rjr27 at cam.ac.uk
Fri Jun 15 13:04:37 PDT 2012
MR_Rosetta can do amazing things, but it might also be worth trying something more manual that has worked for us for several structures with models in the 20% range (e.g. angiotensinogen). Get a number of models from the HHpred search, use the HHpred alignment to superimpose them with phenix.ensembler and turn on the trim option (Ensemble generation settings -> Trim residues deviating more than threshold). This prunes the ensemble back to the conserved core, which (with any luck) should also be preserved in your target structure. Pruning off things that are wrong turns out to be very good for improving the signal-to-noise. After ensembler, run phenix.sculptor (using your HHpred alignment) to trim off the non-conserved side chains from the trimmed ensemble) and then try to solve it in Phaser with the resulting sculpted ensemble.
Randy J. Read
Department of Haematology, University of Cambridge
Cambridge Institute for Medical Research Tel: +44 1223 336500
Wellcome Trust/MRC Building Fax: +44 1223 336827
Hills Road E-mail: rjr27 at cam.ac.uk
Cambridge CB2 0XY, U.K. www-structmed.cimr.cam.ac.uk
On 15 Jun 2012, at 19:58, Heather Condurso wrote:
> Hello all,
> I am working on a structure for which a good molecular replacement model doesn't really exist ( highest sequence similarity is 24% for a protein with completely different function and even worse for proteins with a more similar function). As I try and solve the phase with Se-Met derived protein, I thought I would try and use MR_rosetta and see what happens. I have also tried to use pyrosetta to generate better models, but the structure alignments suggest my protein to be primarily composed of beta sheets so the models from ab initio methods aren't great. I am using the GUI and have given my sca file, a sequence file, the 3 and 9 fragment files, and both the global and local alignment hhr files. I thought that phenix would use the hhr alignments to pull models directly from the pdb, but without giving a model pdb, MR_rosetta fails. So my question is how do I input the models? Should I download each of the pdbs and clean them up a bit? How many should I include? I unde!
> rstand the more I include the more computing time this will take. Can you give any recommendations as to other parameters to change keep the computing time down initially to identify the best model before extensive time is used?
> I appreciate any help you may be able to provide,
> Heather Condurso
> UF-Department of Chemistry
> condurso at ufl.edu
> phenixbb mailing list
> phenixbb at phenix-online.org
More information about the phenixbb