to coordinate error: the parameters are based on base-pair plane(s), averaged over all the base atoms whose uncertainty could conceal out.

Hope this helps a little bit: sorry for not being able to provide a more direct answer. Wilma could provide you with more insights into this issue.

Xiang-Jun