Hi Steve,
I am glad to hear that you find the predictor useful in your work. However,
I am afraid we may not be able to apply your patch as it breaks the
executable Jronn jar, it seems like the models cannot be loaded from within
the Jar file when your patch is applied. Could you tell us a bit more about
your setup? Why you cannot load the models with the old method?
Thanks,
Peter
On 9 May 2012 00:31, Steve Darnell <darnells at dnastar.com> wrote:
> Greetings,
>
> I am having problems with loading resources for the Jronn (protein
> disorder) ModelLoader. With my usage, my program cannot locate the
> modelN.rec associated with the biojava3-protein-disorder module. I know
> that my program can locate other BioJava resources, such as the Blosum65
> matrix in the biojava3-alignment module used by the SeqRes2AtomAligner
> class.
>
> I noticed that ModelLoader.loadModels() uses a different coding pattern
> for getting a resource as a stream than other areas of BioJava.
> ModelLoader.loadModels() uses ClassLoader.getSystemResourceAsStream while
> SubstitutionMartixHelper.getReader uses
> $BIOJAVA_CLASS.class.getResourceAsStream().
>
> The following change to ModelLoader.java (line 163) works in my use case:
>
> > Original
>
> final BufferedReader bfr = new BufferedReader(
> new InputStreamReader(ClassLoader.getSystemResourceAsStream(
> "model" + i + ".rec"),
> "ISO-8859-1"));
>
> < Patch
>
> final BufferedReader bfr = new BufferedReader(
> new InputStreamReader(ModelLoader.class.getResourceAsStream(
> String.format("/model%d.rec", i))
> ));
>
> It appears that using the non-system resource method in the patch code
> searches for resources in more locations that the system resource method in
> the original code (
> http://docs.oracle.com/javase/6/docs/technotes/guides/lang/resources.html).
> The existing Jronn unit test for continues to work with the patch without
> modification.
>
> I am asking for the ModelLoader class in org.biojava3.ronn (from the
> biojava3-protein-disorder module) to be modified as described before the
> next BioJava release, making the code more robust and more consistent with
> existing BioJava coding patterns. I would be happy to address any questions
> or comments.
>
> Best regards,
> Steve
>
> --
> Steve Darnell
> DNASTAR, Inc.
> Madison, WI USA
>
>
> _______________________________________________
> biojava-dev mailing list
> biojava-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/biojava-dev
>