Tophat 1.4.1 paired end alignment error -11 healp

[mediator]

Hi All,
I was running Tophat1.4.1 to do a pair-end alignment but got err=-11. Here is the command I used:

module load bowtie/0.12.7
module load tophat/1.4.1
tophat -m 2 -p 16 -r 210 -o <output dir> <hg19_index> /data/PE_read_1.fastq /data/PE_read_2.fastq


Output:

[Tue Feb 21 10:51:53 2012] Beginning TopHat run (v1.4.1)
-----------------------------------------------
[Tue Feb 21 10:51:53 2012] Preparing output location <output dir>
[Tue Feb 21 10:51:53 2012] Checking for Bowtie index files
[Tue Feb 21 10:51:53 2012] Checking for reference FASTA file
[Tue Feb 21 10:51:53 2012] Checking for Bowtie
Bowtie version: 0.12.7.0
[Tue Feb 21 10:51:53 2012] Checking for Samtools
Samtools Version: 0.1.16
[Tue Feb 21 10:51:53 2012] Generating SAM header for <hg19>
format: fastq
quality scale: phred33 (default)
[Tue Feb 21 10:52:19 2012] Preparing reads
left reads: min. length=50, count=121496435
right reads: min. length=50, count=121483750
[Tue Feb 21 12:08:01 2012] Mapping left_kept_reads against genome with Bowtie
[Tue Feb 21 14:28:21 2012] Processing bowtie hits
[Tue Feb 21 16:06:02 2012] Mapping left_kept_reads_seg1 against genome with Bowtie (1/2)
[Tue Feb 21 16:52:36 2012] Mapping left_kept_reads_seg2 against genome with Bowtie (2/2)
[Tue Feb 21 17:54:56 2012] Mapping right_kept_reads against genome with Bowtie
[Tue Feb 21 20:20:10 2012] Processing bowtie hits
[Tue Feb 21 22:00:24 2012] Mapping right_kept_reads_seg1 against genome with Bowtie (1/2)
[Tue Feb 21 22:58:31 2012] Mapping right_kept_reads_seg2 against genome with Bowtie (2/2)
[Tue Feb 21 23:58:39 2012] Searching for junctions via segment mapping
[FAILED]
Error: segment-based junction search failed with err =-11
Loading left segment hits...

Any suggestions? Thanks!

[pbluescript]

I would try dropping the -m 2 option. The issue might be that by allowing 2 mismatches in an anchor region only 8 bp long, you are getting a lot of potential matches that are making junction mapping difficult.

[mediator]

@pbluescript
Will try that, thanks!