Dear Germana,
at first I would like to mention that even to our most optimistic evaluation loop optimization beyond 12-14 residues is fairly meaningless (you have 20), around 8-10 residue long loops you have a 60-90% chance to get something significantly close to the native solution, but for longer loops it is deteriorating quickly.
But in general you should model more and more final loop conformations as the length of the loop region increases, e.g. for an 8 residue long loop i would generate on the level of 100-500 final conformations but at least 50 final conformations is required to achieve at least 80% of the accuracy provided by the 500 optimization cycle.
In summary, currently there are no methods that can deal efficiently with a 20 residue long loop optimization neither in the Modeller package nor anywhere else.
Andras
"M. Germana Paterlini" wrote: > > I generated several loop conformations using the 'loop' > routine. The top file is similar to example 3 of the manual, with > the addition of gemetrical constraints on the end of loops. > > I am refining 20 a.a. out of a 27-loop segment. > > The refined loop conformations still have residues > located in unfavorable regions of the phi/psi plot, i.e. phi> 0 (non > Gly, or Asn). Typically, there are 1-4 residues in these regions. > Only 2% of my conformations have all residues with phi<0. > Are such occurrencies typical of the refinement procedure? If not, > how can I prevent this for happening? > > Thanks > Germana > > Germana Paterlini > Dept. of Medicinal Chemistry > Minneapolis, MN 55455 > germana1@vwl.medc.umn.edu