Dear GAry,
Answer interspersed below.
All the combinations arising from a PeakContrib are assumed to be
possible. If you know that some of the combinations are impossible, you
should have separate PeakContribs so that all the combinations made sense.
As for weights, weights of zero would mean that you have a lot of
possibilities but do not know which one(s) are true. You would normally
set weights only if you really thought that all the possibilities were
there.
The PeakContrib is to allow you to give independent combinations of
assignments, instead of assigning each dimension individually. Where it
would really matter is with NOESY spectra. You might have a peak that, had
three possible assignments: A47H-T44Hb, L102H-S99Ha, I156H-M172Ha - this
might be because of filtering, or because it was e.g. a 4D HSQC-NOESY-HSQC
peak. You would want your constraint list to have only these three
constraints, not the eight constraints you would get if you just assigned
the dimensions individually. The effect is the same everywhere: if you
want to represent precisely what you know about possible assignments, you
should have separate PeakContribs so that all the possibilities were
actually real. Of course you are doing the assignment, so it is up to you
how precise you need to be. If you can get the correct reslut quicker by
leaving things less precise, you can choose to have only one PeakContrib
in e.g. TOCSY spectra.
Tim is offline now, but if there is anything in Analysis that should be
added I am sure he will do taht when he is back.
Yours,
Rasmus
Individual answers below.
On Tue, 2 May 2006, Gary S. Thompson wrote:
> Rasmus Fogh wrote:
>
> >Dear Gary,
> >
> >If you have two chains that are not symmetrical, you should assign the two
> >chains explicitly. So, a peak that could come from either chain (or, in
> >practice, both) should be assigned as ambigous, with two resonances on
> >each dimension.
> >
> Yes I am doing this so I have resonances for chain A H and chain A N and
> chain B H and chain B N. However, should these reonances be in separate
> PeakContribs?
>
> so using a slightly mangled syntax, consider the case below
>
>
> peak[1].peakDims[1].peakDimContribs[0] = <PeakDimContrib for reonance A H >
> <PeakDimContrib for reonance A H >.resonance = <resonance A H>
>
> peak[1].peakDims[1].peakDimContribs[1] = <PeakDimContrib for reonance B H >
> <PeakDimContrib for reonance B H >.resonance = <resonance B H>
>
> peak[1].peakDims[1].peakDimContribs[0] = <PeakDimContrib for reonance A N >
> <PeakDimContrib for reonance A N >.resonance = <resonance A N>
>
> peak[1].peakDims[1].peakDimContribs[1] = <PeakDimContrib for reonance B N >
> <PeakDimContrib for reonance B N >.resonance = <resonance B N>
>
> now should I have
>
> peak[1].PeakConribs[0] = [<PeakDimContrib for reonance A H >,
> <PeakDimContrib for reonance B H >, <PeakDimContrib for reonance A N >,
> <PeakDimContrib for reonance B N >]
No.
> or should I have
>
> peak[1].PeakConribs[0] = [<PeakDimContrib for reonance A H >,
> <PeakDimContrib for reonance A N >](with weight 0.5)
> peak[1].PeakConribs[1] = [<PeakDimContrib for reonance B H >,
> <PeakDimContrib for reonance B N >](with weight 0.5)
Yes.
>
> where A and B are chain identifiers and H and N are atoms..
>
> Further to this
>
> 1. if I have a peak which I know is residue 13 but I don't know what the
> chain assignment for the peak is what should my my setup be then, whould
> it be:
>
> peak[1].PeakConribs[0] = [<PeakDimContrib for reonance A H >,
> <PeakDimContrib for reonance A N >](with weight 0.0)
> peak[1].PeakConribs[1] = [<PeakDimContrib for reonance B H >,
> <PeakDimContrib for reonance B N >](with weight 0.0)
Yes.
> 2. do all assignments arising from the combinatorial nature of a
> peakContrib have to be reasonable? Thus for example you can have an
> overlapped hcch-tocsy peak where the 13C and 1H resonances are
> overlapped but you know for example what some of the indirect proton
> shifts come from one residue and the rest from another. What should you
> do then two separate peakcontribs with the subsets in them and weights
> of 0...
Ideally yes.
> 3. (most probably for the ccpn team) This is clearly complicated and I
> am not sure i follow the implications and reasoning in all cases. Can
> you expand on why the peakcontrib is there and what it's practical uses
> are in the round covering the common cases...
See beginning of mail.
>
>
> regards
> gary
>
> > A bit of a pain, i know.
> >
> >Yours,
> >
> >Rasmus
> >
> >---------------------------------------------------------------------------
> >Dr. Rasmus H. Fogh Email: [log in to unmask]
> >Dept. of Biochemistry, University of Cambridge,
> >80 Tennis Court Road, Cambridge CB2 1GA, UK. FAX (01223)766002
> >
> >On Tue, 2 May 2006, Gary S. Thompson wrote:
> >
> >
> >
> >>Dear All
> >> I am assigning a dimer which in its unbound state has two fold
> >>symmetry. However, once I bind a ligand to it some of the symmetry (but
> >>not all) is lost. Thus I end up with peaks in my 1H-15N hsqc some of
> >>which are for A subunits and some of which are for B and some of which
> >>are for A and B. In the case of the A+B peaks degenerate peaks should
> >>each pair orf resonances from the two two dimensions and a sigle chain
> >>be in separate peakContribs....
> >>
> >>Also another question which is related. Programmatically what is the
> >>quickest way to identify sets of identical chains in a molecular system?
> >>Do I need to do a sequence comparison ever time I search for a repeat of
> >>the same chain...
> >>
> >>regards
> >>gary
> >>
> >>--
> >>-------------------------------------------------------------------
> >>Dr Gary Thompson
> >>Astbury Centre for Structural Molecular Biology,
> >>University of Leeds, Astbury Building,
> >>Leeds, LS2 9JT, West-Yorkshire, UK Tel. +44-113-3433024
> >>email: [log in to unmask] Fax +44-113-2331407
> >>-------------------------------------------------------------------
> >>
> >>
> >>
> >
> >.
> >
> >
> >
>
>
> --
> -------------------------------------------------------------------
> Dr Gary Thompson
> Astbury Centre for Structural Molecular Biology,
> University of Leeds, Astbury Building,
> Leeds, LS2 9JT, West-Yorkshire, UK Tel. +44-113-3433024
> email: [log in to unmask] Fax +44-113-2331407
> -------------------------------------------------------------------
>
|