On May 12, 2006, at 9:50 AM, <[email protected]> <[email protected]> wrote:
> Hello, > > I have a few questions pertaining to Marvin. > > I'm working with a protein that is 189 residues in length and I > have a total of > 6900 restraints. Has anyone had much luck with Marvin on proteins > of this size > or larger? It is rather large compared to most Marvin structures, but you have a lot of restraints, so that'll help a lot. > > 1.) Would it sound logical to increase the number of structures to > 1000 and just > take the top 5% or would it be best to increase the number of > passes? If I was to > increase the number of passes would it be best to increase the > number of > passes with the linear NOE potential energy function or the pass > with the > Quadratic NOE potential energy function? It'd be more conservative to increase the number of structures per pass, rather than the number of passes. In particular, increasing the number of pass1 structures will help. Pass3 converges quite well because of the quadratic potential shape, but requires good initial likelihoods of course. > > 2.) Is there a script that can search for symmetry partners in > both 15N and > 13C-edited NOE spectra? I presume you mean combining them, matching a C-H ... H-N peak in the 13C spectrum to an N-H .. C-H peak in the 15N. I don't do this, because the peak locations usually don't agree terribly well between different spectra. But there'd be nothing to prevent you from deciding that peak A in the 3dC is the symmetry partner of peak B in the 3dN and deleting the peak assignments that don't match from the appropriate peaks files. But perhaps my experience isn't typical. I could look into writing a script that'd handle this inter-spectra symmetry filter if there's interest. > > 3.) I've noticed that there are a few restraints with degeneracy > of 1 after the > initialMatch script that should have a higher degree of > degeneracy. I've noticed > this with my protein as well as the interleukin-4 example. These > assignments > don't have a symmetry partner either. Has anyone else noticed this? > Did your initialMatch script call the primarySequenceDistance filter? It removes peak assigns for peaks that have an intra-residue peak assignment available. If you can list specific peak names for the IL-4 example, I'd be happy to look at them. --JK
