Mercurial > repos > jjohnson > mmuff
view Galaxy-Workflow-MMuFF_Human_germline_v1.2.ga @ 2:04e170c09922
Add a workflow for noraml tissue samples.
author | Jim Johnson <jj@umn.edu> |
---|---|
date | Fri, 12 Apr 2013 07:51:45 -0500 |
parents | |
children |
line wrap: on
line source
{ "a_galaxy_workflow": "true", "annotation": "", "format-version": "0.1", "name": "MMuFF_Human_germline_v1.2", "steps": { "0": { "annotation": "Human RNA-seq paired reads left mates", "id": 0, "input_connections": {}, "inputs": [ { "description": "Human RNA-seq paired reads left mates", "name": "Forward reads fastq" } ], "name": "Input dataset", "outputs": [], "position": { "left": 240, "top": 287 }, "tool_errors": null, "tool_id": null, "tool_state": "{\"name\": \"Forward reads fastq\"}", "tool_version": null, "type": "data_input", "user_outputs": [] }, "1": { "annotation": "Human RNA-seq paired reads right mates", "id": 1, "input_connections": {}, "inputs": [ { "description": "Human RNA-seq paired reads right mates", "name": "Reverse reads fastq" } ], "name": "Input dataset", "outputs": [], "position": { "left": 249, "top": 383 }, "tool_errors": null, "tool_id": null, "tool_state": "{\"name\": \"Reverse reads fastq\"}", "tool_version": null, "type": "data_input", "user_outputs": [] }, "2": { "annotation": "Known SNPs - dbSNP.vcf", "id": 2, "input_connections": {}, "inputs": [ { "description": "Known SNPs - dbSNP.vcf", "name": "VCF file of known variants (dbSNP)" } ], "name": "Input dataset", "outputs": [], "position": { "left": 2327, "top": 758 }, "tool_errors": null, "tool_id": null, "tool_state": "{\"name\": \"VCF file of known variants (dbSNP)\"}", "tool_version": null, "type": "data_input", "user_outputs": [] }, "3": { "annotation": "Convert quality scores to sanger scale", "id": 3, "input_connections": { "input_file": { "id": 0, "output_name": "output" } }, "inputs": [ { "description": "runtime parameter for tool FASTQ Groomer", "name": "input_type" } ], "name": "FASTQ Groomer", "outputs": [ { "name": "output_file", "type": "fastqsanger" } ], "position": { "left": 510, "top": 264 }, "post_job_actions": {}, "tool_errors": null, "tool_id": "fastq_groomer", "tool_state": "{\"__page__\": 0, \"input_type\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/?.len\\\"\", \"options_type\": \"{\\\"options_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"input_file\": \"null\"}", "tool_version": "1.0.4", "type": "tool", "user_outputs": [] }, "4": { "annotation": "Convert quality scores to sanger scale", "id": 4, "input_connections": { "input_file": { "id": 1, "output_name": "output" } }, "inputs": [ { "description": "runtime parameter for tool FASTQ Groomer", "name": "input_type" } ], "name": "FASTQ Groomer", "outputs": [ { "name": "output_file", "type": "fastqsanger" } ], "position": { "left": 505, "top": 404 }, "post_job_actions": {}, "tool_errors": null, "tool_id": "fastq_groomer", "tool_state": "{\"__page__\": 0, \"input_type\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/?.len\\\"\", \"options_type\": \"{\\\"options_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"input_file\": \"null\"}", "tool_version": "1.0.4", "type": "tool", "user_outputs": [] }, "5": { "annotation": "Map paired reads to reference genome", "id": 5, "input_connections": { "input1": { "id": 3, "output_name": "output_file" }, "singlePaired|input2": { "id": 4, "output_name": "output_file" } }, "inputs": [], "name": "Tophat for Illumina", "outputs": [ { "name": "insertions", "type": "bed" }, { "name": "deletions", "type": "bed" }, { "name": "junctions", "type": "bed" }, { "name": "accepted_hits", "type": "bam" } ], "position": { "left": 884, "top": 305 }, "post_job_actions": { "HideDatasetActiondeletions": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "deletions" }, "HideDatasetActioninsertions": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "insertions" }, "HideDatasetActionjunctions": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "junctions" } }, "tool_errors": null, "tool_id": "tophat", "tool_state": "{\"__page__\": 0, \"refGenomeSource\": \"{\\\"genomeSource\\\": \\\"indexed\\\", \\\"index\\\": \\\"hg19_canonical\\\", \\\"__current_case__\\\": 0}\", \"input1\": \"null\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/?.len\\\"\", \"singlePaired\": \"{\\\"input2\\\": null, \\\"sPaired\\\": \\\"paired\\\", \\\"pParams\\\": {\\\"library_type\\\": \\\"fr-unstranded\\\", \\\"closure_search\\\": {\\\"use_search\\\": \\\"No\\\", \\\"__current_case__\\\": 1}, \\\"__current_case__\\\": 1, \\\"indel_search\\\": {\\\"max_insertion_length\\\": \\\"3\\\", \\\"max_deletion_length\\\": \\\"3\\\", \\\"__current_case__\\\": 1, \\\"allow_indel_search\\\": \\\"Yes\\\"}, \\\"seg_length\\\": \\\"25\\\", \\\"max_intron_length\\\": \\\"500000\\\", \\\"pSettingsType\\\": \\\"full\\\", \\\"min_intron_length\\\": \\\"70\\\", \\\"microexon_search\\\": \\\"No\\\", \\\"min_segment_intron\\\": \\\"50\\\", \\\"max_multihits\\\": \\\"1\\\", \\\"coverage_search\\\": {\\\"use_search\\\": \\\"Yes\\\", \\\"max_coverage_intron\\\": \\\"20000\\\", \\\"__current_case__\\\": 0, \\\"min_coverage_intron\\\": \\\"50\\\"}, \\\"splice_mismatches\\\": \\\"0\\\", \\\"anchor_length\\\": \\\"8\\\", \\\"mate_std_dev\\\": \\\"20\\\", \\\"own_junctions\\\": {\\\"use_junctions\\\": \\\"No\\\", \\\"__current_case__\\\": 1}, \\\"seg_mismatches\\\": \\\"2\\\", \\\"initial_read_mismatches\\\": \\\"2\\\", \\\"max_segment_intron\\\": \\\"500000\\\"}, \\\"__current_case__\\\": 1, \\\"mate_inner_distance\\\": \\\"80\\\"}\"}", "tool_version": "1.5.0", "type": "tool", "user_outputs": [] }, "6": { "annotation": "Keep only mappings with MAPQ of 255", "id": 6, "input_connections": { "input1": { "id": 5, "output_name": "accepted_hits" } }, "inputs": [], "name": "Filter SAM or BAM", "outputs": [ { "name": "output1", "type": "sam" } ], "position": { "left": 1154, "top": 339 }, "post_job_actions": { "HideDatasetActionoutput1": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "output1" } }, "tool_errors": null, "tool_id": "samtools_filter", "tool_state": "{\"__page__\": 0, \"bed_file\": \"null\", \"input1\": \"null\", \"read_group\": \"\\\"\\\"\", \"library\": \"\\\"\\\"\", \"regions\": \"\\\"\\\"\", \"header\": \"\\\"-h\\\"\", \"flag\": \"{\\\"filter\\\": \\\"no\\\", \\\"__current_case__\\\": 0}\", \"mapq\": \"\\\"255\\\"\"}", "tool_version": "1.1.0", "type": "tool", "user_outputs": [] }, "7": { "annotation": "", "id": 7, "input_connections": { "input1": { "id": 6, "output_name": "output1" } }, "inputs": [], "name": "flagstat", "outputs": [ { "name": "output1", "type": "txt" } ], "position": { "left": 1505, "top": 296 }, "post_job_actions": {}, "tool_errors": null, "tool_id": "samtools_flagstat", "tool_state": "{\"__page__\": 0, \"input1\": \"null\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/hg19_canonical.len\\\"\"}", "tool_version": "1.0.0", "type": "tool", "user_outputs": [] }, "8": { "annotation": "Add Read Group info required by GATK tools", "id": 8, "input_connections": { "inputFile": { "id": 6, "output_name": "output1" } }, "inputs": [], "name": "Add or Replace Groups", "outputs": [ { "name": "outFile", "type": "bam" } ], "position": { "left": 1506, "top": 476 }, "post_job_actions": {}, "tool_errors": null, "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/picard/picard_ARRG/1.56.0", "tool_state": "{\"rgpl\": \"\\\"ILLUMINA\\\"\", \"outputFormat\": \"\\\"True\\\"\", \"__page__\": 0, \"rgsm\": \"\\\"1\\\"\", \"readGroupOpts\": \"{\\\"rgOpts\\\": \\\"preSet\\\", \\\"__current_case__\\\": 0}\", \"rgid\": \"\\\"1\\\"\", \"rgpu\": \"\\\"1.5\\\"\", \"rglb\": \"\\\"1\\\"\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/hg19_canonical.len\\\"\", \"inputFile\": \"null\"}", "tool_version": "1.56.0", "type": "tool", "user_outputs": [] }, "9": { "annotation": "Synchronize paired reads", "id": 9, "input_connections": { "input_file": { "id": 8, "output_name": "outFile" } }, "inputs": [], "name": "Paired Read Mate Fixer", "outputs": [ { "name": "out_file", "type": "bam" } ], "position": { "left": 1841, "top": 348 }, "post_job_actions": { "HideDatasetActionout_file": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "out_file" } }, "tool_errors": null, "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/picard/rgPicFixMate/1.56.0", "tool_state": "{\"__page__\": 0, \"outputFormat\": \"\\\"True\\\"\", \"input_file\": \"null\", \"out_prefix\": \"\\\"Fix Mate\\\"\", \"sortOrder\": \"\\\"coordinate\\\"\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/hg19_canonical.len\\\"\"}", "tool_version": "1.56.0", "type": "tool", "user_outputs": [] }, "10": { "annotation": "Remove duplicate reads", "id": 10, "input_connections": { "input_file": { "id": 9, "output_name": "out_file" } }, "inputs": [], "name": "Mark Duplicate reads", "outputs": [ { "name": "out_file", "type": "bam" }, { "name": "html_file", "type": "html" } ], "position": { "left": 2086, "top": 333 }, "post_job_actions": {}, "tool_errors": null, "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/picard/rgPicardMarkDups/1.56.0", "tool_state": "{\"remDups\": \"\\\"True\\\"\", \"assumeSorted\": \"\\\"True\\\"\", \"input_file\": \"null\", \"readRegex\": \"\\\"[a-zA-Z0-9]+:[0-9]:([0-9]+):([0-9]+):([0-9]+).*\\\"\", \"__page__\": 0, \"out_prefix\": \"\\\"Dupes Marked\\\"\", \"optDupeDist\": \"\\\"100\\\"\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/hg19_canonical.len\\\"\"}", "tool_version": "1.56.0", "type": "tool", "user_outputs": [] }, "11": { "annotation": "Reorder BAM for GATK tools", "id": 11, "input_connections": { "inputFile": { "id": 10, "output_name": "out_file" } }, "inputs": [], "name": "Reorder SAM/BAM", "outputs": [ { "name": "outFile", "type": "bam" } ], "position": { "left": 2401, "top": 305 }, "post_job_actions": {}, "tool_errors": null, "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/picard/picard_ReorderSam/1.56.0", "tool_state": "{\"__page__\": 0, \"outputFormat\": \"\\\"True\\\"\", \"allowContigLenDiscord\": \"\\\"False\\\"\", \"allowIncDictConcord\": \"\\\"False\\\"\", \"source\": \"{\\\"indexSource\\\": \\\"built-in\\\", \\\"ref\\\": \\\"hg19_canonical\\\", \\\"__current_case__\\\": 0}\", \"inputFile\": \"null\"}", "tool_version": "1.56.0", "type": "tool", "user_outputs": [] }, "12": { "annotation": "", "id": 12, "input_connections": { "reference_source|input_bam": { "id": 11, "output_name": "outFile" }, "rod_bind_0|rod_bind_type|input_rod": { "id": 2, "output_name": "output" } }, "inputs": [], "name": "Realigner Target Creator", "outputs": [ { "name": "output_interval", "type": "gatk_interval" }, { "name": "output_log", "type": "txt" } ], "position": { "left": 2565, "top": 491 }, "post_job_actions": { "HideDatasetActionoutput_log": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "output_log" } }, "tool_errors": null, "tool_id": "gatk_realigner_target_creator", "tool_state": "{\"__page__\": 0, \"gatk_param_type\": \"{\\\"gatk_param_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"analysis_param_type\": \"{\\\"analysis_param_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"reference_source\": \"{\\\"ref_file\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"hg19_canonical\\\"}, \\\"reference_source_selector\\\": \\\"cached\\\", \\\"input_bam\\\": null, \\\"__current_case__\\\": 0}\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/hg19_canonical.len\\\"\", \"rod_bind\": \"[{\\\"__index__\\\": 0, \\\"rod_bind_type\\\": {\\\"rod_bind_type_selector\\\": \\\"dbsnp\\\", \\\"__current_case__\\\": 0, \\\"input_rod\\\": null}}]\"}", "tool_version": "0.0.4", "type": "tool", "user_outputs": [] }, "13": { "annotation": "Perform local realignments for indels", "id": 13, "input_connections": { "reference_source|input_bam": { "id": 11, "output_name": "outFile" }, "target_intervals": { "id": 12, "output_name": "output_interval" } }, "inputs": [], "name": "Indel Realigner", "outputs": [ { "name": "output_bam", "type": "bam" }, { "name": "output_log", "type": "txt" } ], "position": { "left": 2811, "top": 324 }, "post_job_actions": { "HideDatasetActionoutput_log": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "output_log" } }, "tool_errors": null, "tool_id": "gatk_indel_realigner", "tool_state": "{\"__page__\": 0, \"knowns_only\": \"\\\"False\\\"\", \"gatk_param_type\": \"{\\\"gatk_param_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"lod_threshold\": \"\\\"5.0\\\"\", \"analysis_param_type\": \"{\\\"analysis_param_type_selector\\\": \\\"basic\\\", \\\"__current_case__\\\": 0}\", \"target_intervals\": \"null\", \"reference_source\": \"{\\\"ref_file\\\": {\\\"__class__\\\": \\\"UnvalidatedValue\\\", \\\"value\\\": \\\"hg19_canonical\\\"}, \\\"reference_source_selector\\\": \\\"cached\\\", \\\"input_bam\\\": null, \\\"__current_case__\\\": 0}\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/hg19_canonical.len\\\"\", \"rod_bind\": \"[]\"}", "tool_version": "0.0.6", "type": "tool", "user_outputs": [] }, "14": { "annotation": "Check again for duplicate reads and eliminate", "id": 14, "input_connections": { "input_file": { "id": 13, "output_name": "output_bam" } }, "inputs": [], "name": "Mark Duplicate reads", "outputs": [ { "name": "out_file", "type": "bam" }, { "name": "html_file", "type": "html" } ], "position": { "left": 3174, "top": 291 }, "post_job_actions": { "HideDatasetActionhtml_file": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "html_file" }, "HideDatasetActionout_file": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "out_file" } }, "tool_errors": null, "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/picard/rgPicardMarkDups/1.56.0", "tool_state": "{\"remDups\": \"\\\"True\\\"\", \"assumeSorted\": \"\\\"True\\\"\", \"input_file\": \"null\", \"readRegex\": \"\\\"[a-zA-Z0-9]+:[0-9]:([0-9]+):([0-9]+):([0-9]+).*\\\"\", \"__page__\": 0, \"out_prefix\": \"\\\"Dupes Marked\\\"\", \"optDupeDist\": \"\\\"100\\\"\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/hg19_canonical.len\\\"\"}", "tool_version": "1.56.0", "type": "tool", "user_outputs": [] }, "15": { "annotation": "Determine base calls at each position", "id": 15, "input_connections": { "reference_source|input_bams_0|input_bam": { "id": 14, "output_name": "out_file" } }, "inputs": [], "name": "MPileup", "outputs": [ { "name": "output_mpileup", "type": "pileup" }, { "name": "output_log", "type": "txt" } ], "position": { "left": 3554, "top": 265 }, "post_job_actions": { "HideDatasetActionoutput_log": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "output_log" }, "HideDatasetActionoutput_mpileup": { "action_arguments": {}, "action_type": "HideDatasetAction", "output_name": "output_mpileup" } }, "tool_errors": null, "tool_id": "samtools_mpileup", "tool_state": "{\"__page__\": 0, \"advanced_options\": \"{\\\"max_reads_per_bam\\\": \\\"2500\\\", \\\"advanced_options_selector\\\": \\\"advanced\\\", \\\"extended_BAQ_computation\\\": \\\"False\\\", \\\"region_string\\\": \\\"\\\", \\\"output_per_sample_strand_bias_p_value\\\": \\\"False\\\", \\\"minimum_base_quality\\\": \\\"20\\\", \\\"disable_probabilistic_realignment\\\": \\\"True\\\", \\\"skip_anomalous_read_pairs\\\": \\\"True\\\", \\\"minimum_mapping_quality\\\": \\\"0\\\", \\\"output_per_sample_read_depth\\\": \\\"False\\\", \\\"__current_case__\\\": 0, \\\"position_list\\\": null, \\\"coefficient_for_downgrading\\\": \\\"0\\\"}\", \"chromInfo\": \"\\\"/website/galaxy.msi.umn.edu/PRODUCTION/tool-data/shared/ucsc/chrom/hg19_canonical.len\\\"\", \"genotype_likelihood_computation_type\": \"{\\\"genotype_likelihood_computation_type_selector\\\": \\\"do_not_perform_genotype_likelihood_computation\\\", \\\"__current_case__\\\": 1}\", \"reference_source\": \"{\\\"ref_file\\\": \\\"hg19_canonical\\\", \\\"reference_source_selector\\\": \\\"cached\\\", \\\"input_bams\\\": [{\\\"__index__\\\": 0, \\\"input_bam\\\": null}], \\\"__current_case__\\\": 0}\"}", "tool_version": "0.0.1", "type": "tool", "user_outputs": [] }, "16": { "annotation": "Minimum Base Quality: 30\nMinimum Coverage Depth: 5\nMinimum Frequency of a Specific Allele: .33\nVCF_ID = germline", "id": 16, "input_connections": { "input_file": { "id": 15, "output_name": "output_mpileup" } }, "inputs": [ { "description": "runtime parameter for tool Pileup to VCF", "name": "min_base_qual" } ], "name": "Pileup to VCF", "outputs": [ { "name": "output_file", "type": "vcf" } ], "position": { "left": 3937, "top": 258 }, "post_job_actions": {}, "tool_errors": null, "tool_id": "toolshed.g2.bx.psu.edu/repos/jjohnson/pileup_to_vcf/pileup_to_vcf/2.0", "tool_state": "{\"snps_only\": \"\\\"False\\\"\", \"min_cvrg\": \"\\\"5\\\"\", \"allow_multiples\": \"\\\"True\\\"\", \"input_file\": \"null\", \"__page__\": 0, \"vcf_id\": \"\\\"germline\\\"\", \"cols\": \"{\\\"select_order\\\": \\\"no\\\", \\\"__current_case__\\\": 0}\", \"depth_as\": \"\\\"ref\\\"\", \"min_base_qual\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\", \"min_var_pct\": \"\\\"0.33\\\"\"}", "tool_version": "2.2", "type": "tool", "user_outputs": [] } } }