[Xplor-nih] marvin

[John Kuszewski]

Hi,

The best way to get started with marvin is to build off the eginput  
example.

Stuff you'll need to assemble:

1.  a PSF file for your protein, of course.
2.  a shift table for your protein.
3.  a NOESY peak location table for your protein.

Regarding the shift table, it doesn't have to be in PIPP format.  I  
can also read
NMR-STAR, although you have to be careful to get the degeneracy codes  
correct.

If you want to read NMR-STAR shift tables, add the line "package  
require nmrstar" to the
top of the initMatch* scripts.  Then change the call to  
readPippShiftTable to readNmrStarShiftTable.

In the call to createShiftAssignments, you can set the various  
selections to exclude
atoms that aren't observable in your particular spectrum (eg.,  
backbone amides, if
your 3dC was collected in D2O).

Regarding the NOESY peak location table, it also doesn't have to be  
in PIPP format.
I can read Xeasy and nmrDraw's formats as well.  If you want to use  
one of those formats,
add a "package require xeasy" or "package require nmrdraw" line to  
the top, and change
the call to process3dCPippPeakTable to process3dCXeasyPeakTable or  
process3dCNmrDrawPeakTable.

You'll have to identify each column in the peak table, using the  
flags to your process*PeakTable call,
as illustrated in the CVN example.

Then you'll have to set the peak folding/aliasing flags to match your  
spectrum.  These flags are repeated
for each call to match3d.  The relevant ones are

-nonFoldedPeaksArePositive or -nonFoldedPeaksAreNegative
-signChangesUponFoldingFromHeavyatomDimension
-signChangesUponFoldingFromProtonDimension
-signChangesUponFoldingToProtonDimension
-shiftsAliasedAlongFromHeavyatomDimension or - 
shiftsFoldedAlongFromHeavyatomDimension
-shiftsAliasedAlongFromProtonDimension or - 
shiftsFoldedAlongFromProtonDimension
-shiftsAliasedAlongToProtonDimension or - 
shiftsFoldedAlongToProtonDimension

The meaning of these flags should be obvious.  The last three flags  
are a bit more complex:

-fromProtonSpectralRangePPM   [list -1.1 9.5]
-fromHeavyatomSpectralRangePPM [list 60 120]
-toProtonSpectralRangePPM [list -1.2 6.0]

(The numbers in the above lines are just examples. )  The list of two  
numbers after each flag define the chemical
shifts around which the actual aliasings/foldings take place.  For  
now, getting them right is left in your hands, unless
you're using PIPP peak location tables, which include the necessary  
information in their headers.  I'm working with
Frank Delaglio to see if we can build something general-purpose that  
could read that information out of nmrPipe
spectra, but it's not there yet.

If you don't get the spectral range flag values correct, marvin will  
have a difficult time assigning peaks with folded/aliased
positions.

You'll need to change the names of the  
output .peaks, .shiftAssignments, and .exceptions files to something  
you like.

And if you have a reference structure, replace the name of  
cvn_reference.pdb with the name of the PDB file for your
reference structure.  If you don't have one (as is usually the case,  
of course), just delete those lines.

Finally, you'll need to update the sa_pass* and summarize_pass*  
files.  The necessary changes are all just filename
changes, which should be obvious.

I'm sure you'll run into more questions as you continue.  Ask away.

--JK



On Jun 11, 2007, at 6:19 PM, gary thompson wrote:

> some questions about marvin
>
> where should I start? do i need to alter much other than
>
> 1. alters the input peak and shifts  lists psf  files and file name  
> roots
> 2.  remove opening of reference pdb files e.g . delete
>
> readPDB -fileName ./cvn_reference.pdb
>
> and
>
> initialShiftAssignmentAnalysis \
>     -pot $noe \
>     -referenceStructureFile ./cvn_reference.pdb \
>     -minLikelihood 0.9 \
>
> etc...
>
>
> also
> is there any more documentation on whats going on other than the  
> source code, paper, slides on th web  and the cvn example?
>
>
> regards and many thanks
> gary
>
> n.b. what sort of hardware were the timings you reported (60  
> processors 12 hours) made on?
> _______________________________________________
> Xplor-nih mailing list
> Xplor-nih at nmr.cit.nih.gov
> http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih