Pindel Segmentation Fault

Question

[lucas.a.swanson]

I am trying to run Pindel on a RNA-seq reads-to-genome BAM file, created with the STAR aligner, and I am getting a segfault.

I checked the Pindel code out with svn:

svn checkout https://trac.nbic.nl/svn/pindel/trunk pindel-0.2.4w

A pindel-0.2.4w/test

A pindel-0.2.4w/test/SmallTest

...

A pindel-0.2.4w/Makefile

A pindel-0.2.4w/README

U pindel-0.2.4w

Checked out revision 508.

Apparently this is supposed to be version 0.2.4w? (According to this old answer: https://trac.nbic.nl/pipermail/pindel-users/2013-May/000271.html)

And I used this command to run Pindel:

pindel -i AMLTEST8FI_bam_config.txt -f ${ref_dir}/GRCh37_UCSC.fa -c ALL -o ${output_dir}/AMLTEST8FI --report_close_mapped_reads

The contents of the config file are:

${input_dir}/AMLTEST8FI.bam 130 AMLTEST8FI

The last few lines of stdout before the error are:

...

Reads already used: 16

Far ends already mapped 92

Checksum of far ends: 9870976

Searching some type of variant, replace this with the correct name in child class ...

Total: 9 +6 -3

Sorting and outputing short insertions ...

Short insertions: 1

Total: 148; Close_end_found 148; Far_end_found 92; Used 25.

For LI and BP: 56

LI: maxInsertSize: 130

Breakpoints for large insertions (LI): 0

Mining, Sorting and output LI results: 0 seconds.

Other unassigned breakpoints (BP): 0

Mining, Sorting and output BP results: 0 seconds.

There are 0 split-reads saved for the next cycle.

Processing chromosome: chr1_gl000191_random

Chromosome Size: 106433

currentChromosome->getCompSize()306433

NumBoxes: 61287 BoxSize: 10

NumBoxes: 61287 BoxSize: 10

/opt/sge/spool/node0046/job_scripts/2500151: line 16: 3856 Segmentation fault pindel -i AMLTEST8FI_bam_config.txt -f /genesis/scratch/clingenetics/reference_genomes/UCSC_GRCh37/GRCh37_UCSC.fa -c ALL -o /genesis/scratch/clingenetics/lswanson_test/CLINGEN-747_implement_pindel/step_10/AMLTEST8FI/AMLTEST8FIFLOWCELL_7/AML_Clinical_Pipeline_1.0.0_2013_06_13/star/pindel/AMLTEST8FI --report_close_mapped_reads

Any idea what the problem might be?

~Thanks, Lucas Swanson

Answer 1

[liangkaiye]

please update pindel source to 0.2.5. https://github.com/genome/pindel

[lucas.a.swanson]

Thanks.

I cloned the source code:

$ git clone https://github.com/genome/pindel.git

Cloning into 'pindel'...

remote: Counting objects: 3410, done.

remote: Compressing objects: 100% (885/885), done.

remote: Total 3410 (delta 2591), reused 3328 (delta 2509)

Receiving objects: 100% (3410/3410), 91.60 MiB | 8.99 MiB/s, done.

Resolving deltas: 100% (2591/2591), done.

Checking out files: 100% (160/160), done.

But the 'RELEASE' file suggests that it is still 0.2.4w:

$ cat RELEASE

Pindel 0.2.4w April 25, 2013

-first version after moving to WashU TGI

-changed test runtime max

Pindel 0.2.4t August 13, 2012

• resolved bug in which using pindel with a single incorrect option (like ./pindel -g) caused a segmentation fault

• updated pindel2vcf to version 0.5.0; -G (GATK-compatibility option) now works properly even for unusual SVs; END positions updated according to http://www.1000genomes.org/wiki/Analysis/Variant%20Call%20Format/vcf-variant-call-format-version-41, resolved mysterious ".:10" genotypes, and added -P option to allow fusing all output files into one.

• if an event is near multiple breakdancer events, all of these events will be investigated now

Is the RELEASE file simply out-of-date?

Answer 2

[Allpowerde]

The reason for the segmentation fault seems to have something to do with the bam format produced by your aligner. My alignments are also created by an aligner other than BWA (in my case casava) and I do get the same segmentation fault even with the new version 0.2.5a3, Oct 24 2013 (Note, it is a new version despite the RELEASE still saying "Pindel 0.2.4w April 25, 2013").

However, I followed "Option 3: SAM/BAM-file not produced by BWA" (http://gmt.genome.wustl.edu/pindel/0.2.4/user-manual.html) and created Output4Pindel.txt from the input.bam. The subsequent pindel call ran through successfully (Note, sam2pindel actually needs 6 variables -- Illumina-PairEnd/Illumina-MatePair -- even though the docu says only 5).