Bowtie1 error outputting data

Question

[Sbrillo]

I'm using bowtie1 to align reads from a big dataset (100gb Illumina reads vs 5gb Pacbio reads) and I have created a pipeline to split the alignment in smaller chunks and I'm running the command on an HPC.

I have the output files and it seems that the alignment worked for all 100 chunks.

However in a few cases when I checked the log.file I found this error:

Time loading forward index: 00:00:00 Error outputting data terminate called after throwing an instance of 'int'

The HPC error file is also reporting an error:

line 26: 30288 Aborted bowtie -a -t -5 0 -3 5 -p 32 -v 0 ........

Despite these errors, bowtie is generating the output files and they seem to be correct.

Any suggestions?

[GenoMax]

Is "ungapped" alignments the reason you are using bowtie v.1.x in this case? Have you checked the index (which I assume is made from PacBio reads) using bowtie-inspect?

[Sbrillo]

Thanks for the reply,

Yes my alignment must be ungapped. I checked the index and is fine also because the error is not related to any particular indexed chunks but happens randomly at every new run.

It might be that i have just to play around with memory usage with my HPC.

This is the command-line that i'm using:

bowtie -a -t -5 0 -3 5 -p 32 -v 0 -x $TMPDIR/${PBS_ARRAY_INDEX}_m_pac -q $TMPDIR/m.fastq --suppress 1,2,4,5,6,7,8,9 1> $TMPDIR/male.txt 2> /rds/general/user/mv419/ephemeral/redkmer-droso/data-Y/pacBio_illmapping/logs/${PBS_ARRAY_INDEX}_male_log.txt

During the indexing and the alignments, the files are moved one by one in a temporary directory and then copied in a final directory.

Let me know what do you think.

Best