Re: [Jmol-developers] Re: [Joelib-help] CMLRSS and WWMM and OpenMolecules

[Egon W. <eg...@sc...>]

On Friday 12 March 2004 11:07, Peter Murray-Rust wrote:
> At 10:39 12/03/2004 +0100, Joerg K. Wegner wrote:
> >very fine !
> >
> >>[crossposted, so please be considerate in replies]
> >
> >So, do we need a cross-platform mailing list ?:-}
>
> It could be that we need a chemical developers blog.
>
> >Can a RSS channel used for such things ?
>
> Yes. In general RSS is read-only so it would be coupled to a mailing list
> of some sort (Henry?)

I'm not that clear with the blog concept: a website of someone keeping a 
diary?, or a wiki like communal website?

Anyway, I agree that it is time to organize all chemistry open source 
initiative and come up with solutions which many projects deal with, e.g. 
atom typing and interoperability...

> >>We already have proof of concept for molecules, properties, spectra (no
> >>viewer yet) and crystallography.
> >
> >For a Java based visualization i would recommend the complex SGT:
> >http://www.epic.noaa.gov/java/sgt/
> >
> >I've already used it for the primitive visualization of the JCAMP-DX
> >format in JOELib (only developer interface for uncompressed formats !). No
> >JCAMP-CS support, which could be something for OpenBabel/JOELib.
>
> Christoph, Stefan and I are working on a CMLSpect viewer. We already have
> JCAMP2CML. We want to avoid the posting of JCAMP as it breaks the XML
> model. However it would be possible to convert CMLSpect to JCAMP
>
> >>For example the output of one feed could be input to another service such
> >>as JOELib which computes properties and descriptors. This is still under
> >>construction (V0.2)
> >
> >Great !
>
> ... mainly at the design stage:-) The main technical question is how one
> merges new properties into an old molecule. This is a middleware problem. I
> think it's most like to be relevant to servers at present. "Add your CML
> molecule and get some more properties. We already do this for MOPAC. I
> think that JOELib could easily be factored in - best if someone already had
> a running protocol. Thus if you had a clogp module already running as
> something like:
>
> JOEMol j = new JOEMOl(cmlMolecule);
> double logp = j.getLogP();
>
> then it would be easy to incorporate. I haven't had time to look at how to
> code JOELib and it is easiest if we have high-level routines at this level,
> especially cml2JOE and joe2CML. I have done this for CDK by passing XML
> strings. I know this is inefficient but ultimately we need to work at this
> API level. I therefore have just two calls:
>
> cdkMolecule = getCDKMolecule(cmlMolecule)
> cdkMolecule.doSomCalculation()
> newCmlMolecule = getCMLMolecule(cdkMolecule)
> merge(cmlMolecule, newCMLMolecule)
>
> and we could do the same for JOELIB.

CDK has libio to JOELib... so without the need to write anything (except the 
getJOELibMolecule() wrapper), you can already do:

joelibMol = getJOELibMolecule(getCDKMolecule(cmlMolecule))
joelibMol.doSomeCalc()
newCMLMol = getCMLMolecule(getCDKMolecule(joelibMol))
merge(cmlMolecule, newCMLMolecule)

BTW, Joerg, it would be very nice if you could have a CMLRSS set up that takes 
all current other feeds, calculates properties and puts that only as CMLRSS 
again.

Egon