AMBER Archive (2009)

Subject: Re: [AMBER] makeDIST_RST: no map function for ambiguous restraint

From: David A. Case (case_at_biomaps.rutgers.edu)
Date: Tue May 19 2009 - 07:35:31 CDT

On Sun, May 17, 2009, Peter Varnai wrote:
>
> I use makeDIST_RST to create normal and ambiguous restraints.
...
> #
> AMBIG l5 = H4 8 H2 11 H2 17
>
> I use the 8-column format for the input file:
> # normal
> 20 3LB H1 5 VLB H1 1.8 6.0 #4.04-6.83
> # ambiguous
> 5 VLB H1 0 AMB l5 1.8 6.0
>
> The following command gives error message:
> makeDIST_RST -pdb prep.pdb -ual noe.prep -map sugar.map -rst out.noe
> # makeDIST_RST
> Currently configured for up to 5000 atoms
> Using MAP file sugar.map
> read_maplib:3LB l5 to (H4 8); (H2 11); (H2 17);
> ERROR no map function for l5 AMB :data= 5 VLB H1 0 AMB l5 1.8 6.0
>
> I obtain out.noe that has all normal restraints but not the ambiguous
> one. I checked in each case that the atom names exist in the given
> residue number and it has entry in the map file; eg, residue 8 is VLB
> that has atom named H4 which has a MAPPING line.
>
> Is this the correct way of producing ambiguous restraints? I can
> provide the necessary files if needed.

What you describe should work, but something look amiss in the code to me.
Try changing line 751 of makeDIST_RST.c from

   sscanf( rtyp, "%s", map[nm].rtyp );

to

   sscanf( "AMB", "%s", map[nm].rtyp );

You should probably also post your files if this doesn't fix things. You
might also try adding the "-debug" flag to get more copious output.

Let me know what you find: it looks to me like the code is broken, but it
hasn't been changed for a long time. Maybe you are the only one using the
code right now, or reporting errors; or, maybe I'm misunderstanding how it
should be working....

>
> A further question how to deal with a given cross-peak that was
> assigned to three atom-pairs A-X or B-Y or C-Z. The AMBIG definition
> requires to create two groups A,B,C and X,Y,Z but that would allow for
> all possible distances to contribute to the ambiguous restraint not
> only the three above.

We haven't coded this option; the idea was that the ambiguity comes from
overlap of peaks (several nuclei with almost the same chemical shift), which
would not lead to the type of ambiguity you describe. If the unwanted
combinations (e.g. A-Y or A-Z) correspond to long distances, they will be
effectively ignored anyway, so you might be able to use the current code.
But if you somehow know that A-Y is the wrong assignment even though the two
atoms are close together, that is beyond the capabilities of the current
restraint code.

....thanks for the report....dac

_______________________________________________
AMBER mailing list
AMBER_at_ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber