From: Wayne Boucher <[log in to unmask]>
Subject: Re: Drawing real canvas ContourFile error: getting normal block data
To: [log in to unmask]
Date: Wednesday, January 20, 2010, 5:06 PM

Ah, I just went through the code again and I think it's actually ok after
all, because he is doing a loop inside a loop which means the order
doesn't matter, so it should actually do what it claims.

So I wonder about those three conditions again.  Presumably the isotopes
are ok (so both spectra are (1H, 15N).  And the measurementType I'm pretty
sure must be ok.  Which would leave the acquisition issue.  I'm not sure
why it is even checking that, because why would you care when trying to
map two spectra together.  So that's another Tim question, but meanwhile
can you type the following at the Analsyis python prompt:

>>> def f(spec):
...   for dataDim in spec.sortedDataDims():
...     expDimRef = dataDim.findFirstDataDimRef().expDimRef
..      print spec.experiment.name, spec.name, dataDim.dim, expDimRef.isotopeCodes,
expDimRef.measurementType, expDimRef.expDim.isAcquisition
...

(so just a carriage return on that last line and be careful with that long
line, it is one line not two) and then

>>> spec = top.nmrProject.findFirstExperiment(name="XXX").findFirstDataSource()
>>> f(spec)

where XXX is the name of the experiment (and this is assuming that there
is only one spectrum in the experiment).  Repeat those last two lines for
each of the spectra in question and see what the output says.

I just tried it on an HSQC in our TutorialProject and got:

HSQC 115 1 ('1H',) Shift False
HSQC 115 2 ('15N',) Shift True

and I bet the 1H was the acquisition and the 15N was not, so it just goes
to show.

Wayne

On Wed, 20 Jan 2010, Krisztina Feher wrote:

> Hi Wayne,
>
> Ok, I will skip the 15N HSQC for now and continue with 13C HSQC only. Would be
> nice to sort it out at some point - though, of course, I have some print outs
> about the assignment.
>
> Thanks again,
> Krisztina
>
> --- On Wed, 1/20/10, Wayne Boucher <[log in to unmask]">[log in to unmask]> wrote:
>
>       From: Wayne Boucher <[log in to unmask]">[log in to unmask]>
>       Subject: Re: Drawing real canvas ContourFile error: getting normal
>       block data
>       To: [log in to unmask]">[log in to unmask]
>       Date: Wednesday, January 20, 2010, 4:22 PM
>
>       Hello,
>
>       I'm just tracing through the code to see where that error message
>       comes from and why it was triggered.  For each dim in the two
>       experiments/spectra it matches three things:
>
>       (1) the isotopeCodes, which in most cases, as here, is just a single
>       isotopeCode, e.g. 1H or 15N; so this should not be a problem if they
>       both agree that the two dims are (1H, 15N)
>
>       (2) the experiment measurement type, which is "Shift" by default and I
>       doubt if the default would ever have been changed so I don't think
>       this could be a problem
>
>       (3) whether the dim is the acquisition dim or not, I forget how this
>       is set so it's possible something is going wrong here
>
>       **BUT** the code (to me) looks like it is not doing the checks
>       properly in any case.  So to me it looks like v1 code where links were
>       all ordered, and not v2 code where they are mostly not ordered, and so
>       the coding has to be more careful.  (There is also a dangerous and
>       possibly incorrect piece of Python code in there, to boot.) 
>       Basically, I'm amazed that the kind of error message you see has not
>       come up more often.  (So, the lack of ordering can by luck not be a
>       problem if both spectra happen to end up the same way.)
>
>       I'll have a go at fixing this but I ought to run this past Tim,
>       because he might have an idea why the code is doing what it is doing
>       and I might be misunderstanding something there.
>
>       Wayne
>
>       On Wed, 20 Jan 2010, Krisztina Feher wrote:
>
>       > Hi Wayne,
>       >
>       > I found the original processed file, it had some strange number of
>       points (921 x
>       > 486) with block sizes of 921 and 1. I loaded them, so Analysis does
>       not draw the
>       > lines anymore, but the spectrum is still corrupt (only some
>       horizontally smeared
>       > noise can be seen), so I copied the assigned peaks to a freshly
>       processed spectra.
>       > That solved the 13C HSQC.
>       >
>       > However when I wanted to do the same copying with the 15N HSQC in
>       the Peak Lists
>       > window with Copy Peaks command I got the following:
>       >
>       > Could not map 2nd dimension of source's spectrum.
>       >
>       > In the Peak List I see the 158 peaks assigned in both  F1 and F2
>       dimension to H
>       > and N with corresponding 1H and 15N chemcial shifts. There are also
>       no peaks drawn
>       > in the window either, they are only listed in the Peak Table. 
>       >
>       > What could be the problem with that peak table?
>       >
>       > Thanks a lot,
>       > Krisztina
>       >
>       >
>       >
>       >
>       >
>       > --- On Wed, 1/20/10, Wayne Boucher <[log in to unmask]">[log in to unmask]> wrote:
>       >
>       >       From: Wayne Boucher <[log in to unmask]">[log in to unmask]>
>       >       Subject: Re: Drawing real canvas ContourFile error: getting
>       normal
>       >       block data
>       >       To: [log in to unmask]">[log in to unmask]
>       >       Date: Wednesday, January 20, 2010, 1:34 PM
>       >
>       >       Hello,
>       >
>       >       That is likely to be the problem.  And one thing on the TODO
>       list is
>       >       to
>       >       make it easier to do this kind of thing.  So the program
>       should
>       >       probably
>       >       prompt you to check that the number of points, etc., are still
>       ok.
>       >
>       >       You can modify this in the program via the Experiments -->
>       Spectra
>       >       dialog,
>       >       in the Referencing and File Details tabs.  The tricky bit
>       might be the
>       >       block sizes.  If you're not sure about any of the data then I
>       would
>       >       not delete the existing spectrum but instead open the spectrum
>       afresh
>       >       to
>       >       see what the values are, and then copy them over.  I hope that
>       makes
>       >       sense.
>       >
>       >       Wayne
>       >
>       >       On Wed, 20 Jan 2010, Krisztina Feher wrote:
>       >
>       >       > Hi Wayne,
>       >       >
>       >       > Thanks for the quick relpy. The software can open the
>       spectrum
>       >       freshly. I transferred
>       >       > this project from another computer, I could not find the
>       original
>       >       processed file, so I
>       >       > opened with a newly processed one. I think it  must have
>       been a
>       >       different processing
>       >       > program used originally. Could that be the problem?
>       >       >
>       >       > Krisztina
>       >       >
>       >       >
>       >       > --- On Wed, 1/20/10, Wayne Boucher
>       <[log in to unmask]">[log in to unmask]> wrote:
>       >       >
>       >       >       From: Wayne Boucher <[log in to unmask]">[log in to unmask]>
>       >       >       Subject: Re: Drawing real canvas ContourFile error:
>       getting
>       >       normal block
>       >       >       data
>       >       >       To: [log in to unmask]">[log in to unmask]
>       >       >       Date: Wednesday, January 20, 2010, 12:47 PM
>       >       >
>       >       >       Hello,
>       >       >
>       >       >       This could be a couple of things.  So is that version
>       of
>       >       Analysis capable
>       >       >       of opening a spectrum from fresh and drawing it?  If
>       so it's
>       >       possible that
>       >       >       that error message is indicating that the HSQC files
>       are
>       >       truncated
>       >       >       somehow, or perhaps (although I doubt this) that the
>       endianess
>       >       is wrong.
>       >       >       If it can't open a spectrum from fresh then either the
>       >       compilation or the
>       >       >       linking of the C libraries is probably knackered.
>       >       >
>       >       >       Wayne
>       >       >
>       >       >       On Wed, 20 Jan 2010, Krisztina Feher wrote:
>       >       >
>       >       >       > Hi,
>       >       >       >
>       >       >       > I would need a bit of help with a project loading.
>       It is
>       >       older project
>       >       >       which
>       >       >       > I would like to load now into CCPN 2.1.2 on Ubuntu
>       8.10 with
>       >       all latest
>       >       >       > updates installed. The project should have a 15N
>       HSQC and a
>       >       13C HSQC
>       >       >       spectra
>       >       >       > with assignment on them. The 13C HSQC starts to load
>       and
>       >       then it draws
>       >       >       > magenta lines tighly next to each other diagonally
>       into the
>       >       lower left
>       >       >       > corner of the window, which fill up half of the
>       window with
>       >       the magenta
>       >       >       > lines and then it displays the error message all
>       over again:
>       >       >       >
>       >       >       >>>> Drawing real canvas ContourFile error: getting
>       normal
>       >       block data
>       >       >       > Drawing real canvas ContourFile error: getting
>       normal block
>       >       data
>       >       >       > Drawing real canvas ContourFile error: getting
>       normal block
>       >       data
>       >       >       > Drawing real canvas ContourFile error: getting
>       normal block
>       >       data
>       >       >       > Drawing real canvas ContourFile error: getting
>       normal block
>       >       data
>       >       >       > Drawing real canvas ContourFile error: getting
>       normal block
>       >       data
>       >       >       >
>       >       >       > The picked peaks with assignment are visible, but I
>       cannot
>       >       see the
>       >       >       spectrum
>       >       >       > peaks. Anyone know what is this about and how could
>       I save
>       >       my project with
>       >       >       > the assignment in it?
>       >       >       >
>       >       >       > Thanks a lot,
>       >       >       > Krisztina
>       >       >       >
>       >       >
>       >       >
>       >       >
>       >       >
>       >
>       >
>       >
>       > ???
>
>
>
>