GATK 2.7

[tanglingfung]

sorry, I cannot find specific code for supporting GATK2.7, can you tell me where it is? i want to add the support of --emitRefConfidence to HaplotypeCaller, please advice.

[chapmanb]

Paul; I try to avoid blocks of version specific code like this, so there isn't a specific place to look for 2.7 tweaks. The approach is to check based on version and apply parameters as needed. Here's an example in the filtering code:

https://github.com/chapmanb/bcbio-nextgen/blob/master/bcbio/variation/genotype.py#L269

For adding in the new reference confidence calls, you'd want to add a check here:

https://github.com/chapmanb/bcbio-nextgen/blob/master/bcbio/variation/genotype.py#L97

(version >= 2.7 and len(align_bams) == 1)

Thanks for looking at this.

[choishingwan]

I tried to link the gatk2.7 with bcbio-nextgen by changing the soft link under /share/java to point to the latest installation of gatk on my server (remove the original gatk link and create new one by ln -s -T gatk) However, I've encountered a large chunk of error where the last fragment is the following:

/storage/sam/anaconda/lib/python2.7/site-packages/bcbio/provenance/programs.pyc in get_version(name='picard', dirs=None, config={'algorithm': {'aligner': 'bwa', 'coverage_depth': 'high', 'coverage_interval': 'genome', 'mark_duplicates': 'picard', 'max_errors': 2, 'memory_adjust': {'direction': 'decrease', 'magnitude': 2}, 'num_cores': 1, 'platform': 'illumina', 'quality_format': 'Standard', 'realign': 'gatk', ...}, 'custom_algorithms': {'Minimal': {'aligner': ''}, 'RNA-seq': {'aligner': 'tophat', 'transcript_assemble': True}, 'variant2': {'aligner': 'bwa', 'coverage_depth': 'high', 'coverage_interval': 'exome', 'recalibrate': 'gatk', 'variantcaller': 'gatk'}}, 'log_dir': '/storage/sam/Data/log', 'resources': {'bcbio_variation': {'dir': '/storage/sam/share/java/bcbio_variation', 'jvm_opts': ['-Xms750m', '-Xmx2500m']}, 'bwa': {'cmd': 'bwa', 'cores': 16}, 'freebayes': {'memory': '2g'}, 'gatk': {'dir': '/storage/sam/share/java/gatk', 'jvm_opts': ['-Xms750m', '-Xmx2500m']}, 'gatk-haplotype': {'jvm_opts': ['-Xms2g', '-Xmx5500m']}, 'gatk-vqsr': {'jvm_opts': ['-Xms2g', '-Xmx4000m']}, 'gemini': {'cores': 16}, 'log': {'dir': 'log'}, 'novoalign': {'cores': 16, 'memory': '2G'}, 'picard': {'dir': '/storage/sam/share/java/picard'}, ...}}) 137 p = _get_program_file(dirs) 138 else: 139 p = config["resources"]["program_versions"] 140 with open(p) as in_handle: 141 for line in in_handle: --> 142 prog, version = line.rstrip().split(",") 143 if prog == name and version: 144 return version 145 raise KeyError("Version information not found for %s in %s" % (name, p)) 146

ValueError: need more than 1 value to unpack

---

Did I did something wrong with my gatk installation or was bcbio not supporting the latest gatk? (I have also checked java -version to be 1.7 as per requested by gatk...) Thank you for your helps

[chapmanb]

Thanks much for the report. The symlink approach you described should work, but it seems like something is wrong with your program version file. Could you post a Gist (https://gist.github.com/) of your provenance/programs.txt in the working directory?

Also running bcbio_nextgen.py with a single core (-n 1) during testing will produce less verbose error output and might make the issue easier to spot.

[choishingwan]

Hi, I have made the Gist:

https://gist.github.com/choishingwan/6886050

(Sorry about the format, I am new to Gist...)

When inspecting the content, I remotely remember seeing similar error when using the GATK Queue with the Could not find the main class: org.broadinstitute.sting.gatk.CommandLineGATK Error. However, when using GATK 2.6, I would observe such problem. I can even finish full Queue run based on the best practice from GATK

Thank you for your help

[chapmanb]

The error comes from using a pre-1.7 Java version running GATK 2.7 (or 2.6):

http://stackoverflow.com/questions/10382929/unsupported-major-minor-version-51-0

I know you mention installing 1.7 but it looks like the pipeline is picking up 1.6 instead which causes the issue. I added a check to bcbio-nextgen which should make the origin of the java used more clear in case it is a PATH issue. You can upgrade with bcbio_nextgen.py upgrade -u development or double check the current version on your PATH. Hope this helps.

[choishingwan]

I see I have now solved the problem with some googling.

So what happened with our server is that the original java (java 1.6) was installed at /usr/bin/java whereas our java 1.7 was installed at /software/java-7/jre1.7.0_25/bin/java. In order for us to use the bcbio_nextgen with the latest GATK, we will need to set the JAVA_HOME and also the PATH, pointing at the new java location and that will allow us to run the pipeline without the error message.

Thank you.