Re: [Biojava-l] problem with AnnotationBuilder?

Matthew Pocock Mon, 08 Dec 2003 06:16:13 -0800

Hi francois,

The documentation here is not good. The default behavior of the annotation builder is to simply replace the properties as it goes. If you want list-building behavior, then you can configure this. For example, create a new AnnotationType where you explicitly set the type of NAME to being an un-bounded list of strings.

AnnotationType annT = new AnnotationType.Impl();
PropertyConstraint c_string = new PropertyConstraint.ByClass(String.class);
annT.setConstraints("NAME", c_string, CardinalityConstraint.ONE_OR_MORE);

Now, when top.type.setProperty() is invoked, it will end up routing each value to an object that knows that the values in the "NAME" slot are members of a list, so each value will be appended.

For examples, look inside org.biojava.bio.program.formats, and if you are brave look at the deffinition of AnnotationType.setProperty() and the things it may defer to.

Matthew

Francois Pepin wrote:

Hi everyone,

I think that there might be a problem with AnnotationBuilder (parsing
off Kegg for the curious).

With the following parser:
     LineSplitParser tvp = new LineSplitParser();
     tvp.setEndOfRecord("///");
     tvp.setSplitOffset(12);
     tvp.setContinueOnEmptyTag(true);
     tvp.setTrimTag(true);
     tvp.setTrimValue(false);
     tvp.setMergeSameTag(true);

The simplest of AnnotationBuilder:
AnnotationBuilder tvl=new AnnotationBuilder(AnnotationType.ANY);

and the following text (among others):
NAME        aldehyde dehydrogenase (NAD)
           CoA-independent aldehyde dehydrogenase
           m-methylbenzaldehyde dehydrogenase
           NAD-aldehyde dehydrogenase
           NAD-dependent 4-hydroxynonenal dehydrogenase
           NAD-dependent aldehyde dehydrogenase
           NAD-linked aldehyde dehydrogenase
           propionaldehyde dehydrogenase

I end up having the following value when printing the Annotation:
NAME=propionaldehyde dehydrogenase.

Echo() shows that everything is being read properly:
1  NAME {
2    aldehyde dehydrogenase (NAD)
2    CoA-independent aldehyde dehydrogenase
2    m-methylbenzaldehyde dehydrogenase
2    NAD-aldehyde dehydrogenase
2    NAD-dependent 4-hydroxynonenal dehydrogenase
2    NAD-dependent aldehyde dehydrogenase
2    NAD-linked aldehyde dehydrogenase
2    propionaldehyde dehydrogenase
1  }

Adding the following code (*) in AnnotationBuilder (from line 133), to
check if the values indeed get to be overwritten.

public void value(TagValueContext ctxt, Object value) {
   try {
     Frame top = peek(annotationStack);

*      if (top.annotation.containsProperty(top.tag))
*          System.out.println("replacing"+
*top.annotation.getProperty(top.tag)+ " by "+value);

     top.type.setProperty(top.annotation, top.tag, value);
   } catch (ChangeVetoException cve) {
     throw new AssertionFailure(cve);
   }
 }

This gives us the very interesting output:
replacing aldehyde dehydrogenase (NAD) by CoA-independent aldehyde
dehydrogenase replacing CoA-independent aldehyde dehydrogenase by
m-methylbenzaldehyde dehydrogenase replacing m-methylbenzaldehyde
dehydrogenase by NAD-aldehyde dehydrogenase replacing NAD-aldehyde
dehydrogenase by NAD-dependent 4-hydroxynonenal dehydrogenase replacing
NAD-dependent 4-hydroxynonenal dehydrogenase by NAD-dependent aldehyde
dehydrogenase replacing NAD-dependent aldehyde dehydrogenase by
NAD-linked aldehyde dehydrogenase replacing NAD-linked aldehyde
dehydrogenase by propionaldehyde dehydrogenase

Basically every line gets to be overwritten, so only the last one
remains at the end.

Any ideas how this could be fixed, or did I do something stupid
somewhere?

Thanks,

Francois


_______________________________________________
Biojava-l mailing list  -  [EMAIL PROTECTED]
http://biojava.org/mailman/listinfo/biojava-l

_______________________________________________
Biojava-l mailing list  -  [EMAIL PROTECTED]
http://biojava.org/mailman/listinfo/biojava-l

Re: [Biojava-l] problem with AnnotationBuilder?

Reply via email to