Diagnosing samtools segfault

[Bukowski]

I'm trying to run samtools mpileup on BAM files, part of a tumour/normal pair so I can do some further analysis.

BAM's are aligned with BWA, post-processed with GATK and validated with ValidateSam from Picard. As far as I can tell there is no difference really between the BAM files, except that one causes mpileup to segfault every time.

I don't really know how to diagnose why this is happening - any hints? Errors occuring with Version: 0.1.18 (r982:295) and 0.1.14.

[nickloman]

Sounds like you might need the help of Heng Li (lh3 on here).

But in the meantime you could build samtools with debug flags (you will probably need to edit the Makefile to do this) and then run it through gdb, e.g.

gdb samtools
run "mpileup <other samtools arguments>"

When it segfaults you can then get a backtrace with 'bt'. That will at least show what part of the code is giving the problem.

If it's always happening on a particular alignment you could just filter that line out for now until the problem is fixed.

[Bernt.Popp]

I have the exact same problem with exome BAMs aligned with LifeScope, processed according to GATKs best practices v4 and validated with Picards ValidateSam.
UnifiedGenotyper walks true the files without error, when trying to call the exomes with samtools (Version: 0.1.18 (r982:295) i always get a "Segmentation fault (core dumped)" error.

Is there anything new about this error i.e. a bugfix/workaround?

[Bernt.Popp]

Apperently the Version in most linux repositorys is old altough it has the same version number. Samtools has moved its repository from Sourceforge to github, where you now can pul the latest Version with all Bugfuxes.

https://github.com/samtools/samtools

--> github clone https://github.com/samtools/samtools
--> make samtools
current Version: Version: 0.1.18-r580