On May 22, 2006, at 9:06 PM, Mensur Dlakic wrote: > Hi all, > > I am experimenting with using XPLOR for homology modeling - hope > this statement does not put off the NMR purists out there :-) When > I try to use the TAD protocol, it does not converge to any solution > even after several days. The protein is ~280 residues, and there > are approximately 9000, 1650 and 125 homology-derived NOE, dihedral > and h-bonds restrains, respectively. Is this a hopeless case with > this many restraints, or do I need to change something in the > nmr_torsion.inp that is provided with the version 2.14? > Alternatively, what is the smallest subset of homology-derived NOEs > that could still yield a solution for a protein of this size? I > tried using just N-N distances, but even that has trouble > converging if the distance cut-off is 10 A.
The ordinary NOE potential energy is not suited to homology-derived distance restraints, since a large fraction of those are wrong. A better choice would be to try the Marvin package in xplor, which is far more error-tolerant. It's designed primarily for automatic structure determination, and so the examples are geared toward that. But you can import classic xplor NOEs into Marvin, and you might have better luck then. BTW, where are these homology restraints coming from--Andrej Sali's package? > > Finally, is there a no-hydrogen topology in present version of > XPLOR that I can use instead of protein.par? All of the standard xplor topology files are included in xplor-nih. Note that you could also just delete the hydrogens from your PSF with a "delete sele (name h*) end" command. --JK > > Many thanks for your help, > > Mensur Dlakic > > ====================================================================== > ==== > | Mensur Dlakic, PhD | Tel: (406) > 994-6576 | > | Department of Microbiology | Fax: (406) > 994-4926 | > | Montana State University > | | > | 109 Lewis Hall, P.O. Box 173520 | http://myprofile.cos.com/ > mensur | > | Bozeman, MT 59717-3520 | E-mail: > [email protected] | > ====================================================================== > ==== > > _______________________________________________ > Xplor-nih mailing list > [email protected] > http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
