Hi Brenda, Brenda Patterson wrote: > I am doing molecular replacement with a model of sequence identity 42%. If you used a CCP4 automated MR tool to generate your model, then the "[Post MR] Fill Partial Residues" can conveniently restore side-chains that have been "Schwarzenbachered". > Have > been trying numerous combinations of weighting in refmac of CCP4. However the > geometry of the model is not being maintained. "Geometry" in the sense of things that are restrainted or those that are not? > I am then trying manual fitting > of the model in COOT. However when torsion restraints are applied in the > refine/regularisation option the model comes out of the density. Regularize and Real Space Refinement (RSR) are not the quite the same thing. > how would you tackle this please? If, when using the RSR, you think that not enough emphasis is being placed on the map, then you can change the weight (Set Matrix). > i also have developed CIS peptides in places and have found > that trying to revert these to trans is again causing the model to come out of > the density severely. Oh dear. Olden Coot? There is a trick for even the wary, Refmac will add a CISPEP card to its output where there are unlabelled CISPEPs (from the output of Coot). If there are labelled CISPEPs (but not actually in the structure) Refmac will make the peptide CISPEP. Coot and Refmac are out of step (Coot is behind, it should catch up at some stage). Currently you would have to hand-edit the output of Coot/input to refmac to remove CISPEP cards that are inappropriate. > Does anyone think it's ok not to use the torsion > restraints in COOT Yes, I do. Torsion restraints are not (yet) applied across peptide linkages. > and simply use real space refine/regularisation as it is in > COOT? > As long as your traffic lights are green[1], you should be OK. [1] http://www.ysbl.york.ac.uk/~emsley/coot/screenshots/traffic-lights-example.png Paul.