[Joelib-help] Re: weka : protected vs public

[Joerg K. W. <we...@in...>]

Hi,

i'm not a 'Dr.' but i'm working on that :-)

if you like you can subscribe to the help mailing list, because i'm 
sending the aswers to this list also, to avoid answer redundancy !
http://lists.sourceforge.net/lists/listinfo/joelib-help

> I recompiled with ant, and followed your suggestion
> to create pdf.
> Excellent, PDF created with structures (along with
> atom lables). all the atoms (default c) are highlighted
> with 'square',
> a) How to keep the 'square' off from atoms?
> May be I am demanding too much?
Who cares, much more interesting is, who will implement things :-)
I've added an option to joelib.properties (next release or CVS in the 
next days):
joelib.gui.render.Renderer2DModel.drawCarbonAtoms=false
You can also use:
joelib.gui.render.Renderer2DModel.useKekuleStructure=true
which forces atom labels:

> b) How to suppress the [warning] and [info] messages?
change joelib/src/log4j.properties then recompile.
Switch 'INFO' and 'WARN' to 'ERROR' for the classes you want to suppress 
info's. For debugging you can switch to 'DEBUG', e.g. in JOEMol and you 
will get a HUGE amount of debugging messages.

> c) With smiles I could not see the structures in PDF files
> (obvious: no atom co-ordinates in smiles)
correct, you will need a layout algorithm, e.g.
- CDK, Marvin (commercial) for 2D layout
- Ghemical (force-field), Corina (expert system with torsion angles, 
commercial) for 3D layout. For Corina i've written a external JOELib 
task (or plugin) which a can publish if you want to explicitely use that 
one.

> d) how to create 'csv' files of descriptors for the molecules
> in smiles/sdf format?
Tricky and why ?
If you convert SDF->SDF JOELib will add all descriptors.
This will not work so easily for the SMILES and FLAT format.
Why ?
We need to define the output format !!!
Two solutions and SMILES (only for SMILES output) and TITLE are 
predefined key words:
1. joelib.properties:
joelib.io.types.Smiles.lineStructure=SMILES|TITLE|DESC1|DESC3
joelib.io.types.Flat.lineStructure=TITLE|DESC_WHAT_YOU_LIKE|ONE_MORE

It is clear that you can should not use atom properties, matrices and 
arrays in these formats, which can be handled by SDF, CTX and CML.

2. sh convertSkip.sh command line (more usefull for batch mode). Do not 
forget to quote these entries !!!
[+f<lineStructure>]
[+ff<lineStructureFile>]
[+s<lineStructure>]

e.g.: '+fTITLE|DESC_WHAT_YOU_LIKE|ONE_MORE'
For the flat file you can also define a file, e.g.
myDescs1.flat.txt
TITLE
DESC_WHAT_YOU_LIKE
ONE_MORE

and use
+ffmyDescs1.flat.txt


> Finally regarding weka:
> how to compile entire code?
You can dowload the Weka4JOELib weka.jar binary distribution directly from:
http://sourceforge.net/project/showfiles.php?group_id=39708&package_id=92211
If you want to use the source code you are on your own, please contact 
the Weka mailing list or read their manual or add the source to 
joelib/src and use compile (do not forget to copy the required 
properties file)
http://www.cs.waikato.ac.nz/ml/weka/

With friendly regards, Joerg

> 
> --
> M. Karthikeyan, Ph.D., Scientist
>  _|      _|    _|_|_|  _|
>  _|_|    _|  _|        _|
>  _|  _|  _|  _|        _|
>  _|    _|_|  _|        _|
>  _|      _| o  _|_|_|o _|_|_|_|
>  National    Chemical  Laboratory
>  Pune - 411 008, INDIA
>  Ph: +91-(0)20-5893 457  FAX: 5893 973
>  http://www.ncl-india.org/
> 
> 
> 


-- 
Dipl. Chem. Joerg K. Wegner
Center of Bioinformatics Tuebingen (ZBIT)
Department of Computer Architecture
Univ. Tuebingen, Sand 1, D-72076 Tuebingen, Germany
Phone: (+49/0) 7071 29 78970
Fax: (+49/0) 7071 29 5091
E-Mail: mailto:we...@in...
WWW:    http://www-ra.informatik.uni-tuebingen.de
--
Never mistake motion for action.
                          E. Hemingway