#307 overlapper jobs failed

[Vishal Koparde]

I am getting the following error:

---

/gpfs_fs/atol/tol/Trypanosoma/Trypanosoma_cruzi_CL/gDNA/rawdata/sff_ngs_qc/MaSuRCA_assembly_v3/CA/1-overlapper/0000000002/h0046172081r0028857550 failed, job index 1.
.
.
.
.
/gpfs_fs/atol/tol/Trypanosoma/Trypanosoma_cruzi_CL/gDNA/rawdata/sff_ngs_qc/MaSuRCA_assembly_v3/CA/1-overlapper/0000000002/h0046172081r0046172080 failed, job index 369.
================================================================================

runCA failed.

---

Stack trace:

at /usr/global/blp/MaSuRCA-2.3.2/bin/../CA/Linux-amd64/bin/runCA line 1121.
main::caFailure('369 overlapper jobs failed', undef) called at /usr/global/blp/MaSuRCA-2.3.2/bin/../CA/Linu
x-amd64/bin/runCA line 1271
main::checkOverlapper('normal') called at /usr/global/blp/MaSuRCA-2.3.2/bin/../CA/Linux-amd64/bin/runCA lin
e 1320
main::checkOverlap('normal') called at /usr/global/blp/MaSuRCA-2.3.2/bin/../CA/Linux-amd64/bin/runCA line 5
341

---

Failure message:

369 overlapper jobs failed

---

Any idea how to solve it?

[Sergey Koren]

It looks like you're running MaSuRCA. MaSuRCA uses a fork of an older version of CA and may have bugs that have since been fixed in CA. I suggesting contacting the developers of MaSuRCA directly:
http://www.genome.umd.edu/masurca.html

That said, when overlapper fails, it is almost always a system resources issue and not a software issue. It looks like every one of your jobs failed, which makes me suspect some configuration or memory issue. you can check any of the 1-overlapper/*.out files for the exact error message that caused the failure.

[Vishal Koparde]

Could you please explain what you exactly mean by "system resources" and how I can request them differently?

[Sergey Koren]

The system resources for overlapping (memory and cores) are configured through runCA:
http://wgs-assembler.sourceforge.net/wiki/index.php/RunCA#OVL_Overlapper

However, MaSuRCA controls the spec file for assembly in your run and I don't know how to alter it's behavior. The output files I pointed you to have more information on what caused the overlap jobs to fail which would give more guidance on what needs to be adjusted for the MaSuRCA developers.

[Vishal Koparde]

The .out files you suggested contain the following error message.

/usr/global/blp/MaSuRCA-2.3.2/CA/Linux-amd64/bin/overlap: symbol lookup error: /usr/global/blp/MaSuRCA-2.3.2/CA/Lin
ux-amd64/bin/overlap: undefined symbol: _ZN9jellyfish10mer_dna_ns15mer_base_staticImLi0EE2k_E

Does this help in figuring out whats wrong?

[Vishal Koparde]

Further looking into jellyfish library revealed:

nm -D libjellyfish-2.0.so|grep ZN9jellyfish10mer_dna
000000000022a068 D _ZN9jellyfish10mer_dna_ns17error_different_kE
000000000022a060 D _ZN9jellyfish10mer_dna_ns18error_short_stringE

That is, it does not contain _ZN9jellyfish10mer_dna_ns15mer_base_staticImLi0EE2k_E! Is this a jellyfish version mismatch?

[Sergey Koren]

Yes, that error implies something is wrong in your compilation/installation of MaSuRCA (or rather the compilation of CA bundled with MaSuRCA). It looks like a function the overlap binary was linked to is missing. You can get more details running:
ldd /usr/global/blp/MaSuRCA-2.3.2/CA/Linux-amd64/bin/overlap

This bug is not in CA but in MaSuRCA (jellyfish is not part of the standard CA distribution). Thus, I do not know an appropriate fix. I'd recommend running test examples of MaSuRCA to make sure your installation is working and contacting the MaSuRCA developers for guidance in resolving the error.

[Sergey Koren]

Linking bug in MaSuRCA installation, not related to Celera Assembler.