[Biojava-l] Finding Annotation keys for parsed reports and inconsistent Blast parsing errors...

[Keith James]

>>>>> "DL" == DeAngelo Lampkin <dlampkin at xencor.com> writes:

    DL> Is there some place that lists all the keys that are stored in
    DL> Annotations for various types of reports that are parsed
    DL> (specifically BLAST in my case)? Or are we stuck with manually
    DL> going in,?  listing them all and "seeing before trying".  Say
    DL> it ain't so! :)

Hi,

The keys are currently documented in the builder which creates the
blast result objects -
org.biojava.bio.program.ssbind.BlastLikeSearchBuilder. Not the most
obvious place, admittedly!

The result objects are quite generic and it is left to the builder to
decide what to put in the annotation. It is now possible to create
typed annotations (using org.biojava.bio.AnnotationType) and the
eventual aim is to get the builders to create e.g. 'BlastAnnotation'
where the keys are better defined and described.

    DL> Also, I've been getting inconsistent errors when parsing Blast
    DL> reports. Specifically, I get an error when I try parsing a
    DL> BLAST report using the BlastParser program given in Biojava in
    DL> Anger (nice title BTW) while the Blast2HTML parser in the
    DL> demos directory (also using the BlastLikeSaxParser) runs
    DL> perfectly fine.  So the problem is probably actually in the
    DL> ContentHandler or the SeqSimilarityAdapter.  Anyway, the error
    DL> that the BlastParser program returns is:

    DL> ----------------------------- "org.xml.sax.SAXException:
    DL> Program ncbi-blastp Version 2.2.6 is not supported by the
    DL> biojava blast-like parsing framework at
    DL> org.biojava.bio.program.sax.BlastLikeSAXParser.interpret(BlastLikeSAXParser.java:241)
    DL> at
    DL> org.biojava.bio.program.sax.BlastLikeSAXParser.parse(BlastLikeSAXParser.java:160)
    DL> at BlastParser.main(BlastParser.java:43)"
    DL> -------------------------

You are getting this message because in one case you are using the
parser in strict mode (where it will explictly check for the flavour
and version of blast and die immediately if necessary) and in the
other case (Blast2HTML) the parser is set to lazy mode where it will
have a go but not guarantee anything.

See main method of Blast2HTML:

	/**
	 * Create a SAX Parser
	 */
	XMLReader oParser = (XMLReader) new BlastLikeSAXParser();
	((BlastLikeSAXParser)oParser).setModeLazy();

I believe that strict is the default. Try setting the demo to lazy for
your output and it should work.

Keith

-- 

- Keith James <kdj at sanger.ac.uk> bioinformatics programming support -
- Pathogen Sequencing Unit, The Wellcome Trust Sanger Institute, UK -