Tophat 1.3.3 executing errors

[hajime]

Dear all:

My tophat was updated from 1.3.1 to 1.3.3 last week and
I've met a weird problem recently.

If I run the command like "tophat -p 8 --keep-tmp /my/bowtie_ref/path/refname /my/seq/path/sequence.fq", I'll get the following error msgs:

Error executing: /opt/ngs_tools/tophat-1.3.3/bin/bam_merge ./tophat_out/tmp/left_kept_reads.candidates_and_unspl.bam ./tophat_out/tmp/left_kept_reads.unspl.bam ./tophat_out/tmp/left_kept_reads.candidates.bam

Owing to the temp files were kept, I check wether the files, "left_kept_reads.unspl.bam", and "left_kept_reads.candidates.bam" is in ./tophat_out/tmp/ folder or not. These two files did exist.

However, I ran the command "/opt/ngs_tools/tophat-1.3.3/bin/bam_merge ./tophat_out/tmp/left_kept_reads.candidates_and_unspl.bam ./tophat_out/tmp/left_kept_reads.unspl.bam ./tophat_out/tmp/left_kept_reads.candidates.bam" again, I got the following error msgs:
Warning: no input BAM records found.
GList error (GList.hh:932):Invalid list index: 0

I don't know why there is no input BAM records. SO WIERED!!!

Are there anyone can help me? Thanks a lot

Best,
Yi

[DWW1505]

re: Tophat 1.3.3 executing errors

Long story short: .candidates.bam likely has a header but no data. The script python(.py) doesn't seem to have a
"no data/empty .bam" test before sending work to bam_merge.py.

In our case, candidates.bam had no data due to the reads being selected from a barren area of the
data set and excessive filtering of the reads.

As we are using a Windows7-64 fork of TopHat 1.3.3, the same patch would not
drop in to the Linux trunk. However, the basic code flow is the same and a test at that point should also
generate an alert to the fact that there is a file with no data.

Semper fi, Dave

[hajime]

Thanks for replying, Dave.

I downloaded the test data from Tophat website and used it for testing... it works...

The test data I used before... is only a part of sequence from a real RNA-seq experiment...
I found that the file size and content of xxx.candidiates.bam and xxx.unspl.bam are identical. Therefore, maybe the test data is the major problem

I will use the full sequence data for furthered test. Thanks again.