search

genome / gms

The Genome Modeling System installer
https://github.com/genome/gms/wiki
GNU Lesser General Public License v3.0
78 stars 22 forks source link

somatic-variation fails during create-report step (related to supporting callers?) #125

Closed malachig closed 10 years ago

malachig commented 10 years ago

Since the merge we are getting warnings and errors in somatic-variation create-report step summarized below:

First,

2014-03-12 00:59:59-0500 clia1: Use of uninitialized value in numeric eq (==) at /opt/gms/4DSHC42/sw/genome/lib/perl/Genome/Model/SomaticVariation/Command/CreateReport.pm line 393.
2014-03-12 00:59:59-0500 clia1: Use of uninitialized value $list[0] in sprintf at /opt/gms/4DSHC42/sw/workflow/lib/Workflow/OperationType/Command.pm line 371
2014-03-12 00:59:59-0500 clia1: Removing sites supported by less than  callers
2014-03-12 01:00:01-0500 clia1: Use of uninitialized value in concatenation (.) or string at /opt/gms/4DSHC42/sw/genome/lib/perl/Genome/Model/SomaticVariation/Command/CreateReport.pm line 401.
2014-03-12 01:00:01-0500 clia1: Use of uninitialized value in numeric ge (>=) at /opt/gms/4DSHC42/sw/genome/lib/perl/Genome/Model/SomaticVariation/Command/CreateReport.pm line 418, <GEN17> line 1.

And then later,

2014-03-12 01:10:17-0500 clia1: Use of uninitialized value in split at /opt/gms/4DSHC42/sw/genome/lib/perl/Genome/Model/SomaticVariation/Command/CreateReport.pm line 847.
2014-03-12 01:10:17-0500 clia1: RUN: joinx sort -i /tmp/526.tmpdir/gm-genome_sys-2014-03-12_00_59_59--VqbD/anonymous14 -o /opt/gms/4DSHC42/fs/4DSHC42/info/model_data/CreateReport/b0a07bc193a642409c8e068422f56695/snvs.indels.annotated.tier
2014-03-12 01:10:17-0500 clia1: Failed to open file /tmp/526.tmpdir/gm-genome_sys-2014-03-12_00_59_59--VqbD/anonymous14
2014-03-12 01:10:17-0500 clia1: Command module died or returned undef.
2014-03-12 01:10:17-0500 clia1: ERROR RUNNING COMMAND.  Exit code 2 from: joinx sort -i /tmp/526.tmpdir/gm-genome_sys-2014-03-12_00_59_59--VqbD/anonymous14 -o /opt/gms/4DSHC42/fs/4DSHC42/info/model_data/CreateReport/b0a07bc193a642409c8e068422f56695/snvs.indels.annotated.tier
2014-03-12 01:10:17-0500 clia1: See the command's captured STDERR (if it exists) for more information at /opt/gms/4DSHC42/sw/genome/lib/perl/Genome/Model/SomaticVariation/Command/CreateReport.pm line 860
2014-03-12 01:10:17-0500 clia1:         Genome::Sys::shellcmd('Genome::Sys', 'cmd', 'joinx sort -i /tmp/526.tmpdir/gm-genome_sys-2014-03-12_00_59_...') called at /opt/gms/4DSHC42/sw/genome/lib/perl/Genome/Model/SomaticVariation/Command/CreateReport.pm line 860
2014-03-12 01:10:17-0500 clia1:         Genome::Model::SomaticVariation::Command::CreateReport::_create_review_bed('Genome::Model::SomaticVariation::Command::CreateReport=HASH(0...') called at /opt/gms/4DSHC42/sw/genome/lib/perl/Genome/Model/SomaticVariation/Command/CreateReport.pm line 832
2014-03-12 01:10:17-0500 clia1:         Genome::Model::SomaticVariation::Command::CreateReport::_create_review_files('Genome::Model::SomaticVariation::Command::CreateReport=HASH(0...') called at /opt/gms/4DSHC42/sw/genome/lib/perl/Genome/Model/SomaticVariation/Command/CreateReport.pm line 153

It looks like all of this was added recently by @susannasiebert. We should confer with her before proceeding further.

malachig commented 10 years ago

One thing that is strange about the output above is the warning:

2014-03-12 00:59:59-0500 clia1: Removing sites supported by less than  callers

The number of callers required should be defined there via $self->required_snv_callers. This has a default defined in both Genome/Model/SomaticVariation.pm and Genome/Model/SomaticVariation/Command/CreateReport.pm.

One thing I noticed that may be a red herring is that parameters in the class defs for these modules are using a mix of default => x and default_value => x

apregier commented 10 years ago

required_snv_callers is a new processing profile parameter. The reason that the default in the tool doesn't kick in is because the workflow passes that parameter in from the processing profile. In the in-house Genome instance, we backfilled all of the processing profiles to have the default value. To fix the immediate error, you could backfill the processing profiles in sGMS as well.
Handling these types of release migrations given multiple instances of Genome is a much larger topic.

susannasiebert commented 10 years ago

there are two more parameters beside required_snv_callers that need to be backfilled: tiers_to_review and restrict_to_target_regions. All of them can be backfilled with the default value of 1.

gatoravi commented 10 years ago

I think a new metadata dump for the sGMS will help backfill these processing profiles on the sGMS.

malachig commented 10 years ago

Yes, testing that now.

malachig commented 10 years ago

The new metadata dump is here: https://xfer.genome.wustl.edu/gxfer1/project/gms/testdata/GMS1/export/18177dd5eca44514a47f367d9804e17a-2014.3.12.dat

No problems importing it and running new ref-align builds. Testing somatic-variation now to see if the issue is resolved

malachig commented 10 years ago

Somatic-variation processing profiles have now been updated and somatic-variation builds are once again succeeding. Closing this issue now.