These are both possible. I would write a script that automatically generates modeller top files. The top file will have a definition to pick and model loops that are not well refined in the crystallographic map. You would need a definition how to identify these regions from your experimental data.->this is a more ab initio approach
Alternatively, you could define a set of "user defined retsraints" from crystallographic data in general (you will find more about it in the manual in teh FAQ section). In this second case I would just generate many models without loop modeling, maybe putting the starting conformations in a sufficiently "loose" state so you can better explore and optimize your models ("SET DEVIATION = 6" i.e. in angstroms). ->this is a more comparative modeling approach
Both approaches assume you have some kind of rough initial model (the well defined parts of your structure)
However I have not done it, if that is what you are asking for.
best, Andras
Karsten Suhre wrote: > > Right - and what I am seeking is a way of telling Modeller which loops to > refine on the basis of christallographic data, or maybe another way to > constrain Modeller with christallographic data. I suppose there are a lot of > possible ways to do this by hand, depending on the tools you have, but as I > am still quite new to this domain, I was hoping to find someone who has some > experience with this in order not to reinvent the wheel. > > Kind regards, Karsten. > > On Tuesday 27 August 2002 18:04, you wrote: > > Hello, > > > > In case of loop modeling you need to identify the region(s) where you > > want to apply it, but after that -because it is an ab initio approach- > > all the generated models are individually optimized samples. Probably > > that is what you are looking for. > > It is automatic, beside identifying the segment, you do not need to do > > anything else. > > > > > > > > Andras > > > > Karsten Suhre wrote: > > > Hi! > > > > > > I am looking for a simple method to score a model from Modeller directly > > > against raw experimental cristallography data. The idea is to use models > > > from Modeller for molecular replacement in borderline situations - that > > > is in situations where the standard cristallographers molecular > > > replacement tools fail. Thus, the method should be automatic and not > > > require hand-tuning, so that I can generate lots of different models > > > (e.g. based on different loop refinement parameters) until I find one > > > that does the trick. Do you know about someone who already tried this. > > > > > > Thank you very much, > > > > > > kind regards, Karsten. > > > > > > -- > > > Karsten Suhre > > > Information Genetique & Structurale > > > UMR CNRS 1889 > > > 31, chemin Joseph Aiguier > > > F-13402 Marseille Cedex 20 > > > mailto:Karsten.Suhre@igs.cnrs-mrs.fr > > > http://igs-server.cnrs-mrs.fr/~suhre