Pipeline template template » History » Revision 2
Revision 1 (Bryan Cosca, 05/21/2015 05:24 PM) → Revision 2/25 (Bryan Cosca, 05/21/2015 05:24 PM)
h1. Pipeline template template
h2. Run-Command Template
h2. Crunch Script Template
<pre>
#!/usr/bin/env python
import arvados
import subprocess
import os
import sys
import re
from arvados.collection import Collection as coll
import arvados_tools
import shutil
arvados_tools.spawn_new_task_per_file('input','.*realigned.bqsrCal.bam$',if_sequence=0, and_end_task=True)
this_job = arvados.current_job()
this_task = arvados.current_task()
tmpdir = arvados.current_task().tmpdir
input_1 = this_task['parameters']['input_1']
input_1_path = os.path.join(arvados.get_job_param_mount("input"),input_1)
tmp_input_1_path = os.path.join(tmpdir,input_1) # If we need to copy to tmpdir
shutil.copyfile(input_1_path,tmp_input_1_path) # If we need to copy to tmpdir
</pre>
<pre>
samtools_path = arvados_tools.get_file_path('samtools','^samtools$')
gatk_path = arvados_tools.get_file_path('gatk_jar','^GenomeAnalysisTK.jar$')
reference_path = arvados_tools.get_file_path('reference','.*f(ast)?a(.gz)?$')
dbsnp_path = arvados_tools.get_file_path('dbsnp','^dbsnp.*vcf$')
</pre>
<pre>
gatk_inserttool_args = []
gatk_inserttool_output_name
gatk_inserttool_output_path
gatk_inserttool_pipe = subprocess.check_output()
</pre>
<pre>
samtools_inserttool_args = []
</pre>
h2. Script Parameter Template
h3. Latest arvados_sdk_version:
https://arvados.org/projects/arvados/repository/revisions/master/show/sdk/python