[Biojava-dev] a bug in WU-BLAST parser

[mark.schreiber at novartis.com]

Hi -

I will take a look at this, however, if you know of a solution could you 
post it to me and I will commit it to CVS.

Thanks.

- Mark





"Hongyu Zhang" <hzhang at ceres-inc.com>
Sent by: biojava-dev-bounces at portal.open-bio.org
10/08/2005 06:02 AM

 
        To:     <biojava-dev at biojava.org>
        cc:     Raj Thavamani <rthavamani at ceres-inc.com>, (bcc: Mark 
Schreiber/GP/Novartis)
        Subject:        [Biojava-dev] a bug in WU-BLAST parser


I am reporting an error in the WU-BLAST parser to the Biojava
developer's list. The Biojava version is the latest 1.4, and the bug is
in file src/org/biojava/bio/program/sax/WuBlastSummaryLineHelper.java.
The error happened when the description field of the BLAST hit summary
lines is empty. For example:

 

ADL26502                                                      1430
8.5e-145  1

 

Cause of the error:

 

A parameter, iGrab, in the code, affects the behavior of the parser
dependent on the WU-BLAST version. When the version is BLASTX, TBLASTX
or TBLASTN, igrab is set to 4, and the code at the line 120 will try to
read the non-existed "Frame" field from the WU-BLAST summary lines.
Usually, the description field in the summary line is not empty, so this
line of code will grab the last word from the description field as the
"Frame" field mistakenly. This mistake usually won't matter to the
following codes and therefore is hidden in most situations. In the
example above, however, since the description field of the hit is empty,
the code will mistakenly shift and read the next "High Score" field
(1430 in this case) as the "Frame" field and cause the StringTokenizer
to throw an error.

 

 

Hongyu Zhang, Ph.D.

Computational Biologist

Ceres Inc.

1535 Rancho Conejo Blvd

Thousand oaks, CA 91320

Phone: (805)376-6504 ext. 1204

Fax: (805)376-6537

 


**********************************************************************

This email message is for the sole use of the intended recipient(s) and 
may contain confidential and privileged information.  Any unauthorized 
review, use, disclosure or distribution is prohibited.  If you are not the 
intended recipient, please contact the sender by reply email and destroy 
all copies of the original message.  Ceres, Inc. declines any liability 
for any viruses or other potentially harmful code which may be transmitted 
by or accompanying this email or any attachment.

**********************************************************************


_______________________________________________
biojava-dev mailing list
biojava-dev at biojava.org
http://biojava.org/mailman/listinfo/biojava-dev