diff --git a/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/Assembly-decontamination-VGP9-tests.yml b/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/Assembly-decontamination-VGP9-tests.yml index e0f853eb5..3f710d1f5 100644 --- a/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/Assembly-decontamination-VGP9-tests.yml +++ b/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/Assembly-decontamination-VGP9-tests.yml @@ -2,16 +2,20 @@ job: Scaffolded assembly (fasta): class: File - location: https://zenodo.org/records/10067431/files/Scaffolded_assembly.fasta?download=1 + location: https://zenodo.org/records/13326708/files/Scaffolded_assembly.fasta?download=1 filetype: fasta - Database for Kraken2: viral2019-03 + Database for Kraken2: k2_pluspf_20210517 outputs: - full contaminant + mito scaffold list: - file: test-data/full contaminant + mito scaffold list.txt - compare: diff - sort: true + mito_scaff_names: + asserts: + has_text: + text: "SCAFFOLD_82" final decontaminated assembly: decompress: true asserts: not_has_text: - text: "_1100" + text: "scaffold_115" + contaminants_table: + asserts: + has_text: + text: "SCAFFOLD_101" diff --git a/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/Assembly-decontamination-VGP9.ga b/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/Assembly-decontamination-VGP9.ga index b83991631..becfb8caa 100644 --- a/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/Assembly-decontamination-VGP9.ga +++ b/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/Assembly-decontamination-VGP9.ga @@ -11,8 +11,11 @@ ], "format-version": "0.1", "license": "BSD-3-Clause", - "release": "0.1.6", + "release": "0.2", "name": "Assembly-decontamination-VGP9", + "report": { + "markdown": "\n# Workflow Execution Report\n\n### Worflow ran on: \n\n```galaxy\ninvocation_time()\n```\n\n# List of contaminants\n\n\r\n```galaxy\nhistory_dataset_as_table(output=\"contaminants_table\")\n```\r\n\n\n# List of Mitochondrial scaffolds\n\r\n```galaxy\nhistory_dataset_as_table(output=\"mito_scaff_names\")\n```\r\n\n\n\n\n## Workflow\n```galaxy\nworkflow_display()\n```\n" + }, "steps": { "0": { "annotation": "", @@ -31,7 +34,7 @@ "outputs": [], "position": { "left": 0, - "top": 431.99609375 + "top": 579.578125 }, "tool_id": null, "tool_state": "{\"optional\": false, \"format\": [\"fasta\"], \"tag\": null}", @@ -58,7 +61,7 @@ "outputs": [], "position": { "left": 72.984375, - "top": 545.19140625 + "top": 692.7734375 }, "tool_id": null, "tool_state": "{\"restrictOnConnections\": true, \"parameter_type\": \"text\", \"optional\": false}", @@ -70,7 +73,7 @@ { "label": null, "output_name": "output", - "uuid": "342dd6ab-f498-4a16-805f-3b7761497e91" + "uuid": "9b1c5517-60c3-49b5-9e9e-f85e883b6a5f" } ] }, @@ -95,8 +98,8 @@ } ], "position": { - "left": 342.8125, - "top": 227.6796875 + "left": 342.8124888681364, + "top": 375.78555810947694 }, "post_job_actions": { "ChangeDatatypeActionoutput": { @@ -105,6 +108,11 @@ }, "action_type": "ChangeDatatypeAction", "output_name": "output" + }, + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" } }, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sed_tool/9.3+galaxy1", @@ -143,9 +151,15 @@ ], "position": { "left": 652.953125, - "top": 176.06640625 + "top": 323.6484375 + }, + "post_job_actions": { + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } }, - "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_dustmasker_wrapper/2.14.1+galaxy2", "tool_shed_repository": { "changeset_revision": "cbf3f518b668", @@ -181,8 +195,8 @@ } ], "position": { - "left": 972.0078125, - "top": 207.47265625 + "left": 972.0000745777726, + "top": 355.5823870316441 }, "post_job_actions": { "ChangeDatatypeActionoutput": { @@ -191,6 +205,11 @@ }, "action_type": "ChangeDatatypeAction", "output_name": "output" + }, + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" } }, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sed_tool/9.3+galaxy1", @@ -241,7 +260,7 @@ ], "position": { "left": 1372.5625, - "top": 0 + "top": 147.58203125 }, "post_job_actions": { "HideDatasetActionunclassified_out_s": { @@ -292,7 +311,7 @@ ], "position": { "left": 1365.8984375, - "top": 392.03515625 + "top": 539.6171875 }, "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/2.14.1+galaxy2", @@ -311,9 +330,48 @@ }, "7": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_grep_tool/9.3+galaxy1", + "content_id": "Cut1", "errors": null, "id": 7, + "input_connections": { + "input": { + "id": 5, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Cut", + "outputs": [ + { + "name": "out_file1", + "type": "tabular" + } + ], + "position": { + "left": 1727.19921875, + "top": 53.078125 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Cut1", + "tool_state": "{\"columnList\": \"c1,c2,c3\", \"delimiter\": \"T\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.0.2", + "type": "tool", + "uuid": "bb39eb8b-cad8-44f1-aa56-64a11d985e3c", + "when": null, + "workflow_outputs": [] + }, + "8": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_grep_tool/9.3+galaxy1", + "errors": null, + "id": 8, "input_connections": { "infile": { "id": 5, @@ -331,9 +389,15 @@ ], "position": { "left": 1727.8671875, - "top": 57.37109375 + "top": 204.953125 + }, + "post_job_actions": { + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" + } }, - "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_grep_tool/9.3+galaxy1", "tool_shed_repository": { "changeset_revision": "fbf99087e067", @@ -348,11 +412,11 @@ "when": null, "workflow_outputs": [] }, - "8": { + "9": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/parse_mito_blast/parse_mito_blast/1.0.2+galaxy0", "errors": null, - "id": 8, + "id": 9, "input_connections": { "blastout": { "id": 6, @@ -374,7 +438,7 @@ ], "position": { "left": 1741, - "top": 369.26171875 + "top": 516.84375 }, "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/iuc/parse_mito_blast/parse_mito_blast/1.0.2+galaxy0", @@ -389,16 +453,74 @@ "type": "tool", "uuid": "4350e47a-2394-4381-a274-e238985195df", "when": null, - "workflow_outputs": [] + "workflow_outputs": [ + { + "label": "mito_scaff_names", + "output_name": "mito_scaff_names", + "uuid": "1478af1e-52f4-48fe-b966-0a61e2014a58" + } + ] }, - "9": { + "10": { + "annotation": "", + "content_id": "Filter1", + "errors": null, + "id": 10, + "input_connections": { + "input": { + "id": 7, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Filter", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 2074.40625, + "top": 0 + }, + "post_job_actions": { + "HideDatasetActionout_file1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "out_file1" + }, + "RenameDatasetActionout_file1": { + "action_arguments": { + "newname": "List of contaminants" + }, + "action_type": "RenameDatasetAction", + "output_name": "out_file1" + } + }, + "tool_id": "Filter1", + "tool_state": "{\"cond\": \"c1!='U'\", \"header_lines\": \"0\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1.1", + "type": "tool", + "uuid": "f5d80335-9b98-40ca-b706-4bb54d9801e6", + "when": null, + "workflow_outputs": [ + { + "label": "contaminants_table", + "output_name": "out_file1", + "uuid": "64156e28-aaf0-4578-9a7e-d6d21774c2c8" + } + ] + }, + "11": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sed_tool/9.3+galaxy1", "errors": null, - "id": 9, + "id": 11, "input_connections": { "infile": { - "id": 7, + "id": 8, "output_name": "output" } }, @@ -413,7 +535,7 @@ ], "position": { "left": 2061.3515625, - "top": 85.98828125 + "top": 233.5703125 }, "post_job_actions": { "ChangeDatatypeActionoutput": { @@ -422,6 +544,11 @@ }, "action_type": "ChangeDatatypeAction", "output_name": "output" + }, + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" } }, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sed_tool/9.3+galaxy1", @@ -438,14 +565,14 @@ "when": null, "workflow_outputs": [] }, - "10": { + "12": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sed_tool/9.3+galaxy1", "errors": null, - "id": 10, + "id": 12, "input_connections": { "infile": { - "id": 8, + "id": 9, "output_name": "mito_scaff_names" } }, @@ -460,9 +587,15 @@ ], "position": { "left": 2259.0390625, - "top": 370.75390625 + "top": 518.3359375 + }, + "post_job_actions": { + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" + } }, - "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sed_tool/9.3+galaxy1", "tool_shed_repository": { "changeset_revision": "fbf99087e067", @@ -477,14 +610,14 @@ "when": null, "workflow_outputs": [] }, - "11": { + "13": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_find_and_replace/9.3+galaxy1", "errors": null, - "id": 11, + "id": 13, "input_connections": { "infile": { - "id": 9, + "id": 11, "output_name": "output" } }, @@ -499,9 +632,15 @@ ], "position": { "left": 2321.61328125, - "top": 195.1875 + "top": 342.76953125 + }, + "post_job_actions": { + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } }, - "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_find_and_replace/9.3+galaxy1", "tool_shed_repository": { "changeset_revision": "fbf99087e067", @@ -509,26 +648,26 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"caseinsensitive\": \"false\", \"find_and_replace\": [], \"find_pattern\": \"kraken:taxid\\\\|[0-9]+\", \"global\": \"true\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"is_regex\": \"true\", \"replace_pattern\": \"\", \"searchwhere\": {\"__current_case__\": 0, \"searchwhere_select\": \"line\"}, \"skip_first_line\": \"false\", \"wholewords\": \"true\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"find_and_replace\": [{\"__index__\": 0, \"find_pattern\": \" kraken:taxid\\\\|[0-9]+\", \"replace_pattern\": null, \"is_regex\": true, \"global\": true, \"caseinsensitive\": false, \"wholewords\": true, \"skip_first_line\": false, \"searchwhere\": {\"searchwhere_select\": \"line\", \"__current_case__\": 0}}], \"infile\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "9.3+galaxy1", "type": "tool", "uuid": "c37db543-a966-4ed2-88b9-87b317465144", "when": null, "workflow_outputs": [] }, - "12": { + "14": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/9.3+galaxy1", "errors": null, - "id": 12, + "id": 14, "input_connections": { "inputs": [ { - "id": 10, + "id": 12, "output_name": "output" }, { - "id": 11, + "id": 13, "output_name": "outfile" } ] @@ -544,7 +683,7 @@ ], "position": { "left": 2611.0546875, - "top": 276.36328125 + "top": 423.9453125 }, "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/9.3+galaxy1", @@ -567,18 +706,18 @@ } ] }, - "13": { + "15": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/gfastats/gfastats/1.3.6+galaxy0", "errors": null, - "id": 13, + "id": 15, "input_connections": { "input_file": { "id": 0, "output_name": "output" }, "target_condition|exclude_bed": { - "id": 12, + "id": 14, "output_name": "out_file1" } }, @@ -593,7 +732,7 @@ ], "position": { "left": 3050.21875, - "top": 650.49609375 + "top": 798.078125 }, "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/gfastats/gfastats/1.3.6+galaxy0", @@ -620,6 +759,6 @@ "tags": [ "VGP_curated" ], - "uuid": "f6d8df21-f1dc-493c-bfab-e2cbecb8873d", - "version": 1 -} + "uuid": "3f6673a2-caf4-4156-b5da-cc6bde2ddc8d", + "version": 8 +} \ No newline at end of file diff --git a/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/CHANGELOG.md b/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/CHANGELOG.md index 3414f50e8..ee2feefeb 100644 --- a/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/CHANGELOG.md +++ b/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/CHANGELOG.md @@ -1,5 +1,9 @@ # Changelog +## [0.2] 2024-08-15 + +- Bug Fix: a missing regex caused the list of contaminants to empty out and the contaminants were not removed. + ## [0.1.6] 2024-06-10 ### Automatic update diff --git a/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/test-data/full contaminant + mito scaffold list.txt b/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/test-data/full contaminant + mito scaffold list.txt deleted file mode 100644 index 50433e8b3..000000000 --- a/workflows/VGP-assembly-v2/Assembly-decontamination-VGP9/test-data/full contaminant + mito scaffold list.txt +++ /dev/null @@ -1,2 +0,0 @@ -scaffold_1058 -scaffold_1100