diff --git a/workflows/VGP-assembly-v2/Fetch-Related-Genomes/.dockstore.yml b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/.dockstore.yml new file mode 100644 index 0000000000..9b8cccab1d --- /dev/null +++ b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/.dockstore.yml @@ -0,0 +1,11 @@ +version: 1.2 +workflows: +- name: main + subclass: Galaxy + publish: true + primaryDescriptorPath: /Fetch-Related-Genomes.ga + testParameterFiles: + - /Fetch-Related-Genomes-tests.yml + authors: + - name: Delphine Lariviere + orcid: 0000-0001-6421-3484 diff --git a/workflows/VGP-assembly-v2/Fetch-Related-Genomes/CHANGELOG.md b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/CHANGELOG.md new file mode 100644 index 0000000000..8d9be5a6b3 --- /dev/null +++ b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/CHANGELOG.md @@ -0,0 +1,5 @@ +# Changelog + +## [0.1] - 2026-05-12 + +- Initial release: given a reference assembly, the workflow uses RefSeq Masher to find related-species genomes, applies user-defined taxonomic and assembly-quality filters (assembly level, type, BioSample sex, sequencing technology, taxonomic class, chromosome count), and downloads the genome fastas of the n closest matching related species. diff --git a/workflows/VGP-assembly-v2/Fetch-Related-Genomes/Fetch-Related-Genomes-tests.yml b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/Fetch-Related-Genomes-tests.yml new file mode 100644 index 0000000000..1b8dbad849 --- /dev/null +++ b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/Fetch-Related-Genomes-tests.yml @@ -0,0 +1,93 @@ +- doc: Test 1 - Baseline, no filters applied, minimum assembly level set to Contig, sequences not downloaded + job: + Assembly: + class: File + location: https://zenodo.org/records/20074725/files/Haplotype_1.fasta + filetype: fasta + hashes: + - hash_function: SHA-1 + hash_value: a0ee25fd9f7cf223ca40ff530e1201589fade212 + Search assemblies among the n closest related species *: 30 + Download n related species: 1 + "Download sequences? ": false + Minimum Level Assembly: Contig + Assembly Type: All + Assembly BioSample Sex: Any + Assembly Sequencing Technology: "" + Assembly Taxonomic Class: "" + Minimum Number Of Chromosomes: 0 + Maximum Number Of Chromosomes: 0 + Keep empty values when filtering on number of chromosomes: true + outputs: + All Refseq Matches: + asserts: + has_n_lines: + n: 1 + delta: 1000 + Result Species: + asserts: + has_n_lines: + n: 1 + delta: 100 + +- doc: Test 2 - Taxonomic class and sequencing technology filters + job: + Assembly: + class: File + location: https://zenodo.org/records/20074725/files/Haplotype_1.fasta + filetype: fasta + hashes: + - hash_function: SHA-1 + hash_value: a0ee25fd9f7cf223ca40ff530e1201589fade212 + Search assemblies among the n closest related species *: 30 + Download n related species: 1 + "Download sequences? ": true + Minimum Level Assembly: Contig + Assembly Type: All + Assembly BioSample Sex: Any + Assembly Sequencing Technology: "Illumina" + Assembly Taxonomic Class: "Bacilli,Bacteroidia,Gammaproteobacteria" + Minimum Number Of Chromosomes: 0 + Maximum Number Of Chromosomes: 0 + Keep empty values when filtering on number of chromosomes: true + outputs: + All Refseq Matches: + asserts: + has_n_lines: + n: 1 + delta: 1000 + Filtered Related Species Genome Report: + asserts: + has_text: + text: "Illumina" + Related Species Genomes: + element_tests: {} + +- doc: Test 3 - Chromosome count filter (strict, drops rows with empty chromosome count) + job: + Assembly: + class: File + location: https://zenodo.org/records/20074725/files/Haplotype_1.fasta + filetype: fasta + hashes: + - hash_function: SHA-1 + hash_value: a0ee25fd9f7cf223ca40ff530e1201589fade212 + Search assemblies among the n closest related species *: 30 + Download n related species: 1 + "Download sequences? ": true + Minimum Level Assembly: Contig + Assembly Type: All + Assembly BioSample Sex: Any + Assembly Sequencing Technology: "" + Assembly Taxonomic Class: "" + Minimum Number Of Chromosomes: 1 + Maximum Number Of Chromosomes: 5 + Keep empty values when filtering on number of chromosomes: false + outputs: + All Refseq Matches: + asserts: + has_n_lines: + n: 1 + delta: 1000 + Related Species Genomes: + element_tests: {} diff --git a/workflows/VGP-assembly-v2/Fetch-Related-Genomes/Fetch-Related-Genomes.ga b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/Fetch-Related-Genomes.ga new file mode 100644 index 0000000000..0d514dc367 --- /dev/null +++ b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/Fetch-Related-Genomes.ga @@ -0,0 +1,3292 @@ +{ + "a_galaxy_workflow": "true", + "annotation": "Given a reference assembly, search NCBI for related-species genomes using RefSeq Masher, apply user-defined taxonomic and assembly-quality filters (assembly level, type, BioSample sex, sequencing technology, taxonomic class, number of chromosomes), and download the genome fastas of the n closest matching related species. Useful for assembling a synteny reference set ahead of manual curation workflows such as JBrowse2 precuration tracks.", + "comments": [], + "creator": [ + { + "class": "Person", + "identifier": "https://orcid.org/0000-0001-6421-3484", + "name": "Delphine Lariviere" + }, + { + "class": "Organization", + "name": "IWC" + } + ], + "format-version": "0.1", + "license": "MIT", + "name": "Fetch Related Genomes", + "report": { + "markdown": "\n# Workflow Execution Report\n\n## Workflow Inputs\n```galaxy\ninvocation_inputs()\n```\n\n## Workflow Outputs\n```galaxy\ninvocation_outputs()\n```\n\n## Workflow\n```galaxy\nworkflow_display()\n```\n" + }, + "steps": { + "0": { + "annotation": "Reference used to search related genomes.", + "content_id": null, + "errors": null, + "id": 0, + "input_connections": {}, + "inputs": [ + { + "description": "Reference used to search related genomes.", + "name": "Assembly" + } + ], + "label": "Assembly", + "name": "Input dataset", + "outputs": [], + "position": { + "left": 0, + "top": 0 + }, + "tool_id": null, + "tool_state": "{\"optional\": false, \"format\": [\"fasta\"], \"tag\": null}", + "tool_version": null, + "type": "data_input", + "uuid": "f104df80-258d-42f5-911b-ebd55e3a7af5", + "when": null, + "workflow_outputs": [] + }, + "1": { + "annotation": "Download genomes fasta. Otherwise just return the potential candidates.", + "content_id": null, + "errors": null, + "id": 1, + "input_connections": {}, + "inputs": [ + { + "description": "Download genomes fasta. Otherwise just return the potential candidates.", + "name": "Download sequences? " + } + ], + "label": "Download sequences? ", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 61.74049540666487, + "top": 105.64114243923694 + }, + "tool_id": null, + "tool_state": "{\"default\": false, \"validators\": [], \"parameter_type\": \"boolean\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "5589479f-a94b-4cab-af30-a8935d4fe583", + "when": null, + "workflow_outputs": [] + }, + "2": { + "annotation": "The filters will be applied to the number of species specified in the parameter \"Search assemblies among the n closest related species\". Then the n(this parameter) most related species will be downloaded. Note: A large number of species will cause mashmap to run out of memory.", + "content_id": null, + "errors": null, + "id": 2, + "input_connections": {}, + "inputs": [ + { + "description": "The filters will be applied to the number of species specified in the parameter \"Search assemblies among the n closest related species\". Then the n(this parameter) most related species will be downloaded. Note: A large number of species will cause mashmap to run out of memory.", + "name": "Download n related species" + } + ], + "label": "Download n related species", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 144.47965542950726, + "top": 214.4218965438264 + }, + "tool_id": null, + "tool_state": "{\"default\": 5, \"validators\": [{\"min\": null, \"max\": null, \"negate\": false, \"type\": \"in_range\"}], \"parameter_type\": \"integer\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "60a89922-6479-4ff6-9c74-45faabf61f36", + "when": null, + "workflow_outputs": [] + }, + "3": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 3, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "Minimum Level Assembly" + } + ], + "label": "Minimum Level Assembly", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 232.67263112079243, + "top": 352.0474828512039 + }, + "tool_id": null, + "tool_state": "{\"default\": \"Complete Genome\", \"multiple\": false, \"validators\": [], \"restrictions\": [\"Complete Genome\", \"Chromosome\", \"Scaffold\", \"Contig\"], \"parameter_type\": \"text\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "ee83d451-b4f5-4051-ae3d-7979e3b3b533", + "when": null, + "workflow_outputs": [] + }, + "4": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 4, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "Assembly Type" + } + ], + "label": "Assembly Type", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 309.10130714615275, + "top": 476.9913099287952 + }, + "tool_id": null, + "tool_state": "{\"multiple\": false, \"validators\": [], \"restrictions\": [\"All\", \"Diploid\", \"Haploid\", \"Diploid or Haploid\"], \"parameter_type\": \"text\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "e0a58533-f3dd-4513-af06-d69d4f33a721", + "when": null, + "workflow_outputs": [] + }, + "5": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 5, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "Assembly BioSample Sex" + } + ], + "label": "Assembly BioSample Sex", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 377.3791151061009, + "top": 584.6799155518996 + }, + "tool_id": null, + "tool_state": "{\"multiple\": false, \"validators\": [], \"restrictions\": [\"Any\", \"Male\", \"Female\"], \"parameter_type\": \"text\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "58ba9a9e-74f8-40c6-91a2-e1ebed86694a", + "when": null, + "workflow_outputs": [] + }, + "6": { + "annotation": "Text that must be included in the sequence technology field of NCBI Dataset. Leave Empty to keep everything.", + "content_id": null, + "errors": null, + "id": 6, + "input_connections": {}, + "inputs": [ + { + "description": "Text that must be included in the sequence technology field of NCBI Dataset. Leave Empty to keep everything.", + "name": "Assembly Sequencing Technology" + } + ], + "label": "Assembly Sequencing Technology", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 445.8753126055241, + "top": 713.2976251356673 + }, + "tool_id": null, + "tool_state": "{\"multiple\": false, \"validators\": [], \"parameter_type\": \"text\", \"optional\": true}", + "tool_version": null, + "type": "parameter_input", + "uuid": "6917e47f-d37c-40a8-9893-58d652bca8aa", + "when": null, + "workflow_outputs": [] + }, + "7": { + "annotation": "Filter on Class. Comma separated and quoted values (e.g. 'Mammalia','Aves'). . Leave Empty to keep everything.", + "content_id": null, + "errors": null, + "id": 7, + "input_connections": {}, + "inputs": [ + { + "description": "Filter on Class. Comma separated and quoted values (e.g. 'Mammalia','Aves'). . Leave Empty to keep everything.", + "name": "Assembly Taxonomic Class" + } + ], + "label": "Assembly Taxonomic Class", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 535.7138726337298, + "top": 823.8251883442948 + }, + "tool_id": null, + "tool_state": "{\"multiple\": false, \"validators\": [], \"parameter_type\": \"text\", \"optional\": true}", + "tool_version": null, + "type": "parameter_input", + "uuid": "88f238cd-1424-4684-a79c-79e683ff1562", + "when": null, + "workflow_outputs": [] + }, + "8": { + "annotation": "Minimum number of chromosomes in the assemblies. Set to 0 to keep everything", + "content_id": null, + "errors": null, + "id": 8, + "input_connections": {}, + "inputs": [ + { + "description": "Minimum number of chromosomes in the assemblies. Set to 0 to keep everything", + "name": "Minimum Number Of Chromosomes" + } + ], + "label": "Minimum Number Of Chromosomes", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 619.5057456808113, + "top": 923.976374881292 + }, + "tool_id": null, + "tool_state": "{\"default\": 0, \"validators\": [{\"min\": null, \"max\": null, \"negate\": false, \"type\": \"in_range\"}], \"parameter_type\": \"integer\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "d54d72f2-78a8-4c19-a2d3-848ac9dfe883", + "when": null, + "workflow_outputs": [] + }, + "9": { + "annotation": "Maximum number of chromosomes in the assemblies. Set to 0 to keep everything", + "content_id": null, + "errors": null, + "id": 9, + "input_connections": {}, + "inputs": [ + { + "description": "Maximum number of chromosomes in the assemblies. Set to 0 to keep everything", + "name": "Maximum Number Of Chromosomes" + } + ], + "label": "Maximum Number Of Chromosomes", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 688.57899390296, + "top": 1036.3173588321774 + }, + "tool_id": null, + "tool_state": "{\"default\": 0, \"validators\": [{\"min\": null, \"max\": null, \"negate\": false, \"type\": \"in_range\"}], \"parameter_type\": \"integer\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "2cecbe72-5ec4-47bd-b43c-e28b7f999ff3", + "when": null, + "workflow_outputs": [] + }, + "10": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 10, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "Keep empty values when filtering on number of chromosomes" + } + ], + "label": "Keep empty values when filtering on number of chromosomes", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 782.0136378245954, + "top": 1184.1270602599857 + }, + "tool_id": null, + "tool_state": "{\"default\": false, \"validators\": [], \"parameter_type\": \"boolean\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "c0f2076b-ecb6-4eb0-a678-cb0ad6c832ba", + "when": null, + "workflow_outputs": [] + }, + "11": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 11, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "Search assemblies among the n closest related species *" + } + ], + "label": "Search assemblies among the n closest related species *", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 865.2441050136213, + "top": 1320.9230962430759 + }, + "tool_id": null, + "tool_state": "{\"default\": 50, \"validators\": [{\"min\": null, \"max\": null, \"negate\": false, \"type\": \"in_range\"}], \"parameter_type\": \"integer\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "91878559-d85e-47aa-9286-ba5e585aacf8", + "when": null, + "workflow_outputs": [] + }, + "12": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 12, + "input_connections": { + "input_param_type|input_param": { + "id": 3, + "output_name": "output" + } + }, + "inputs": [], + "label": "Assembly Level Filter", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_text", + "type": "expression.json" + } + ], + "position": { + "left": 1567.0240057971228, + "top": 861.3437961377745 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + }, + "HideDatasetActionoutput_param_text": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_text" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"text\", \"__current_case__\": 0, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": \"Contig\", \"to\": \"c3 in ['Contig','Scaffold','Chromosome','Complete Genome']\"}, {\"__index__\": 1, \"from\": \"Complete Genome\", \"to\": \"c3=='Complete Genome'\"}, {\"__index__\": 2, \"from\": \"Chromosome\", \"to\": \"c3 in ['Chromosome','Complete Genome']\"}, {\"__index__\": 3, \"from\": \"Scaffold\", \"to\": \"c3 in ['Scaffold','Chromosome','Complete Genome']\"}]}, \"output_param_type\": \"text\", \"unmapped\": {\"on_unmapped\": \"fail\", \"__current_case__\": 1}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "b3926f2c-bbc3-47ac-9e29-6e8d5c6b1e70", + "when": null, + "workflow_outputs": [] + }, + "13": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 13, + "input_connections": { + "input_param_type|input_param": { + "id": 4, + "output_name": "output" + } + }, + "inputs": [], + "label": "Type Value", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_text", + "type": "expression.json" + } + ], + "position": { + "left": 1594.724711081088, + "top": 1272.9381959836467 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + }, + "HideDatasetActionoutput_param_text": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_text" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"text\", \"__current_case__\": 0, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": \"Diploid\", \"to\": \"=='diploid'\"}, {\"__index__\": 1, \"from\": \"Haploid\", \"to\": \"=='haploid'\"}, {\"__index__\": 2, \"from\": \"Diploid or Haploid\", \"to\": \" in ['diploid',haploid']\"}, {\"__index__\": 3, \"from\": \"All\", \"to\": \"False\"}]}, \"output_param_type\": \"text\", \"unmapped\": {\"on_unmapped\": \"input\", \"__current_case__\": 0}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "b4f6c514-5b5e-44ee-b627-675ca188fd7b", + "when": null, + "workflow_outputs": [] + }, + "14": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 14, + "input_connections": { + "input_param_type|input_param": { + "id": 5, + "output_name": "output" + } + }, + "inputs": [], + "label": "Sex Value", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_text", + "type": "expression.json" + } + ], + "position": { + "left": 1620.3803966462413, + "top": 1574.7660948503603 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + }, + "HideDatasetActionoutput_param_text": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_text" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"text\", \"__current_case__\": 0, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": \"Any\", \"to\": \"False\"}, {\"__index__\": 1, \"from\": \"Male\", \"to\": \"=='male'\"}, {\"__index__\": 2, \"from\": \"Female\", \"to\": \"=='female'\"}]}, \"output_param_type\": \"text\", \"unmapped\": {\"on_unmapped\": \"input\", \"__current_case__\": 0}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "f24b269c-f2a1-439d-b6c1-feb158235c72", + "when": null, + "workflow_outputs": [] + }, + "15": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 15, + "input_connections": { + "input_param_type|input_param": { + "id": 6, + "output_name": "output" + } + }, + "inputs": [], + "label": "Technology Bool", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_boolean", + "type": "expression.json" + } + ], + "position": { + "left": 1565.4556263562374, + "top": 1046.1673356396343 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"text\", \"__current_case__\": 0, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": []}, \"output_param_type\": \"boolean\", \"unmapped\": {\"on_unmapped\": \"input\", \"__current_case__\": 0}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "33a4abc5-516d-4a74-af1a-69e1cd641f2f", + "when": null, + "workflow_outputs": [] + }, + "16": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "errors": null, + "id": 16, + "input_connections": { + "components_1|param_type|component_value": { + "id": 6, + "output_name": "output" + } + }, + "inputs": [], + "label": "Technology Value", + "name": "Compose text parameter value", + "outputs": [ + { + "name": "out1", + "type": "expression.json" + } + ], + "position": { + "left": 1959.0458747427067, + "top": 1184.2155682892512 + }, + "post_job_actions": { + "HideDatasetActionout1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "tool_shed_repository": { + "changeset_revision": "e188c9826e0f", + "name": "compose_text_param", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"components\": [{\"__index__\": 0, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"'\"}}, {\"__index__\": 1, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": {\"__class__\": \"ConnectedValue\"}}}, {\"__index__\": 2, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"'.lower() in c4.lower()\"}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.1", + "type": "tool", + "uuid": "a483c731-292b-4312-a262-04ac41e7f9fc", + "when": null, + "workflow_outputs": [] + }, + "17": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_text_file_with_recurring_lines/9.5+galaxy3", + "errors": null, + "id": 17, + "input_connections": { + "token_set_0|line": { + "id": 7, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Create text file", + "outputs": [ + { + "name": "outfile", + "type": "txt" + } + ], + "position": { + "left": 699.4410427990773, + "top": 522.5332123190486 + }, + "post_job_actions": { + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_text_file_with_recurring_lines/9.5+galaxy3", + "tool_shed_repository": { + "changeset_revision": "ab83aa685821", + "name": "text_processing", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"token_set\": [{\"__index__\": 0, \"line\": {\"__class__\": \"ConnectedValue\"}, \"repeat_select\": {\"repeat_select_opts\": \"user\", \"__current_case__\": 0, \"times\": \"1\"}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "9.5+galaxy3", + "type": "tool", + "uuid": "d0eb5db4-d9cb-4fa9-9033-ef9a433bc310", + "when": null, + "workflow_outputs": [] + }, + "18": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 18, + "input_connections": { + "input_param_type|input_param": { + "id": 7, + "output_name": "output" + } + }, + "inputs": [], + "label": "Class Bool ", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_boolean", + "type": "expression.json" + } + ], + "position": { + "left": 810.2900493414509, + "top": 356.013768643739 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"text\", \"__current_case__\": 0, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": []}, \"output_param_type\": \"boolean\", \"unmapped\": {\"on_unmapped\": \"input\", \"__current_case__\": 0}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "451fcf35-c498-4e4d-afc9-492bbd74fadc", + "when": null, + "workflow_outputs": [] + }, + "19": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 19, + "input_connections": { + "input_param_type|input_param": { + "id": 8, + "output_name": "output" + } + }, + "inputs": [], + "label": "Minimum Chromosomes Bool", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_boolean", + "type": "expression.json" + } + ], + "position": { + "left": 1633.2509964184928, + "top": 1748.9595613157232 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"integer\", \"__current_case__\": 1, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": \"0\", \"to\": \"False\"}]}, \"output_param_type\": \"boolean\", \"unmapped\": {\"on_unmapped\": \"default\", \"__current_case__\": 2, \"default_value\": \"True\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "97561427-9d35-4cd6-82a8-e36b532b9397", + "when": null, + "workflow_outputs": [] + }, + "20": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 20, + "input_connections": { + "input_param_type|input_param": { + "id": 8, + "output_name": "output" + } + }, + "inputs": [], + "label": "Minimum Chromosomes Value", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_text", + "type": "expression.json" + } + ], + "position": { + "left": 1642.8759503629524, + "top": 1951.1687181065486 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + }, + "HideDatasetActionoutput_param_text": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_text" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"integer\", \"__current_case__\": 1, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": []}, \"output_param_type\": \"text\", \"unmapped\": {\"on_unmapped\": \"input\", \"__current_case__\": 0}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "27d6ea1f-2119-43fd-a85a-5a5e225001f7", + "when": null, + "workflow_outputs": [] + }, + "21": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 21, + "input_connections": { + "input_param_type|input_param": { + "id": 9, + "output_name": "output" + } + }, + "inputs": [], + "label": "Maximum Chromosomes Bool", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_boolean", + "type": "expression.json" + } + ], + "position": { + "left": 1648.6324367285483, + "top": 2112.161143675939 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"integer\", \"__current_case__\": 1, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": \"0\", \"to\": \"False\"}]}, \"output_param_type\": \"boolean\", \"unmapped\": {\"on_unmapped\": \"default\", \"__current_case__\": 2, \"default_value\": \"True\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "6cd35618-a269-423f-8526-c234d93d4d27", + "when": null, + "workflow_outputs": [] + }, + "22": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 22, + "input_connections": { + "input_param_type|input_param": { + "id": 9, + "output_name": "output" + } + }, + "inputs": [], + "label": "Maximum Chromosomes Value", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_text", + "type": "expression.json" + } + ], + "position": { + "left": 1650.8014835096346, + "top": 2306.915922476275 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + }, + "HideDatasetActionoutput_param_text": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_text" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"integer\", \"__current_case__\": 1, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": []}, \"output_param_type\": \"text\", \"unmapped\": {\"on_unmapped\": \"input\", \"__current_case__\": 0}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "d3bb5ee9-2926-46bc-bdf4-64d2c5901e9f", + "when": null, + "workflow_outputs": [] + }, + "23": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/nml/refseq_masher/refseq_masher_matches/0.1.2", + "errors": null, + "id": 23, + "input_connections": { + "input|fasta": { + "id": 0, + "output_name": "output" + }, + "top_n_results": { + "id": 11, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "RefSeq Masher Matches", + "outputs": [ + { + "name": "output_path_tab", + "type": "tabular" + } + ], + "position": { + "left": 654.265018821712, + "top": 17.031679872592985 + }, + "post_job_actions": { + "HideDatasetActionoutput_path_tab": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_path_tab" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/nml/refseq_masher/refseq_masher_matches/0.1.2", + "tool_shed_repository": { + "changeset_revision": "1ec42f033bb4", + "name": "refseq_masher", + "owner": "nml", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"adv\": {\"min_kmer_threshold\": \"8\", \"output_type\": \"tab\", \"verbosity\": \"-vv\"}, \"input\": {\"type\": \"fasta\", \"__current_case__\": 0, \"fasta\": {\"__class__\": \"ConnectedValue\"}}, \"top_n_results\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.2", + "type": "tool", + "uuid": "48096e3c-89e9-4bf0-80d1-8c71437aacbc", + "when": null, + "workflow_outputs": [] + }, + "24": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "errors": null, + "id": 24, + "input_connections": { + "components_1|param_type|component_value": { + "id": 13, + "output_name": "output_param_text" + } + }, + "inputs": [], + "label": "Filter on Type expression", + "name": "Compose text parameter value", + "outputs": [ + { + "name": "out1", + "type": "expression.json" + } + ], + "position": { + "left": 2110.711917310815, + "top": 1369.5361298832447 + }, + "post_job_actions": { + "HideDatasetActionout1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "tool_shed_repository": { + "changeset_revision": "e188c9826e0f", + "name": "compose_text_param", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"components\": [{\"__index__\": 0, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"c6\"}}, {\"__index__\": 1, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.1", + "type": "tool", + "uuid": "4cbfd6f6-3d86-4d7c-8865-23d4520741c1", + "when": null, + "workflow_outputs": [] + }, + "25": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 25, + "input_connections": { + "input_param_type|input_param": { + "id": 13, + "output_name": "output_param_text" + } + }, + "inputs": [], + "label": "Type bool", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_boolean", + "type": "expression.json" + } + ], + "position": { + "left": 2414.945257133506, + "top": 1511.8574599186832 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + }, + "HideDatasetActionoutput_param_text": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_text" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"text\", \"__current_case__\": 0, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": \"False\", \"to\": \"False\"}]}, \"output_param_type\": \"boolean\", \"unmapped\": {\"on_unmapped\": \"default\", \"__current_case__\": 2, \"default_value\": \"True\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "3cf3ccdb-e131-4b0b-a2fd-841db158e902", + "when": null, + "workflow_outputs": [] + }, + "26": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 26, + "input_connections": { + "input_param_type|input_param": { + "id": 14, + "output_name": "output_param_text" + } + }, + "inputs": [], + "label": "Sex bool", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_boolean", + "type": "expression.json" + } + ], + "position": { + "left": 2124.5729894097485, + "top": 1564.4370779727476 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + }, + "HideDatasetActionoutput_param_text": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_text" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"text\", \"__current_case__\": 0, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": \"False\", \"to\": \"False\"}]}, \"output_param_type\": \"boolean\", \"unmapped\": {\"on_unmapped\": \"default\", \"__current_case__\": 2, \"default_value\": \"True\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "2881e9f6-8787-406d-83b0-4225cb668dcf", + "when": null, + "workflow_outputs": [] + }, + "27": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "errors": null, + "id": 27, + "input_connections": { + "components_1|param_type|component_value": { + "id": 14, + "output_name": "output_param_text" + } + }, + "inputs": [], + "label": "Filter on Sex expression", + "name": "Compose text parameter value", + "outputs": [ + { + "name": "out1", + "type": "expression.json" + } + ], + "position": { + "left": 2136.891242288172, + "top": 1715.7489342986767 + }, + "post_job_actions": { + "HideDatasetActionout1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "tool_shed_repository": { + "changeset_revision": "e188c9826e0f", + "name": "compose_text_param", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"components\": [{\"__index__\": 0, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"c5\"}}, {\"__index__\": 1, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.1", + "type": "tool", + "uuid": "77703d15-7b45-47e9-afec-1b87c3c51c9e", + "when": null, + "workflow_outputs": [] + }, + "28": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_find_and_replace/9.5+galaxy3", + "errors": null, + "id": 28, + "input_connections": { + "infile": { + "id": 17, + "output_name": "outfile" + } + }, + "inputs": [], + "label": null, + "name": "Replace", + "outputs": [ + { + "name": "outfile", + "type": "input" + } + ], + "position": { + "left": 835.9218928393063, + "top": 773.6525813785477 + }, + "post_job_actions": { + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_find_and_replace/9.5+galaxy3", + "tool_shed_repository": { + "changeset_revision": "ab83aa685821", + "name": "text_processing", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"find_and_replace\": [{\"__index__\": 0, \"find_pattern\": \"\\\\s\", \"replace_pattern\": \"\", \"is_regex\": true, \"global\": true, \"caseinsensitive\": false, \"wholewords\": false, \"skip_first_line\": false, \"searchwhere\": {\"searchwhere_select\": \"line\", \"__current_case__\": 0}}, {\"__index__\": 1, \"find_pattern\": \",\", \"replace_pattern\": \"\\\\x27,\\\\x27\", \"is_regex\": true, \"global\": true, \"caseinsensitive\": false, \"wholewords\": false, \"skip_first_line\": false, \"searchwhere\": {\"searchwhere_select\": \"line\", \"__current_case__\": 0}}, {\"__index__\": 2, \"find_pattern\": \"(\\\\w+)\", \"replace_pattern\": \"\\\\L$1\", \"is_regex\": true, \"global\": true, \"caseinsensitive\": false, \"wholewords\": false, \"skip_first_line\": false, \"searchwhere\": {\"searchwhere_select\": \"line\", \"__current_case__\": 0}}], \"infile\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "9.5+galaxy3", + "type": "tool", + "uuid": "a67da362-d5e6-4592-81b3-9f4f4bbb4b84", + "when": null, + "workflow_outputs": [] + }, + "29": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "errors": null, + "id": 29, + "input_connections": { + "components_1|param_type|component_value": { + "id": 20, + "output_name": "output_param_text" + }, + "when": { + "id": 19, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "Filter on Minimum chromosomes expression", + "name": "Compose text parameter value", + "outputs": [ + { + "name": "out1", + "type": "expression.json" + } + ], + "position": { + "left": 2179.376078846294, + "top": 1935.6193294130028 + }, + "post_job_actions": { + "HideDatasetActionout1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "tool_shed_repository": { + "changeset_revision": "e188c9826e0f", + "name": "compose_text_param", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"components\": [{\"__index__\": 0, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"c7>\"}}, {\"__index__\": 1, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.1", + "type": "tool", + "uuid": "e1660f54-8f4d-4ea8-842d-f03eebb8d816", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "30": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "errors": null, + "id": 30, + "input_connections": { + "components_1|param_type|component_value": { + "id": 22, + "output_name": "output_param_text" + }, + "when": { + "id": 21, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "Filter on Maximum chromosomes expression", + "name": "Compose text parameter value", + "outputs": [ + { + "name": "out1", + "type": "expression.json" + } + ], + "position": { + "left": 2180.925977852387, + "top": 2226.2330468776486 + }, + "post_job_actions": { + "HideDatasetActionout1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "tool_shed_repository": { + "changeset_revision": "e188c9826e0f", + "name": "compose_text_param", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"components\": [{\"__index__\": 0, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"c7<\"}}, {\"__index__\": 1, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.1", + "type": "tool", + "uuid": "d82e67a3-d58c-4853-91ed-6f80fc08876f", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "31": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/column_arrange_by_header/bg_column_arrange_by_header/0.3", + "errors": null, + "id": 31, + "input_connections": { + "param_input": { + "id": 23, + "output_name": "output_path_tab" + } + }, + "inputs": [], + "label": null, + "name": "Column arrange", + "outputs": [ + { + "name": "output", + "type": "tabular" + } + ], + "position": { + "left": 938.9657923232129, + "top": 89.46858671265326 + }, + "post_job_actions": { + "RenameDatasetActionoutput": { + "action_arguments": { + "newname": "All Refseq Matches" + }, + "action_type": "RenameDatasetAction", + "output_name": "output" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/column_arrange_by_header/bg_column_arrange_by_header/0.3", + "tool_shed_repository": { + "changeset_revision": "dcc6bb900eee", + "name": "column_arrange_by_header", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"discard_remaining\": true, \"param_input\": {\"__class__\": \"ConnectedValue\"}, \"rep_param_columns\": [{\"__index__\": 0, \"param_column\": \"assembly_accession\"}, {\"__index__\": 1, \"param_column\": \"taxonomic_species\"}, {\"__index__\": 2, \"param_column\": \"taxonomic_genus\"}, {\"__index__\": 3, \"param_column\": \"taxonomic_family\"}, {\"__index__\": 4, \"param_column\": \"taxonomic_order\"}, {\"__index__\": 5, \"param_column\": \"taxonomic_class\"}, {\"__index__\": 6, \"param_column\": \"taxonomic_phylum\"}, {\"__index__\": 7, \"param_column\": \"taxonomic_superkingdom\"}, {\"__index__\": 8, \"param_column\": \"distance\"}, {\"__index__\": 9, \"param_column\": \"pvalue\"}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.3", + "type": "tool", + "uuid": "241d0db8-bcfc-44c2-847e-3aadb6f7f6e2", + "when": null, + "workflow_outputs": [ + { + "label": "All Refseq Matches", + "output_name": "output", + "uuid": "89d6ad2a-7a6b-4e89-a033-c9415211f574" + } + ] + }, + "32": { + "annotation": "", + "content_id": "param_value_from_file", + "errors": null, + "id": 32, + "input_connections": { + "input1": { + "id": 28, + "output_name": "outfile" + } + }, + "inputs": [], + "label": null, + "name": "Parse parameter value", + "outputs": [ + { + "name": "text_param", + "type": "expression.json" + } + ], + "position": { + "left": 1037.4026986170777, + "top": 552.1875171241254 + }, + "post_job_actions": { + "HideDatasetActiontext_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "text_param" + } + }, + "tool_id": "param_value_from_file", + "tool_state": "{\"input1\": {\"__class__\": \"ConnectedValue\"}, \"param_type\": \"text\", \"remove_newlines\": true, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.0", + "type": "tool", + "uuid": "0b773d81-982d-44de-8a39-4f57075be809", + "when": null, + "workflow_outputs": [] + }, + "33": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "errors": null, + "id": 33, + "input_connections": { + "components_1|param_type|component_value": { + "id": 29, + "output_name": "out1" + }, + "when": { + "id": 10, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Compose text parameter value", + "outputs": [ + { + "name": "out1", + "type": "expression.json" + } + ], + "position": { + "left": 2568.2170618473956, + "top": 2283.1616434799244 + }, + "post_job_actions": { + "HideDatasetActionout1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "tool_shed_repository": { + "changeset_revision": "e188c9826e0f", + "name": "compose_text_param", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"components\": [{\"__index__\": 0, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"c9>\"}}, {\"__index__\": 1, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": {\"__class__\": \"ConnectedValue\"}}}, {\"__index__\": 2, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \" or c9==''\"}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.1", + "type": "tool", + "uuid": "02d054e2-dbc3-4067-bc23-a669104a0215", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "34": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "errors": null, + "id": 34, + "input_connections": { + "components_1|param_type|component_value": { + "id": 30, + "output_name": "out1" + }, + "when": { + "id": 10, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Compose text parameter value", + "outputs": [ + { + "name": "out1", + "type": "expression.json" + } + ], + "position": { + "left": 2635.5872589143787, + "top": 2552.515656171764 + }, + "post_job_actions": { + "HideDatasetActionout1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "tool_shed_repository": { + "changeset_revision": "e188c9826e0f", + "name": "compose_text_param", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"components\": [{\"__index__\": 0, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"c9>\"}}, {\"__index__\": 1, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": {\"__class__\": \"ConnectedValue\"}}}, {\"__index__\": 2, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \" or c9==''\"}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.1", + "type": "tool", + "uuid": "d780baf4-4736-4407-9346-db011055f182", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "35": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "errors": null, + "id": 35, + "input_connections": { + "components_1|param_type|component_value": { + "id": 32, + "output_name": "text_param" + } + }, + "inputs": [], + "label": "Class Value", + "name": "Compose text parameter value", + "outputs": [ + { + "name": "out1", + "type": "expression.json" + } + ], + "position": { + "left": 1282.1737495510542, + "top": 632.5388878761851 + }, + "post_job_actions": { + "HideDatasetActionout1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/compose_text_param/compose_text_param/0.1.1", + "tool_shed_repository": { + "changeset_revision": "e188c9826e0f", + "name": "compose_text_param", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"components\": [{\"__index__\": 0, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"c6.lower() in ['\"}}, {\"__index__\": 1, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": {\"__class__\": \"ConnectedValue\"}}}, {\"__index__\": 2, \"param_type\": {\"select_param_type\": \"text\", \"__current_case__\": 0, \"component_value\": \"']\"}}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.1", + "type": "tool", + "uuid": "b7d56e67-5b3f-464d-b725-c75dfdb087c4", + "when": null, + "workflow_outputs": [] + }, + "36": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 36, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 33, + "output_name": "out1" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 29, + "output_name": "out1" + }, + "style_cond|type_cond|pick_from_2|value": { + "id": 20, + "output_name": "output_param_text" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "text_param", + "type": "expression.json" + } + ], + "position": { + "left": 3026.739546233963, + "top": 2289.6142208486376 + }, + "post_job_actions": { + "HideDatasetActiontext_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "text_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"text\", \"__current_case__\": 0, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 2, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "232295de-d26c-404b-8b47-0f709e588489", + "when": null, + "workflow_outputs": [] + }, + "37": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 37, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 34, + "output_name": "out1" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 30, + "output_name": "out1" + }, + "style_cond|type_cond|pick_from_2|value": { + "id": 22, + "output_name": "output_param_text" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "text_param", + "type": "expression.json" + } + ], + "position": { + "left": 3008.964664091213, + "top": 2564.8856835454276 + }, + "post_job_actions": { + "HideDatasetActiontext_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "text_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"text\", \"__current_case__\": 0, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 2, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "994d0de4-f747-4530-ac34-a47c73b2f8bb", + "when": null, + "workflow_outputs": [] + }, + "38": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 38, + "input_connections": { + "cond": { + "id": 35, + "output_name": "out1" + }, + "input": { + "id": 31, + "output_name": "output" + }, + "when": { + "id": 18, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "Filter on Type 2", + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 1145.6560892417244, + "top": 250.18981726621388 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": {\"__class__\": \"ConnectedValue\"}, \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "41f9fd28-aed4-4fd6-a1ae-c213619bbf8f", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "39": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 39, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 38, + "output_name": "out_file1" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 31, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "data_param", + "type": "input" + } + ], + "position": { + "left": 1438.7315908170772, + "top": 384.3264071087019 + }, + "post_job_actions": { + "RenameDatasetActiondata_param": { + "action_arguments": { + "newname": "Filtered Refseq Matches" + }, + "action_type": "RenameDatasetAction", + "output_name": "data_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"data\", \"__current_case__\": 4, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "38cfc5bc-6cb7-4a54-ba46-3bc4a29c9e8c", + "when": null, + "workflow_outputs": [ + { + "label": "Filtered Refseq Matches", + "output_name": "data_param", + "uuid": "1827562b-252e-489e-9e67-ffcc3b4aa80a" + } + ] + }, + "40": { + "annotation": "", + "content_id": "Cut1", + "errors": null, + "id": 40, + "input_connections": { + "input": { + "id": 39, + "output_name": "data_param" + } + }, + "inputs": [], + "label": null, + "name": "Cut", + "outputs": [ + { + "name": "out_file1", + "type": "tabular" + } + ], + "position": { + "left": 1570.270285341267, + "top": 194.23912596377386 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Cut1", + "tool_state": "{\"columnList\": \"c1\", \"delimiter\": \"T\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.0.2", + "type": "tool", + "uuid": "8fc793e7-330b-4faa-99f2-1050ede956f7", + "when": null, + "workflow_outputs": [] + }, + "41": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_tail_tool/9.5+galaxy3", + "errors": null, + "id": 41, + "input_connections": { + "infile": { + "id": 40, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Select last", + "outputs": [ + { + "name": "outfile", + "type": "input" + } + ], + "position": { + "left": 1704.3713571423743, + "top": 467.08047533886383 + }, + "post_job_actions": { + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_tail_tool/9.5+galaxy3", + "tool_shed_repository": { + "changeset_revision": "ab83aa685821", + "name": "text_processing", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"complement\": \"+\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"num_lines\": \"2\", \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "9.5+galaxy3", + "type": "tool", + "uuid": "e8b94d1b-c296-49e1-9e39-06abf7ead1e2", + "when": null, + "workflow_outputs": [] + }, + "42": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 42, + "input_connections": { + "input": { + "id": 41, + "output_name": "outfile" + } + }, + "inputs": [], + "label": null, + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 1973.3770368350831, + "top": 411.9589954730579 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": \"\\\"GCA\\\" in c1\", \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "ff02967b-484e-4eca-82fb-d11a072ba5f6", + "when": null, + "workflow_outputs": [] + }, + "43": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 43, + "input_connections": { + "input": { + "id": 41, + "output_name": "outfile" + } + }, + "inputs": [], + "label": null, + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 1972.6013169425814, + "top": 628.5167254529836 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": \"\\\"GCF\\\" in c1\", \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "2df2c2f5-32aa-475b-9c69-01effd09a5fa", + "when": null, + "workflow_outputs": [] + }, + "44": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/mvdbeek/concatenate_multiple_datasets/cat_multiple/0.2", + "errors": null, + "id": 44, + "input_connections": { + "input": [ + { + "id": 42, + "output_name": "out_file1" + }, + { + "id": 43, + "output_name": "out_file1" + } + ] + }, + "inputs": [], + "label": null, + "name": "Concatenate multiple datasets", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 2283.551931964627, + "top": 590.7729902796618 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/mvdbeek/concatenate_multiple_datasets/cat_multiple/0.2", + "tool_shed_repository": { + "changeset_revision": "201c568972c3", + "name": "concatenate_multiple_datasets", + "owner": "mvdbeek", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2", + "type": "tool", + "uuid": "dca04711-c0e7-4fa9-8505-c73171d0f617", + "when": null, + "workflow_outputs": [] + }, + "45": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/ncbi_datasets/datasets_download_genome/18.21.0+galaxy0", + "errors": null, + "id": 45, + "input_connections": { + "query|subcommand|text_or_file|inputfile": { + "id": 44, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "NCBI Datasets Genomes", + "outputs": [ + { + "name": "genome_data_report", + "type": "tabular" + } + ], + "position": { + "left": 2296.155913853448, + "top": 948.1794042407084 + }, + "post_job_actions": { + "HideDatasetActiongenome_fasta": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "genome_fasta" + }, + "RenameDatasetActiongenome_data_report": { + "action_arguments": { + "newname": "All Related Species Genome Report" + }, + "action_type": "RenameDatasetAction", + "output_name": "genome_data_report" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/ncbi_datasets/datasets_download_genome/18.21.0+galaxy0", + "tool_shed_repository": { + "changeset_revision": "6ba50cbabbb7", + "name": "ncbi_datasets", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"file_choices\": {\"report_columns\": [\"accession\", \"assminfo-name\", \"assminfo-submitter\", \"organism-name\", \"assminfo-type\", \"assminfo-level\", \"assminfo-biosample-sex\", \"assminfo-sequencing-tech\", \"assmstats-total-number-of-chromosomes\"], \"include\": [], \"decompress\": false}, \"filters\": {\"reference\": false, \"annotated\": false, \"assembly_level\": null, \"assembly_version\": \"latest\", \"assembly_source\": null, \"chromosomes\": null, \"exclude_atypical\": false, \"mag\": null, \"released_before\": null, \"released_after\": null, \"search\": []}, \"query\": {\"subcommand\": {\"download_by\": \"accession\", \"__current_case__\": 0, \"text_or_file\": {\"text_or_file\": \"file\", \"__current_case__\": 1, \"inputfile\": {\"__class__\": \"ConnectedValue\"}}}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "18.21.0+galaxy0", + "type": "tool", + "uuid": "78000bc7-519c-4f2c-831b-ad6d28770f50", + "when": null, + "workflow_outputs": [ + { + "label": "All Related Species Genome Report", + "output_name": "genome_data_report", + "uuid": "6e512d53-74e7-4079-8ab5-6b3df96e205a" + } + ] + }, + "46": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/column_arrange_by_header/bg_column_arrange_by_header/0.3", + "errors": null, + "id": 46, + "input_connections": { + "param_input": { + "id": 45, + "output_name": "genome_data_report" + } + }, + "inputs": [], + "label": null, + "name": "Column arrange", + "outputs": [ + { + "name": "output", + "type": "tabular" + } + ], + "position": { + "left": 2563.862584125438, + "top": 895.8207933298966 + }, + "post_job_actions": { + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" + }, + "RenameDatasetActionoutput": { + "action_arguments": { + "newname": "All Refseq Matches" + }, + "action_type": "RenameDatasetAction", + "output_name": "output" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/column_arrange_by_header/bg_column_arrange_by_header/0.3", + "tool_shed_repository": { + "changeset_revision": "dcc6bb900eee", + "name": "column_arrange_by_header", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"discard_remaining\": false, \"param_input\": {\"__class__\": \"ConnectedValue\"}, \"rep_param_columns\": [{\"__index__\": 0, \"param_column\": \"Assembly Accession\"}, {\"__index__\": 1, \"param_column\": \"Assembly Name\"}, {\"__index__\": 2, \"param_column\": \"Assembly Level\"}, {\"__index__\": 3, \"param_column\": \"Assembly Sequencing Tech\"}, {\"__index__\": 4, \"param_column\": \"Assembly BioSample Sex\"}, {\"__index__\": 5, \"param_column\": \"Assembly Type\"}, {\"__index__\": 6, \"param_column\": \"Assembly Stats Total Number of Chromosomes\"}], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.3", + "type": "tool", + "uuid": "d24a863d-bb6d-49fc-aae6-93d555553006", + "when": null, + "workflow_outputs": [] + }, + "47": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 47, + "input_connections": { + "cond": { + "id": 12, + "output_name": "output_param_text" + }, + "input": { + "id": 46, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 2651.3936219836983, + "top": 1201.261411333804 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": {\"__class__\": \"ConnectedValue\"}, \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "f5fccca0-9e9f-49dc-9fd4-f0ee80aef726", + "when": null, + "workflow_outputs": [] + }, + "48": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 48, + "input_connections": { + "cond": { + "id": 16, + "output_name": "out1" + }, + "input": { + "id": 47, + "output_name": "out_file1" + }, + "when": { + "id": 15, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "Filter on Technology", + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 2795.8742335040324, + "top": 1429.6922421328197 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": {\"__class__\": \"ConnectedValue\"}, \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "67f87a6c-de0d-4720-8532-9bb4ab10c80f", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "49": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 49, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 48, + "output_name": "out_file1" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 47, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "data_param", + "type": "input" + } + ], + "position": { + "left": 3099.258616166827, + "top": 1303.9324694359668 + }, + "post_job_actions": { + "HideDatasetActiondata_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "data_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"data\", \"__current_case__\": 4, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "4897b9d0-ec5a-4775-b150-0f823a507614", + "when": null, + "workflow_outputs": [] + }, + "50": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 50, + "input_connections": { + "cond": { + "id": 24, + "output_name": "out1" + }, + "input": { + "id": 49, + "output_name": "data_param" + }, + "when": { + "id": 25, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "Filter on Type", + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 3261.6498894807896, + "top": 1755.6532846844327 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": {\"__class__\": \"ConnectedValue\"}, \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "30ba40bc-9897-426b-9431-cb97da2e7fdc", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "51": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 51, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 50, + "output_name": "out_file1" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 49, + "output_name": "data_param" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "data_param", + "type": "input" + } + ], + "position": { + "left": 3510.918424520682, + "top": 1523.6691850913755 + }, + "post_job_actions": { + "HideDatasetActiondata_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "data_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"data\", \"__current_case__\": 4, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "27611965-5085-452d-b384-2794530c9a87", + "when": null, + "workflow_outputs": [] + }, + "52": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 52, + "input_connections": { + "cond": { + "id": 27, + "output_name": "out1" + }, + "input": { + "id": 51, + "output_name": "data_param" + }, + "when": { + "id": 26, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "Filter on Sex", + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 3554.8683940034525, + "top": 2022.281836665902 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": {\"__class__\": \"ConnectedValue\"}, \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "aa19d60a-1e1f-4e7e-95c3-3379254eb26d", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "53": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 53, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 52, + "output_name": "out_file1" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 51, + "output_name": "data_param" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "data_param", + "type": "input" + } + ], + "position": { + "left": 3810.5265312051297, + "top": 1812.019093279344 + }, + "post_job_actions": { + "HideDatasetActiondata_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "data_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"data\", \"__current_case__\": 4, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "f3dbd8bf-0ca1-454d-9300-c9658aaa7810", + "when": null, + "workflow_outputs": [] + }, + "54": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 54, + "input_connections": { + "cond": { + "id": 36, + "output_name": "text_param" + }, + "input": { + "id": 53, + "output_name": "data_param" + }, + "when": { + "id": 19, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "Filter on Maximum Chromosomes", + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 4123.121558550249, + "top": 1678.278447697747 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": {\"__class__\": \"ConnectedValue\"}, \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "9acca961-6dac-45d4-ac5e-ef7d552f2f21", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "55": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 55, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 54, + "output_name": "out_file1" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 49, + "output_name": "data_param" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "data_param", + "type": "input" + } + ], + "position": { + "left": 4271.8132076413585, + "top": 1925.131374881494 + }, + "post_job_actions": { + "HideDatasetActiondata_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "data_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"data\", \"__current_case__\": 4, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "16ba855d-928b-4e9c-810f-22327676cd3b", + "when": null, + "workflow_outputs": [] + }, + "56": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 56, + "input_connections": { + "cond": { + "id": 37, + "output_name": "text_param" + }, + "input": { + "id": 55, + "output_name": "data_param" + }, + "when": { + "id": 21, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "Filter on Minimum Chromosomes 2", + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 4496.952134937062, + "top": 2146.866661879531 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": {\"__class__\": \"ConnectedValue\"}, \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.1.1", + "type": "tool", + "uuid": "64c659c2-9f3e-48d2-8db1-9062976cd043", + "when": "$(inputs.when)", + "workflow_outputs": [] + }, + "57": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 57, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 56, + "output_name": "out_file1" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 55, + "output_name": "data_param" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "data_param", + "type": "input" + } + ], + "position": { + "left": 4616.665655144189, + "top": 1669.2857468984946 + }, + "post_job_actions": { + "RenameDatasetActiondata_param": { + "action_arguments": { + "newname": "Filtered Genome Data" + }, + "action_type": "RenameDatasetAction", + "output_name": "data_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"data\", \"__current_case__\": 4, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "0065158a-491d-4bb9-b613-ae781ec65696", + "when": null, + "workflow_outputs": [ + { + "label": "Filtered Genome Data", + "output_name": "data_param", + "uuid": "2cbcf762-1532-49b8-9a1a-1799731b592f" + } + ] + }, + "58": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_easyjoin_tool/9.5+galaxy3", + "errors": null, + "id": 58, + "input_connections": { + "infile1": { + "id": 57, + "output_name": "data_param" + }, + "infile2": { + "id": 31, + "output_name": "output" + } + }, + "inputs": [], + "label": "Result Species", + "name": "Join", + "outputs": [ + { + "name": "output", + "type": "input" + } + ], + "position": { + "left": 4860.459772326877, + "top": 1426.4262458289288 + }, + "post_job_actions": { + "RenameDatasetActionoutput": { + "action_arguments": { + "newname": "Result Species" + }, + "action_type": "RenameDatasetAction", + "output_name": "output" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_easyjoin_tool/9.5+galaxy3", + "tool_shed_repository": { + "changeset_revision": "ab83aa685821", + "name": "text_processing", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"column1\": \"1\", \"column2\": \"1\", \"empty_string_filler\": \"0\", \"header\": false, \"ignore_case\": false, \"infile1\": {\"__class__\": \"ConnectedValue\"}, \"infile2\": {\"__class__\": \"ConnectedValue\"}, \"jointype\": \" \", \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "9.5+galaxy3", + "type": "tool", + "uuid": "04c77150-7e79-4427-ab88-10963214f93c", + "when": null, + "workflow_outputs": [ + { + "label": "Result Species", + "output_name": "output", + "uuid": "20c6934e-c1eb-41dc-9452-8305b3a2c03f" + } + ] + }, + "59": { + "annotation": "", + "id": 59, + "input_connections": { + "Candidate file. ": { + "id": 58, + "input_subworkflow_step_id": 0, + "output_name": "output" + }, + "Keep n": { + "id": 2, + "input_subworkflow_step_id": 1, + "output_name": "output" + }, + "when": { + "id": 1, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Download sequences", + "outputs": [], + "position": { + "left": 5516.908405364885, + "top": 1663.6273455172739 + }, + "subworkflow": { + "a_galaxy_workflow": "true", + "annotation": "", + "comments": [], + "creator": [ + { + "class": "Person", + "identifier": "https://orcid.org/0000-0001-6421-3484", + "name": "Delphine Lariviere" + }, + { + "class": "Organization", + "name": "IWC" + } + ], + "format-version": "0.1", + "license": "MIT", + "name": "Download sequences", + "report": { + "markdown": "\n# Workflow Execution Report\n\n## Workflow Inputs\n```galaxy\ninvocation_inputs()\n```\n\n## Workflow Outputs\n```galaxy\ninvocation_outputs()\n```\n\n## Workflow\n```galaxy\nworkflow_display()\n```\n" + }, + "steps": { + "0": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 0, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "Candidate file. " + } + ], + "label": "Candidate file. ", + "name": "Input dataset", + "outputs": [], + "position": { + "left": 0.0, + "top": 419.76679959111556 + }, + "tool_id": null, + "tool_state": "{\"optional\": false, \"tag\": null}", + "tool_version": null, + "type": "data_input", + "uuid": "b3e0f07e-553a-4397-9f34-0e7b8566c594", + "when": null, + "workflow_outputs": [] + }, + "1": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 1, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "Keep n" + } + ], + "label": "Keep n", + "name": "Input parameter", + "outputs": [], + "position": { + "left": 22.19141971346084, + "top": 582.4475236684381 + }, + "tool_id": null, + "tool_state": "{\"validators\": [{\"min\": null, \"max\": null, \"negate\": false, \"type\": \"in_range\"}], \"parameter_type\": \"integer\", \"optional\": false}", + "tool_version": null, + "type": "parameter_input", + "uuid": "39cc6a38-87f4-456b-aee0-faebb80dedfb", + "when": null, + "workflow_outputs": [] + }, + "2": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_tail_tool/9.5+galaxy3", + "errors": null, + "id": 2, + "input_connections": { + "infile": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Select last", + "outputs": [ + { + "name": "outfile", + "type": "input" + } + ], + "position": { + "left": 552.6728072500974, + "top": 529.988427032152 + }, + "post_job_actions": { + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_tail_tool/9.5+galaxy3", + "tool_shed_repository": { + "changeset_revision": "ab83aa685821", + "name": "text_processing", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"complement\": \"+\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"num_lines\": \"2\", \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "9.5+galaxy3", + "type": "tool", + "uuid": "407a5d09-5438-4978-8278-27fe0197ca33", + "when": null, + "workflow_outputs": [] + }, + "3": { + "annotation": "", + "content_id": "Show beginning1", + "errors": null, + "id": 3, + "input_connections": { + "input": { + "id": 2, + "output_name": "outfile" + }, + "lineNum": { + "id": 1, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Select first", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 720.5031510158143, + "top": 727.6879912410002 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Show beginning1", + "tool_state": "{\"header\": false, \"input\": {\"__class__\": \"RuntimeValue\"}, \"lineNum\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.0.2", + "type": "tool", + "uuid": "55429cf3-108d-4e34-b5cf-2ca257e56f7a", + "when": null, + "workflow_outputs": [] + }, + "4": { + "annotation": "", + "content_id": "Cut1", + "errors": null, + "id": 4, + "input_connections": { + "input": { + "id": 3, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Cut", + "outputs": [ + { + "name": "out_file1", + "type": "tabular" + } + ], + "position": { + "left": 871.3858365142088, + "top": 480.212937372733 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Cut1", + "tool_state": "{\"columnList\": \"c1\", \"delimiter\": \"T\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.0.2", + "type": "tool", + "uuid": "bee6acc8-fa62-41a6-b97e-153e739085d1", + "when": null, + "workflow_outputs": [] + }, + "5": { + "annotation": "", + "content_id": "wc_gnu", + "errors": null, + "id": 5, + "input_connections": { + "input1": { + "id": 4, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Line/Word/Character count", + "outputs": [ + { + "name": "out_file1", + "type": "tabular" + } + ], + "position": { + "left": 1044.7224187144432, + "top": 270.5949419630715 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "wc_gnu", + "tool_state": "{\"include_header\": false, \"input1\": {\"__class__\": \"ConnectedValue\"}, \"options\": [\"lines\"], \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.0.0", + "type": "tool", + "uuid": "bf941e92-9b91-49c6-bb85-f196e0554611", + "when": null, + "workflow_outputs": [] + }, + "6": { + "annotation": "", + "content_id": "param_value_from_file", + "errors": null, + "id": 6, + "input_connections": { + "input1": { + "id": 5, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Parse parameter value", + "outputs": [ + { + "name": "integer_param", + "type": "expression.json" + } + ], + "position": { + "left": 1255.3476994987934, + "top": 423.06292023636524 + }, + "post_job_actions": { + "HideDatasetActioninteger_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "integer_param" + } + }, + "tool_id": "param_value_from_file", + "tool_state": "{\"input1\": {\"__class__\": \"ConnectedValue\"}, \"param_type\": \"integer\", \"remove_newlines\": true, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.1.0", + "type": "tool", + "uuid": "294b6a43-fd7e-4155-b3eb-6221af13992d", + "when": null, + "workflow_outputs": [] + }, + "7": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 7, + "input_connections": { + "input_param_type|input_param": { + "id": 6, + "output_name": "integer_param" + } + }, + "inputs": [], + "label": null, + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_boolean", + "type": "expression.json" + } + ], + "position": { + "left": 1464.2323588180182, + "top": 226.39163162398745 + }, + "post_job_actions": { + "HideDatasetActionoutput_param_boolean": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_param_boolean" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"integer\", \"__current_case__\": 1, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": \"0\", \"to\": \"False\"}]}, \"output_param_type\": \"boolean\", \"unmapped\": {\"on_unmapped\": \"default\", \"__current_case__\": 2, \"default_value\": \"True\"}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "20a6fa7b-9c6e-412c-b5c3-74e4083534f0", + "when": null, + "workflow_outputs": [] + }, + "8": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/ncbi_datasets/datasets_download_genome/18.21.0+galaxy0", + "errors": null, + "id": 8, + "input_connections": { + "query|subcommand|text_or_file|inputfile": { + "id": 4, + "output_name": "out_file1" + }, + "when": { + "id": 7, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": null, + "name": "NCBI Datasets Genomes", + "outputs": [ + { + "name": "sequence_report", + "type": "input" + }, + { + "name": "genome_fasta", + "type": "input" + }, + { + "name": "genome_data_report", + "type": "tabular" + } + ], + "position": { + "left": 1728.7591001542162, + "top": 473.6861453296942 + }, + "post_job_actions": { + "HideDatasetActiongenome_fasta": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "genome_fasta" + }, + "RenameDatasetActiongenome_data_report": { + "action_arguments": { + "newname": "Filtered Related Species Genome Report" + }, + "action_type": "RenameDatasetAction", + "output_name": "genome_data_report" + }, + "RenameDatasetActionsequence_report": { + "action_arguments": { + "newname": "Related Species Sequence Reports" + }, + "action_type": "RenameDatasetAction", + "output_name": "sequence_report" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/ncbi_datasets/datasets_download_genome/18.21.0+galaxy0", + "tool_shed_repository": { + "changeset_revision": "6ba50cbabbb7", + "name": "ncbi_datasets", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"file_choices\": {\"report_columns\": [\"accession\", \"assminfo-name\", \"assminfo-submitter\", \"organism-name\", \"assminfo-type\", \"assminfo-level\", \"assminfo-biosample-sex\", \"assminfo-sequencing-tech\", \"assmstats-total-number-of-chromosomes\"], \"include\": [\"genome\", \"seq-report\"], \"decompress\": false}, \"filters\": {\"reference\": false, \"annotated\": false, \"assembly_level\": null, \"assembly_version\": \"latest\", \"assembly_source\": null, \"chromosomes\": null, \"exclude_atypical\": false, \"mag\": null, \"released_before\": null, \"released_after\": null, \"search\": []}, \"query\": {\"subcommand\": {\"download_by\": \"accession\", \"__current_case__\": 0, \"text_or_file\": {\"text_or_file\": \"file\", \"__current_case__\": 1, \"inputfile\": {\"__class__\": \"ConnectedValue\"}}}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "18.21.0+galaxy0", + "type": "tool", + "uuid": "9550b81e-a4e4-4493-a6f3-2a77ccfa43b7", + "when": "$(inputs.when)", + "workflow_outputs": [ + { + "label": "Related Species Sequence Reports", + "output_name": "sequence_report", + "uuid": "fb0d75d6-fc36-4bba-82d8-9e4d1058221c" + }, + { + "label": "Filtered Related Species Genome Report", + "output_name": "genome_data_report", + "uuid": "febc321c-819f-4a55-abb7-05f3a4fd440c" + } + ] + }, + "9": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "errors": null, + "id": 9, + "input_connections": { + "input_param_type|input_param": { + "id": 7, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": "This step fails if no related species is found that matches the filters", + "name": "Map parameter value", + "outputs": [ + { + "name": "output_param_text", + "type": "expression.json" + } + ], + "position": { + "left": 1939.8116134753584, + "top": 0 + }, + "post_job_actions": { + "RenameDatasetActionoutput_param_text": { + "action_arguments": { + "newname": "This step fails if no related species is found that matches the filters" + }, + "action_type": "RenameDatasetAction", + "output_name": "output_param_text" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/map_param_value/map_param_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "5ac8a4bf7a8d", + "name": "map_param_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"input_param_type\": {\"type\": \"boolean\", \"__current_case__\": 3, \"input_param\": {\"__class__\": \"ConnectedValue\"}, \"mappings\": [{\"__index__\": 0, \"from\": true, \"to\": \"True\"}]}, \"output_param_type\": \"text\", \"unmapped\": {\"on_unmapped\": \"fail\", \"__current_case__\": 1}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "f096ca43-81ef-42c9-8bef-f634bf134c6d", + "when": null, + "workflow_outputs": [ + { + "label": "This step fails if no related species is found that matches the filters", + "output_name": "output_param_text", + "uuid": "d70ce5f3-03f8-4526-ae98-d9b2b1a53837" + } + ] + }, + "10": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "errors": null, + "id": 10, + "input_connections": { + "style_cond|type_cond|pick_from_0|value": { + "id": 8, + "output_name": "genome_fasta" + }, + "style_cond|type_cond|pick_from_1|value": { + "id": 4, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Pick parameter value", + "outputs": [ + { + "name": "data_param", + "type": "input" + } + ], + "position": { + "left": 2030.8500754379747, + "top": 491.6294119416157 + }, + "post_job_actions": { + "HideDatasetActiondata_param": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "data_param" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/pick_value/pick_value/0.2.0", + "tool_shed_repository": { + "changeset_revision": "b19e21af9c52", + "name": "pick_value", + "owner": "iuc", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"style_cond\": {\"pick_style\": \"first\", \"__current_case__\": 0, \"type_cond\": {\"param_type\": \"data\", \"__current_case__\": 4, \"pick_from\": [{\"__index__\": 0, \"value\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"value\": {\"__class__\": \"ConnectedValue\"}}]}}, \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "0.2.0", + "type": "tool", + "uuid": "5e8cd165-638e-4638-bfa3-5dbd8d9b17b6", + "when": null, + "workflow_outputs": [] + }, + "11": { + "annotation": "", + "content_id": "__FLATTEN__", + "errors": null, + "id": 11, + "input_connections": { + "input": { + "id": 10, + "output_name": "data_param" + }, + "when": { + "id": 7, + "output_name": "output_param_boolean" + } + }, + "inputs": [], + "label": null, + "name": "Flatten collection", + "outputs": [ + { + "name": "output", + "type": "input" + } + ], + "position": { + "left": 2347.3328957860285, + "top": 436.3033183719074 + }, + "post_job_actions": { + "RenameDatasetActionoutput": { + "action_arguments": { + "newname": "Related Species Genomes" + }, + "action_type": "RenameDatasetAction", + "output_name": "output" + } + }, + "tool_id": "__FLATTEN__", + "tool_state": "{\"input\": {\"__class__\": \"ConnectedValue\"}, \"join_identifier\": \"_\", \"__page__\": 0, \"__rerun_remap_job_id__\": null}", + "tool_uuid": null, + "tool_version": "1.0.0", + "type": "tool", + "uuid": "fd26c739-8e83-4a45-ba38-672fc89ed59b", + "when": "$(inputs.when)", + "workflow_outputs": [ + { + "label": "Related Species Genomes", + "output_name": "output", + "uuid": "f2323592-3250-4b05-993a-002a86109c26" + } + ] + } + }, + "tags": [], + "uuid": "07b6d2ff-c4cf-4705-8aca-bd9099200ef9" + }, + "tool_id": null, + "type": "subworkflow", + "uuid": "bfd278e3-a863-4f49-877c-86322905a823", + "when": "$(inputs.when)", + "workflow_outputs": [ + { + "label": "Related Species Sequence Reports", + "output_name": "Related Species Sequence Reports", + "uuid": "31ae1f96-ac5f-472a-bf21-e231ef6028c8" + }, + { + "label": "This step fails if no related species is found that matches the filters", + "output_name": "This step fails if no related species is found that matches the filters", + "uuid": "cfa734e0-d721-4de0-b28f-23cb06f5797e" + }, + { + "label": "Filtered Related Species Genome Report", + "output_name": "Filtered Related Species Genome Report", + "uuid": "d058f0bf-5a34-487c-8ed2-ab97043a09a5" + }, + { + "label": "Related Species Genomes", + "output_name": "Related Species Genomes", + "uuid": "3fa2fcb6-0b23-4d2b-a2ca-4026947c0e5e" + } + ] + } + }, + "tags": [ + "VGP", + "assembly", + "genome-assembly", + "ncbi-datasets", + "refseq-masher" + ], + "uuid": "0aeab89d-f347-4b09-aebb-4b10652ab3bf", + "version": 40, + "release": "0.1" +} diff --git a/workflows/VGP-assembly-v2/Fetch-Related-Genomes/README.md b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/README.md new file mode 100644 index 0000000000..3b8329481d --- /dev/null +++ b/workflows/VGP-assembly-v2/Fetch-Related-Genomes/README.md @@ -0,0 +1,61 @@ +# Fetch Related Genomes + +This workflow identifies and downloads genomes from species closely related to a reference assembly. It is intended as a preparation step for downstream curation workflows that need a panel of related-species genomes for synteny alignment (for example, the [JBrowse2 precuration tracks workflow](../Precuration-Jbrowse-tracks-alignments)). + +Given a reference assembly, the workflow: + +1. Searches RefSeq with [RefSeq Masher](https://github.com/phac-nml/refseq_masher) to find the most similar genomes. +2. Queries [NCBI Datasets](https://www.ncbi.nlm.nih.gov/datasets/) for assembly and BioSample metadata for each candidate. +3. Filters candidates by user-defined criteria (assembly level, assembly type, BioSample sex, sequencing technology, taxonomic class, chromosome count). +4. Downloads the genome fastas of the *n* closest matching related species. + +## Inputs + +### Required Inputs + +1. **Assembly** [fasta] - Reference assembly used to search for related genomes (typically your in-progress VGP assembly). +2. **Search assemblies among the n closest related species \*** [integer] - How many of the closest RefSeq matches to consider before filtering. +3. **Download n related species** [integer] - Final number of related-species fastas to download. Note: a large number of species can cause downstream mashmap alignment to run out of memory. +4. **Download sequences?** [boolean] - When `true`, downloads the genome fastas of the selected related species (the `Related Species Genomes` output collection). When `false`, only the filtered reports are produced - useful for exploring filter combinations before committing to a download. + +### Filter Parameters + +5. **Minimum Level Assembly** [text] - Minimum NCBI assembly level to accept. One of `Complete Genome`, `Chromosome`, `Scaffold`, `Contig`. Default: `Complete Genome`. +6. **Assembly Type** [text] - Required assembly type (e.g. `haploid`, `diploid`). Leave empty to keep everything. +7. **Assembly BioSample Sex** [text] - Required BioSample sex value (e.g. `female`, `male`). Leave empty to keep everything. +8. **Assembly Sequencing Technology** [text] - Substring that must appear in the NCBI Datasets sequencing-technology field (e.g. `PacBio`). Leave empty to keep everything. +9. **Assembly Taxonomic Class** [text] - Comma-separated taxonomic classes to keep (e.g. `Mammalia,Aves`). Leave empty to keep everything. +10. **Minimum Number Of Chromosomes** [integer] - Minimum number of chromosomes. Set to `0` to disable. +11. **Maximum Number Of Chromosomes** [integer] - Maximum number of chromosomes. Set to `0` to disable. +12. **Keep empty values when filtering on number of chromosomes** [boolean] - If `true`, assemblies with missing chromosome counts are kept when chromosome filters are applied. + +## Outputs + +1. **All Refseq Matches** [tabular] - Full RefSeq Masher match table for the reference assembly. +2. **Filtered Refseq Matches** [tabular] - RefSeq matches kept after applying the closest-*n* selection. +3. **All Related Species Genome Report** [tabular] - NCBI Datasets genome report for all candidate matches. +4. **Filtered Related Species Genome Report** [tabular] - Genome report after applying the assembly-quality filters. +5. **Filtered Genome Data** [tabular] - Filter-passing rows of the genome data report joined back with the RefSeq Masher columns (one row per kept species). +6. **Result Species** [tabular] - Final list of selected species (the top `Download n related species` rows that survived all filters). +7. **Related Species Sequence Reports** [tabular] - NCBI Datasets sequence-level reports for the kept genomes. +8. **Related Species Genomes** [collection of fasta] - Downloaded genome fastas of the closest related species, ready to feed into a synteny workflow. Only produced when `Download sequences?` is `true`. + +### Validation output + +9. **This step fails if no related species is found that matches the filters** [text] - Internal sentinel. The workflow fails fast at this step if every candidate is filtered out, so you get a clear error instead of empty downstream outputs. + +## Pairing with downstream workflows + +The **Related Species Genomes** output collection is designed to be used directly as the `Related Species` collection input of the [JBrowse2 precuration tracks workflow](../Precuration-Jbrowse-tracks-alignments). If sequence names collide across species, prefix scaffold names per species before downstream use: + +```bash +zcat Species_1.fasta.gz | sed 's/^>/>Species_1_/' | gzip > Species_1_renamed.fasta.gz +``` + +## Citation + +If you use this workflow, please cite the underlying tools: + +- RefSeq Masher +- NCBI Datasets +- Other tools as listed in the workflow