Parsing Blast Results In Java

Question

[weslfield]

Hey guys, so I am in the process of creating a Java app with a GUI and all based on metagenomic analysis. So my question revolves around running BLAST searches and returning the results in the most efficient manner. I have been using Biojava3 for some other functionality in my software, but the BLASTing seems a little bit of an issue, especially with the most recent release (BioJava3 doesn't have a fully developed BLAST parser yet). For my purposes, I only really need the IDs of the hits to extract phylogenetic information from them, I don't need the sequences of the hits themselves. Can someone guide me to the best way to accomplish this? I have done it using Perl/bioperl very easily, but am struggling in Java. Thanks for the help and suggestions!

Answer 1

[Pierre Lindenbaum]

generating a java parser for blast+XML ? easy:

${JAVA_HOME}/bin/xjc  -p blast -dtd "http://www.ncbi.nlm.nih.gov/dtd/NCBI_BlastOutput.dtd"

and you're done.

See also:

BioJava and Blast+

How to perform a Blast Search from a Java Application?

Answer 2

[Neilfws]

The simplest solution would be to run BLAST with options which return simplified, easy-to-parse output (such as tab-delimited).

Assuming that you are running the newer BLAST+ programs, the option is:

-outfmt <string>

where string can take the values 0 - 11. Values of 6, 7 and 10 are tab-delimited, tab-delimited + comment lines and CSV, respectively. You can further customize delimited output using specifiers, e.g. 'sseqid' for subject sequence ID.