Hello,
We have just put new releases of Analysis (1.0.12) and the FormatConverter
(1.0.15) live on the web, available from the usual spot:
http://www.ccpn.ac.uk/downloads/downloads.html
(And the FC release also available on SourceForge and on the MSD website.)
For people who already have 1.0.11 of Analysis we would recommend using
the upgrade mechanism (it's simpler), available inside Analysis (under
Project --> Upgrade).
This email is going to be rather lengthy because there are a lot of
changes. In particular I am almost certainly going to forget to mention
half a dozen things. So expect follow-up emails (e.g. from Tim and Wim)
to add to the length.
I'm first going to mention general major issues, then major Analysis
changes (and a bigger list of Analyhsis changes is at the bottom of this
email, many of which people will already have because of the upgrade
mechanism). Then I will get onto more detailed data model changes, which
is probably of interest to less people.
Because of the data model changes we expect an increase in bugs being
reported for this release. In particular, there might be lots of
"backward-compatibility" issues with existing project XML files which we
have not come across yet.
*** To be safe I would back up all existing projects. ***
Please email us any errors and even any warnings you get, except for
warnings you get from loading up old XML files. (Warnings will normally
mention the word "deprecated".)
For reasons mentioned below, the ccp package now has sub-packages. So
where you used to have, for example, projectDir/ccp/MolSystem.xml you will
now have projectDir/ccp/molecule/MolSystem.xml. Converting old data will
not (for now) remove the old files so be warned you will still have the
older XML files left around after conversion. So don't get confused by
this.
In Analysis the biggest change is that you can now use precalculated (or
"stored") contour files instead of having contours calculated on-the-fly.
This will be useful in various situations, e.g. you are on the road with a
portable and cannot bring all your data along, or you have so much data
that you want better performance. Having these precalculated contours
means that the raw spectrum data files do not have to be loaded into
memory, which saves memory and time. The downside is that the levels are
fixed, and worse, you cannot pick peaks. (Of course you can place them by
hand still.) So the general idea will probably be to first pick peaks and
then create the precalculated contours and go from there.
The semantics between on-the-fly and precalculated contours is a bit
tricky. We've tried to set it up to do what we think people probably
want. But we welcome feedback on this new feature. One compromise we
made (in order to save the files getting bloated in size) is to lump the
contours into positive and negative contours, and not, for example, to
distinguish between different positive contour levels. (This is the way
ANSIG did it as well.) So color schemes will not work the way you might
expect (i.e. only the first color will be used).
This new feature is available under the Experiment --> Edit Spectra
dialog, under the "Contour Files" button. (We have also re-organised the
columns in that Edit Spectra dialog to make the more useful things appear
more at the left, but that is a separate matter.)
Another change, which Sparky people will welcome, is that you can now move
the peak annotation labels. Again, we have done something which we hope
will prove useful, but we welcome feedback here as well. There is also a
pointer from the label to the peak if they are far enough away from each
other (where "far enough" is currently hard-wired to something or other,
that will no doubt become more flexible in future).
To move a label by hand you select the peak and then use the q key while
pointing at where you want the location to be with the mouse cursor.
There is also an "auto" move option under the right mouse menu Peak -->
Label Auto Arrange, and the positions can be reset to the default under
the right mouse menu Peak --> Label Position Reset.
A couple of known issues with labels. The font size does not yet change,
that will happen in the next release but only for Tk and not for OpenGL.
And for some reason, in the OpenGL version the text gets completely
clipped if the beginning of the text appears to the left of the window, I
cannot see why yet (each character is being drawn separately, since that
is the glut mechanism for drawing text, so you would have thought that
each would be clipped independently). (This problem was not noticed in
the past because the text used to be right of the peak.)
There is also now some support for pseudo-3D experiments, where you have a
bunch of 2D experiments all collected together in one (pseudo)-3D file.
(And I guess pseudo-4D ought to be possible as well!) For now this is
only available for imported Azara data because I don't know how to
recognise this situation in other formats. (Any help welcome on that
score.) This is early days on this so expect further improvements in
time.
As mentioned, these releases involve some major changes to the data model.
In the forthcoming 1.1 version of the data model and API, most links with
hicard != 1 (e.g. freqDataDim.dataDimRefs) will no longer be (ordered)
lists but (unordered) sets. So we recommend already changing code from
the style:
freqDataDim.dataDimRefs[0]
(which will be illegal in 1.1) to:
freqDataDim.findFirstDataDimRef()
There is a new haddock package which Tim is best placed to explain.
There are now "top object" classes in all packages. In particular, for
example, there is a new "top object" class, NmrProject under Project which
is the ancestor of all Nmr package classes. So instead of doing:
project.nmrExperiments
to get all (NMR) experiments you should now do:
project.currentNmrProject.experiments
More generally there is project.nmrProjects, but for now we are assuming
only one nmrProject per project, and the easiest (and hopefully future
compatible) way of getting at that is to use currentNmrProject.
Similarly there is now a package AnalysisProject which is the ancestor of
all Analysis package classes. Some Analysis data that used to be
application-specific data on project are now attributes/roles on
analysisProject.
Further, all the ccp packages have had an extra level added into their
name at the top. This was because ccp by itself was getting rather
cluttered. So there are now four sub-packages directly under ccp:
general, lims, molecule and nmr. The list of which packages they contain
is:
ccp.general: Affiliation, Annotation, Citation, DbRef, Instrument,
Method, Taxonomy, Template
ccp.lims: Classification, Crystallization, ExpBlueprint, Experiment,
Holder, Location, Protocol, RefSampleComponent, RiskAssessment, Sample,
Target
ccp.molecule: ChemComp, ChemCompCharge, ChemCompCoord, ChemCompLabel,
ChemElement, LabelledMolecule, MolStructure, MolSystem, Molecule,
Stereochemistry
ccp.nmr: Nmr, NmrConstraint, NmrEntry, NmrExpPrototype, NmrReference
And note that MolStructure used to be Coordinates, NmrEntry used to be
BmrbEntry, and Stereochemistry used to be RefStereochemistry.
The memops and ccpnmr packages have remained as they were.
Wayne
Analysis changes:
Updated Analysis to deal with new and forthcoming changes to the data
model. In particular all data model packages now have top objects. There
is also a new package for HADDOCK projects.
Added the ability to create stored (precalculated) contour files. This is
useful if disk space or speed (because of memory) is an issue. (Before
now all contours were calculated on-the-fly.) This is available from the
Experiment -> Edit Spectra dialog, with the "Contour Files" button.
Added the ability to move peak annotations (labels). With a peak
selected, you press 'q' and the annotation will move to the location
pointed to by the mouse cursor.
Added functions and menu options for auto peak label arranging ('w' is the
keyboard shortcut), and resetting to the default.
Added a pointer from the peak annotation (label) to the peak center if
they are far enough apart.
Added 1D window slice options to the window section of the right mouse
menu.
Added a confirmation upon peak list deletion.
When viewing constraints, added annotation '*' for fixedResonances that
represent more than normal atoms sets (e.g. whole residue in HADDOCK)
Added attempt to resolve prochirals (doesn't deal with ensembles yet)
for violation estimation
Added better quick peak annotations where there are multiple chain codes.
Added functions to create a transpose peak list from an existing one. This
is accessible from the Crosspeaks -> Current Peak Lists dialog.
Spectrum window strips can now be horizontal instead of vertical (the
default).
Added intra-residue filter option in the edit assignment popup.
Added a system macro to remove all marks and rulers - takes the 'n' key
shortcut.
Added more options for displaying nucleotides in the atom browser, so that
you can display base and sugar atoms on separate rows.
Added option and popup to follow shift changes, e.g. during titrations or
temperature series.
Added option to deassign a given dimension number of the selected peaks,
in the Crosspeaks -> Edit Peak List dialog.
Added option to only use assigned peaks when calculating the reference NOE
intensity.
Added option to use NOE sum or min dist for calculating the structural
value for distance constraints, and when displaying distances on the
structure viewer.
Added propagation of experiment type to selection in top table of the
Experiment -> Experiment Types dialog.
Added options to create synthetic peaklists. These include synthetic NOEs
from a structure, a peaklist using shifts from correlated resonances (e.g.
make 2D 13C-HSQC from higher dimensionality assignments), and a function
to make a transposed peaklist from an existing one.
Rates analysis can now be done using pseudo-3D data files from Azara.
Correspondingly the experiment series dialog has been upgraded so that it
can deal with either separate experiments per sample or with pseudo-3D
experiments (where one dimension is sampled).
Change to give peak dimensions for sampled data an annotation.
Added use of ButtonScrollbar to change orthogonal view of sampled
(non-frequency) dimensions.
Changed default display of strip midpoints and strip numbers to on.
Changed Window -> New Window dialog so that you can choose which spectra
will be displayed (and appear in the toolbar) before the window opens.
In the structure viewer there are changes to enable display of distances
between whole atom sets rather than giving connections between every
possible atom pair.
Code migration for *Basic modules. Mostly adding findFirstObjects(),
sortedObjects() and new package heads (top objects).
Displaying spin system strips now works with bound resonances even if
they're not in the same spin system. (Esp. important for N-CO)
Obtaining NOE distances from peak intensities is now done in a generalised
manner using an r^-6 curve (although bins still exist).
First version of Network anchoring module.
Windows should pan and zoom more smoothly with the mouse
