This may help:

I have been using an average model constructed from the 5 best models out of 50 by the objective function, to calculate site specific and total uncertainties. This works relatively well when using multiple templates (4). I have tested the site uncertainty to model error and they are correlated at .6 for regions which are aligned to the templates. This uncertainty is not correlated for regions in loops or gaps. I have only tested this in helix-loop-helix proteins!

I am not sure what others think, I hope this helps.

Michael Buck

************************************************ Michael Buck NCSU Genetics mjbuck@unity.ncsu.edu Phone (919)515-5759 Fax (919)515-3355 http://www4.ncsu.edu/~mjbuck *************************************************

At 11:04 AM 4/2/02 +0200, margot wrote: >Hello modelling community, > >ist there any accepted algorithm to generate coordinate uncertainties for >homology models (or ensembles of models)? If anyone knows of anything >published >on that matter I'd be grateful for the reference, as well as unpublished >insightful material. > >And another question: Is there any additional bulletin board dealing with >protein (biomolecule) modelling in a broader sense than Modeller usage? >thanks, > margot