Egon I've sent you this message + the 2 screenshot described bellow.
Don't know if you received it.. Anyway my message never reached the
cdk user list, so I try again without screenshot attached :)
Quoting Egon Willighagen <egon.willigha...@gmail.com>:
On Tue, Mar 3, 2009 at 11:14 AM, Vincent Le Guilloux
<vincent.le-guill...@univ-orleans.fr> wrote:
First of all, the 2D depiction seems to be slightly broken: sometimes
two atoms overlap, making one of them disapearing (it can be seen when
resizing the Jtable column for example). Anyone knows about that? Is
the code bellow up to date? Should I use any other class etc. ?
Please attach screenshots.
overlap.png shows an example. The circle on the top molecule shows
where the overlap occurs. Here there should be an oxygen, just like in
the molecule bellow. When resizing the cell, sometimes it's the oxygen that is
drawn, sometimes it's the fluorine. And some other times, both atom
names are drawn.
Also some other weird things happens when I use the sliders to
navigates through molecules of the table (as seen in ss2.png). Some
molecules (not all) appears broken when I release the slider, but when
I resize the column (-> repaint), they go back to normal... o_O
Quoting Egon Willighagen <egon.willigha...@gmail.com>:
Just delete the original 2D coordinates first:
for (IAtom atom : molecule.atoms()) atom.setPoint2d(null);
Isn't it a waste of time? I mean, if 2D depiction coordinates are
already set, obviously it's a waste of time to recalculate them... Im
I missing something? My point was to know if the current algorithm
detects such case (Z = 0), and in which case wouldn't recalculates 2D
coordinates.
Quoting Egon Willighagen <egon.willigha...@gmail.com>:
For large sets you can best use the IteratingMDLReader, but I do not
think JTable supports caching, so you will run out of memory there
before reading the MDL will be a problem, I think.
When I send the data to the JTable, they are already parsed to
IAtomContainer, so I guess there is no cache problem here. However, I
read my data from a mysql database, so a simple iteration is not that
simple in my case... Anyway I will see :)
Quoting Egon Willighagen <egon.willigha...@gmail.com>:
In my program, I will have to load a huge set of molecule, and 500 of
them will be displayed in a JTable as described bellow... I haven't
tested yet to open such a number of molecules, so I was wondering if
there could be any performance issue using the following 2D
depiction/drawing code and MDL reading? And if so, if there is any way
to improve it?
Use a table widget that support caching... that is, that only keeps in
memory the things that it depicts at that time.
It's already the case I guess... See above.
BTW, alternatively, you could consider Bioclipse, which is already
doing much of what you wrote down here. Not sure where you're end goal
is, but Bioclipse is an excellent end-user platform, where the CDK is
just a library with cheminformatics functionality.
I know, bioclipse is a really nice initiative and I will probably use
it for other tasks.
However, we are not end user only. Our in-house program have some
specific algorithm, that we are currently improving along with others
new ones, related to chemoinformatics that are not present in
bioclipse, knime ...
As far as I know, CDK is what I need as its the best free java library
still in developpement; I'm currently switching to it, taking the
opportunity to add some features such as 2D visualisation of database
in JTable that were not present previously (and many other things I
hope...).
Ok that will be it for now. Sorry for all those questions... And many
thanks in advance for your answers.
No worries.
Thanks again :)
--
Vincent Le Guilloux
Phd Student - ICOA - UMR CNRS 6005
Div. of chemoinformatic and molecular modeling.
University of Orléans
Phone: ++33 2 38 49 45 77
Fax : ++33 2 38 41 72 81
[Pièce jointe retirée : type d'origine de la pièce jointe: "image/png",
nom: "mol_broken.png"]
[Pièce jointe retirée : type d'origine de la pièce jointe: "image/png",
nom: "overlap.png"]
------------------------------------------------------------------------------
Open Source Business Conference (OSBC), March 24-25, 2009, San Francisco, CA
-OSBC tackles the biggest issue in open source: Open Sourcing the Enterprise
-Strategies to boost innovation and cut costs with open source participation
-Receive a $600 discount off the registration fee with the source code: SFAD
http://p.sf.net/sfu/XcvMzF8H
_______________________________________________
Cdk-user mailing list
Cdk-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/cdk-user
