On Friday 28 November 2008, Donnie Berkholz wrote: > On 16:04 Fri 28 Nov , Ethan A Merritt wrote: > > On Friday 28 November 2008, Ben Akiyama wrote: > > > I am measuring and comparing torsion angles in several solved > > > crystal structures of RNA helices using AMIGOS. I was wondering if > > > anyone knows of a program that can give me an idea of the error in > > > these measurements based on coordinate error, b-factors, etc. > > > > If you mean real torsion angles, the only quantitative way to do this > > is to do full-matrix refinement, invert the matrix, and derive error > > estimates from the off-diagonal terms. That is not something you can > > do after the fact from a previously refined structure. > > I'm trying to answer a related question right now. Does anyone have a > good feeling for how this method would compare with
> (1) propagated error from the coordinate esu and also with In order to propagate error estimates from individual ADPs and atomic coordinates, you have to assume that the errors are independent. But they are not. Consider the real-life scenario of a flexible loop refined using the usual set of crystallographic restraints. The uncertainty in the atomic position is large, and probably the ADPs are large to match. But the error/uncertainty in bond lengths and angles are not so large, because they have been restrained during refinement. That is, the whole loop and all the atoms in it may flop 2A to one side or to the other, but the restraints prevent the CA from flopping 2A to one side while CB flops 2A to the other. Conversely, since their are no restraints on the torsion angles, their uncertainty may be if anything underestimated by error propagation. > (2) repeated refinements using randomized coordinates against the same data > set? I suspect that it depends on the data quality, and on the individual structure. But since it is easy to conduct the experiment, why not try it out on your protein refinement of interest? -- Ethan A Merritt Biomolecular Structure Center University of Washington, Seattle 98195-7742
