Pipeline template template » History » Version 4
Bryan Cosca, 05/21/2015 06:22 PM
| 1 | 1 | Bryan Cosca | h1. Pipeline template template |
|---|---|---|---|
| 2 | |||
| 3 | h2. Run-Command Template |
||
| 4 | |||
| 5 | 4 | Bryan Cosca | <pre> |
| 6 | "PrintReads":{ |
||
| 7 | "script_version":"29009a1c1f8a9653042c5853832881aca4141cf2", |
||
| 8 | "repository":"arvados", |
||
| 9 | "script":"run-command", |
||
| 10 | "script_parameters":{ |
||
| 11 | "input":{ |
||
| 12 | "output_of":"Indel-Realignment" |
||
| 13 | }, |
||
| 14 | "reference":{ |
||
| 15 | "dataclass":"Collection", |
||
| 16 | "default":"3514b8e5da0e8d109946bc809b20a78a+5698", |
||
| 17 | "link_name":"human_g1k_v37 reference data", |
||
| 18 | "title":"Base-Recalibration Input Reference genome (FASTA)" |
||
| 19 | }, |
||
| 20 | "command":[ |
||
| 21 | "java", |
||
| 22 | "-Xmx60g", |
||
| 23 | "-jar", |
||
| 24 | "$(dir $(gatk3))/GenomeAnalysisTK.jar", |
||
| 25 | "-T", |
||
| 26 | "PrintReads", |
||
| 27 | "-R", |
||
| 28 | "$(glob $(dir $(reference))/*.fasta)", |
||
| 29 | { |
||
| 30 | "foreach":"iterator", |
||
| 31 | "command":[ |
||
| 32 | "-I", |
||
| 33 | "$(iterator)" |
||
| 34 | ] |
||
| 35 | }, |
||
| 36 | "-BQSR", |
||
| 37 | "$(bqsr_table)", |
||
| 38 | "-nct", |
||
| 39 | "16", |
||
| 40 | "-o", |
||
| 41 | "$(outputname)" |
||
| 42 | ], |
||
| 43 | "outputname":{ |
||
| 44 | "value":{ |
||
| 45 | "list":"iterator", |
||
| 46 | "index":"0", |
||
| 47 | "command":"$(basename $(iterator)).bqsrCal.bam" |
||
| 48 | } |
||
| 49 | }, |
||
| 50 | "bqsr_table":{ |
||
| 51 | "value":{ |
||
| 52 | "list":"iterator", |
||
| 53 | "index":"0", |
||
| 54 | "command":"$(dir $(bqsr))/$(basename $(basename $(iterator))).recal_data.table" |
||
| 55 | } |
||
| 56 | }, |
||
| 57 | "input_dir":"$(dir $(input))", |
||
| 58 | "task.foreach":[ |
||
| 59 | "iterator" |
||
| 60 | ], |
||
| 61 | "iterator":{ |
||
| 62 | "value":{ |
||
| 63 | "group":"input_dir", |
||
| 64 | "regex":"(.*)\\.realigned.bam" |
||
| 65 | } |
||
| 66 | }, |
||
| 67 | "bqsr":{ |
||
| 68 | "output_of":"Base-Recalibration" |
||
| 69 | }, |
||
| 70 | "gatk3":{ |
||
| 71 | "dataclass":"Collection", |
||
| 72 | "default":"2e98fdc8e90f4c48a0714b711767c9ce+76", |
||
| 73 | "link_name":"Genome Analysis Toolkit 3.2-2", |
||
| 74 | "title":"PrintReads Input Version of GATK3 jar" |
||
| 75 | }, |
||
| 76 | "dbsnp":{ |
||
| 77 | "dataclass":"Collection", |
||
| 78 | "default":"8ac324bfa3dfff1ff81ed34b433869b1+6712", |
||
| 79 | "title":"Single Nucleotide Polymorphism database", |
||
| 80 | "description":"Base-Recalibration Input DBsnp" |
||
| 81 | } |
||
| 82 | }, |
||
| 83 | "runtime_constraints":{ |
||
| 84 | "max_tasks_per_node":1, |
||
| 85 | "min_nodes":1, |
||
| 86 | "docker_image":"bcosc/arv-base-java", |
||
| 87 | "arvados_sdk_version":"29009a1c1f8a9653042c5853832881aca4141cf2" |
||
| 88 | }, |
||
| 89 | "output_name":false |
||
| 90 | } |
||
| 91 | </pre> |
||
| 92 | 1 | Bryan Cosca | |
| 93 | h2. Crunch Script Template |
||
| 94 | <pre> |
||
| 95 | #!/usr/bin/env python |
||
| 96 | |||
| 97 | import arvados |
||
| 98 | import subprocess |
||
| 99 | import os |
||
| 100 | import sys |
||
| 101 | import re |
||
| 102 | from arvados.collection import Collection as coll |
||
| 103 | import arvados_tools |
||
| 104 | import shutil |
||
| 105 | |||
| 106 | arvados_tools.spawn_new_task_per_file('input','.*realigned.bqsrCal.bam$',if_sequence=0, and_end_task=True) |
||
| 107 | |||
| 108 | this_job = arvados.current_job() |
||
| 109 | this_task = arvados.current_task() |
||
| 110 | tmpdir = arvados.current_task().tmpdir |
||
| 111 | |||
| 112 | input_1 = this_task['parameters']['input_1'] |
||
| 113 | input_1_path = os.path.join(arvados.get_job_param_mount("input"),input_1) |
||
| 114 | tmp_input_1_path = os.path.join(tmpdir,input_1) # If we need to copy to tmpdir |
||
| 115 | shutil.copyfile(input_1_path,tmp_input_1_path) # If we need to copy to tmpdir |
||
| 116 | </pre> |
||
| 117 | |||
| 118 | |||
| 119 | <pre> |
||
| 120 | samtools_path = arvados_tools.get_file_path('samtools','^samtools$') |
||
| 121 | gatk_path = arvados_tools.get_file_path('gatk_jar','^GenomeAnalysisTK.jar$') |
||
| 122 | reference_path = arvados_tools.get_file_path('reference','.*f(ast)?a(.gz)?$') |
||
| 123 | dbsnp_path = arvados_tools.get_file_path('dbsnp','^dbsnp.*vcf$') |
||
| 124 | </pre> |
||
| 125 | |||
| 126 | <pre> |
||
| 127 | gatk_inserttool_args = [] |
||
| 128 | gatk_inserttool_output_name |
||
| 129 | gatk_inserttool_output_path |
||
| 130 | gatk_inserttool_pipe = subprocess.check_output() |
||
| 131 | </pre> |
||
| 132 | |||
| 133 | <pre> |
||
| 134 | samtools_inserttool_args = [] |
||
| 135 | </pre> |
||
| 136 | |||
| 137 | h2. Script Parameter Template |
||
| 138 | |||
| 139 | |||
| 140 | 3 | Bryan Cosca | h2. Latest arvados_sdk_version: |
| 141 | 2 | Bryan Cosca | |
| 142 | https://arvados.org/projects/arvados/repository/revisions/master/show/sdk/python |