From dfbee2a47cae5d89316f6c29f884a1af3bf96e08 Mon Sep 17 00:00:00 2001 From: Nolan Woods Date: Tue, 16 Aug 2022 14:06:32 -0700 Subject: [PATCH] Clean up Kristens workflows --- workflow/workflows/BLAST_Check.ga | 494 +-- workflow/workflows/Curated_GI_Detection.ga | 350 +- workflow/workflows/Detect_genomic_islands.ga | 212 +- workflow/workflows/Island_Check.ga | 2919 ++++++++--------- .../Salmonella_Curated_GI_Detection.ga | 232 +- 5 files changed, 2065 insertions(+), 2142 deletions(-) diff --git a/workflow/workflows/BLAST_Check.ga b/workflow/workflows/BLAST_Check.ga index e2ea3ba..4498307 100644 --- a/workflow/workflows/BLAST_Check.ga +++ b/workflow/workflows/BLAST_Check.ga @@ -13,27 +13,27 @@ "inputs": [ { "description": "", - "name": "Predicted GIs" + "name": "Fasta" } ], - "label": "Predicted GIs", + "label": "Fasta", "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 297, - "height": 63, - "left": -2154, - "right": -1954, - "top": 234, - "width": 200, - "x": -2154, - "y": 234 + "bottom": 425.9699014692164, + "height": 33.2833251953125, + "left": 429.95022446361935, + "right": 530.4502244636194, + "top": 392.6865762739039, + "width": 100.5, + "x": 429.95022446361935, + "y": 392.6865762739039 }, "tool_id": null, - "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", + "tool_state": "{\"optional\": false, \"tag\": \"\", \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "df998bc1-c517-438b-a3b9-2b1433409960", + "uuid": "2e74c941-fce6-431e-9aa4-825301ff8be1", "workflow_outputs": [] }, "1": { @@ -45,58 +45,67 @@ "inputs": [ { "description": "", - "name": "Fasta" + "name": "Predicted GIs" } ], - "label": "Fasta", + "label": "Predicted GIs", "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 483.140625, - "height": 63, - "left": -2145, - "right": -1945, - "top": 420.140625, - "width": 200, - "x": -2145, - "y": 420.140625 + "bottom": 623.3828307194495, + "height": 33.2833251953125, + "left": 423.9552056611474, + "right": 524.4552056611474, + "top": 590.099505524137, + "width": 100.5, + "x": 423.9552056611474, + "y": 590.099505524137 }, "tool_id": null, - "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", + "tool_state": "{\"optional\": false, \"tag\": \"\", \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "0cc3432d-180c-453e-98f4-bed53a1e164d", + "uuid": "2ff5d6fe-a5fc-4e41-8658-53c9df028c8a", "workflow_outputs": [] }, "2": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", + "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", "errors": null, "id": 2, "input_connections": { - "infiles_0|input_type_selection|infile": { + "input_file": { "id": 0, "output_name": "output" } }, - "inputs": [], - "label": "Remove Comments", - "name": "AWK Script", + "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", + "name": "input_file" + }, + { + "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", + "name": "mask_data_file" + } + ], + "label": "Genomes BLAST Database", + "name": "NCBI BLAST+ makeblastdb", "outputs": [ { "name": "outfile", - "type": "txt" + "type": "data" } ], "position": { - "bottom": 414.5, - "height": 137, - "left": -1882, - "right": -1682, - "top": 277.5, - "width": 200, - "x": -1882, - "y": 277.5 + "bottom": 597.9529819773204, + "height": 98.54998779296875, + "left": 1207.014943592584, + "right": 1307.514943592584, + "top": 499.40299418435166, + "width": 100.5, + "x": 1207.014943592584, + "y": 499.40299418435166 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -105,53 +114,48 @@ "output_name": "outfile" } }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", "tool_shed_repository": { - "changeset_revision": "df0c8f33c8ea", - "name": "awkscript", - "owner": "brinkmanlab", + "changeset_revision": "2889433c7ae1", + "name": "ncbi_blast_plus", + "owner": "devteam", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"$0 !~ /^#/ {print}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1", + "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"RuntimeValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "0.3.3", "type": "tool", - "uuid": "128d8539-0229-4a00-be44-4a4c3b5ad6aa", + "uuid": "1f450a2d-4ef2-47da-a17d-3a1016925d7c", "workflow_outputs": [] }, "3": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", + "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", "errors": null, "id": 3, "input_connections": { - "input_file": { + "infiles_0|input_type_selection|infile": { "id": 1, "output_name": "output" } }, - "inputs": [ - { - "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", - "name": "mask_data_file" - } - ], - "label": "Genomes_BLAST_Database", - "name": "NCBI BLAST+ makeblastdb", + "inputs": [], + "label": "Remove Comments", + "name": "AWK Script", "outputs": [ { "name": "outfile", - "type": "data" + "type": "txt" } ], "position": { - "bottom": 698, - "height": 189, - "left": -1368, - "right": -1168, - "top": 509, - "width": 200, - "x": -1368, - "y": 509 + "bottom": 648.3151559402693, + "height": 71.35000610351562, + "left": 631.7661570079291, + "right": 732.2661570079291, + "top": 576.9651498367537, + "width": 100.5, + "x": 631.7661570079291, + "y": 576.9651498367537 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -160,17 +164,17 @@ "output_name": "outfile" } }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", + "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", "tool_shed_repository": { - "changeset_revision": "2889433c7ae1", - "name": "ncbi_blast_plus", - "owner": "devteam", + "changeset_revision": "df0c8f33c8ea", + "name": "awkscript", + "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"ConnectedValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "0.3.3", + "tool_state": "{\"code\": \"$0 !~ /^#/ {print}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1", "type": "tool", - "uuid": "f2eeae19-7311-4e37-81ce-ca3eab907ed7", + "uuid": "a6f43e4d-1ff1-416d-b499-025bf6bf4efc", "workflow_outputs": [] }, "4": { @@ -180,11 +184,16 @@ "id": 4, "input_connections": { "inputs": { - "id": 2, + "id": 3, "output_name": "outfile" } }, - "inputs": [], + "inputs": [ + { + "description": "runtime parameter for tool Concatenate datasets", + "name": "inputs" + } + ], "label": null, "name": "Concatenate datasets", "outputs": [ @@ -194,14 +203,14 @@ } ], "position": { - "bottom": 258, - "height": 116, - "left": -1864, - "right": -1664, - "top": 142, - "width": 200, - "x": -1864, - "y": 142 + "bottom": 678.6621890850921, + "height": 61.100006103515625, + "left": 815.4726854011194, + "right": 915.9726854011194, + "top": 617.5621829815765, + "width": 100.5, + "x": 815.4726854011194, + "y": 617.5621829815765 }, "post_job_actions": { "ChangeDatatypeActionout_file1": { @@ -224,10 +233,10 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"inputs\": {\"__class__\": \"ConnectedValue\"}, \"queries\": [], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"inputs\": {\"__class__\": \"RuntimeValue\"}, \"queries\": [], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.1.0", "type": "tool", - "uuid": "9743034e-6142-44fb-9589-818673298fdb", + "uuid": "2708c62b-91ea-4b71-9f16-0fce9a9737f0", "workflow_outputs": [] }, "5": { @@ -241,7 +250,12 @@ "output_name": "out_file1" } }, - "inputs": [], + "inputs": [ + { + "description": "runtime parameter for tool Cut", + "name": "input" + } + ], "label": "To BED", "name": "Cut", "outputs": [ @@ -251,14 +265,14 @@ } ], "position": { - "bottom": 314.640625, - "height": 95, - "left": -1587.859375, - "right": -1387.859375, - "top": 219.640625, - "width": 200, - "x": -1587.859375, - "y": 219.640625 + "bottom": 710.2947538005772, + "height": 50.916656494140625, + "left": 997.5622376399253, + "right": 1098.0622376399253, + "top": 659.3780973064365, + "width": 100.5, + "x": 997.5622376399253, + "y": 659.3780973064365 }, "post_job_actions": { "ChangeDatatypeActionoutput": { @@ -288,10 +302,10 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"complement\": \"\", \"cut_type_options\": {\"cut_element\": \"-f\", \"__current_case__\": 0, \"list\": \"1,4,5\"}, \"delimiter\": \"\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"complement\": \"\", \"cut_type_options\": {\"cut_element\": \"-f\", \"__current_case__\": 0, \"list\": \"1\\n4\\n5\\n\"}, \"delimiter\": \"\", \"input\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1.0", "type": "tool", - "uuid": "430ecfe4-eba3-4d4b-9b48-91256419c5b4", + "uuid": "a99850a0-dace-407e-bbe0-0e9f28dc2a95", "workflow_outputs": [] }, "6": { @@ -301,7 +315,7 @@ "id": 6, "input_connections": { "fasta_source|fasta": { - "id": 1, + "id": 0, "output_name": "output" }, "input": { @@ -309,7 +323,16 @@ "output_name": "output" } }, - "inputs": [], + "inputs": [ + { + "description": "runtime parameter for tool bedtools GetFastaBed", + "name": "fasta_source" + }, + { + "description": "runtime parameter for tool bedtools GetFastaBed", + "name": "input" + } + ], "label": null, "name": "bedtools GetFastaBed", "outputs": [ @@ -319,14 +342,14 @@ } ], "position": { - "bottom": 417.140625, - "height": 147, - "left": -1334.5, - "right": -1134.5, - "top": 270.140625, - "width": 200, - "x": -1334.5, - "y": 270.140625 + "bottom": 434.1693998877682, + "height": 78.04998779296875, + "left": 1209.1294018190297, + "right": 1309.6294018190297, + "top": 356.11941209479943, + "width": 100.5, + "x": 1209.1294018190297, + "y": 356.11941209479943 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -349,10 +372,10 @@ "owner": "iuc", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"fasta_source\": {\"fasta_source_selector\": \"history\", \"__current_case__\": 0, \"fasta\": {\"__class__\": \"ConnectedValue\"}}, \"input\": {\"__class__\": \"ConnectedValue\"}, \"name\": \"false\", \"split\": \"false\", \"strand\": \"false\", \"tab\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"fasta_source\": {\"fasta_source_selector\": \"history\", \"__current_case__\": 0, \"fasta\": {\"__class__\": \"RuntimeValue\"}}, \"input\": {\"__class__\": \"RuntimeValue\"}, \"name\": \"false\", \"split\": \"false\", \"strand\": \"false\", \"tab\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "2.27.1", "type": "tool", - "uuid": "d1cb4922-9789-4979-9c1b-77effc532ce9", + "uuid": "51610527-c75b-46dd-b6b3-536afdc6ff09", "workflow_outputs": [] }, "7": { @@ -362,7 +385,7 @@ "id": 7, "input_connections": { "db_opts|histdb": { - "id": 3, + "id": 2, "output_name": "outfile" }, "query": { @@ -370,8 +393,17 @@ "output_name": "output" } }, - "inputs": [], - "label": "GI_blastn", + "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "db_opts" + }, + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "query" + } + ], + "label": "GI blastn", "name": "NCBI BLAST+ blastn", "outputs": [ { @@ -380,14 +412,14 @@ } ], "position": { - "bottom": 624, - "height": 189, - "left": -1045, - "right": -845, - "top": 435, - "width": 200, - "x": -1045, - "y": 435 + "bottom": 521.1290761463678, + "height": 98.61666870117188, + "left": 1415.0497208780316, + "right": 1515.5497208780316, + "top": 422.5124074451959, + "width": 100.5, + "x": 1415.0497208780316, + "y": 422.5124074451959 }, "post_job_actions": { "ChangeDatatypeActionoutput1": { @@ -410,10 +442,10 @@ "owner": "devteam", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"ConnectedValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"RuntimeValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.3.3", "type": "tool", - "uuid": "326c1904-2584-482f-a6ab-411ad5c83618", + "uuid": "b8110196-7807-4841-935d-6dbe31f02bef", "workflow_outputs": [] }, "8": { @@ -428,7 +460,7 @@ } }, "inputs": [], - "label": "BLAST_Hit_Filter", + "label": "BLAST Hit Filter", "name": "AWK Script", "outputs": [ { @@ -437,14 +469,14 @@ } ], "position": { - "bottom": 596, - "height": 137, - "left": -688, - "right": -488, - "top": 459, - "width": 200, - "x": -688, - "y": 459 + "bottom": 579.2512519039325, + "height": 61.16667175292969, + "left": 1604.900485366138, + "right": 1705.400485366138, + "top": 518.0845801510028, + "width": 100.5, + "x": 1604.900485366138, + "y": 518.0845801510028 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -467,10 +499,10 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "519adf18-77ea-4431-823a-3825f8b91315", + "uuid": "7c326117-af60-42cd-a9ed-8a528b05cdab", "workflow_outputs": [] }, "9": { @@ -484,7 +516,12 @@ "output_name": "outfile" } }, - "inputs": [], + "inputs": [ + { + "description": "runtime parameter for tool Filter empty datasets", + "name": "input" + } + ], "label": null, "name": "Filter empty datasets", "outputs": [ @@ -494,28 +531,28 @@ } ], "position": { - "bottom": 768.5, - "height": 137, - "left": -579, - "right": -379, - "top": 631.5, - "width": 200, - "x": -579, - "y": 631.5 - }, - "post_job_actions": {}, + "bottom": 614.1858452184877, + "height": 71.35000610351562, + "left": 1791.4926101912313, + "right": 1891.9926101912313, + "top": 542.835839114972, + "width": 100.5, + "x": 1791.4926101912313, + "y": 542.835839114972 + }, + "post_job_actions": { + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" + } + }, "tool_id": "__FILTER_EMPTY_DATASETS__", - "tool_state": "{\"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"input\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.0.0", "type": "tool", - "uuid": "2cfc0170-d2a0-4951-b2d9-ae3cbd34c16b", - "workflow_outputs": [ - { - "label": null, - "output_name": "output", - "uuid": "5a4a3708-2879-423a-84f0-aa756b7b60df" - } - ] + "uuid": "5f94f81f-3d8d-4f00-875e-8b4728d21d3c", + "workflow_outputs": [] }, "10": { "annotation": "", @@ -528,8 +565,13 @@ "output_name": "output" } }, - "inputs": [], - "label": "Sort_BLAST_Hits", + "inputs": [ + { + "description": "runtime parameter for tool Sort", + "name": "infile" + } + ], + "label": "Sort BLAST Hits", "name": "Sort", "outputs": [ { @@ -538,14 +580,14 @@ } ], "position": { - "bottom": 616.5, - "height": 116, - "left": -420, - "right": -220, - "top": 500.5, - "width": 200, - "x": -420, - "y": 500.5 + "bottom": 650.8669326554483, + "height": 50.91667175292969, + "left": 1981.840856751399, + "right": 2082.340856751399, + "top": 599.9502609025186, + "width": 100.5, + "x": 1981.840856751399, + "y": 599.9502609025186 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -561,10 +603,10 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1.1", "type": "tool", - "uuid": "fe51dedc-8b91-4c57-9d5d-eb063c9107f3", + "uuid": "de530266-f665-4458-aa0d-e2a4712a5959", "workflow_outputs": [] }, "11": { @@ -579,7 +621,7 @@ } }, "inputs": [], - "label": "Grouping_Hits", + "label": "Grouping Hits", "name": "AWK Script", "outputs": [ { @@ -588,14 +630,14 @@ } ], "position": { - "bottom": 529.5, - "height": 116, - "left": -178, - "right": 22, - "top": 413.5, - "width": 200, - "x": -178, - "y": 413.5 + "bottom": 686.0174089118616, + "height": 61.16667175292969, + "left": 2170.1244752798507, + "right": 2270.6244752798507, + "top": 624.8507371589319, + "width": 100.5, + "x": 2170.1244752798507, + "y": 624.8507371589319 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -618,10 +660,10 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"BEGIN {\\n getline first\\n split(first,last_line)\\n gap_dist=14000\\n min_length=8000\\n min_cov=0.95\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($2 == last_line[2]) && ($8 == last_line[8]) && (($6 < $7) ? ($6 <= last_line[7]+gap_dist) : ($7 <= last_line[6]+gap_dist))) {\\n if ($8 == \\\"+\\\") {\\n last_line[6] = (last_line[6] < $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] > $7 ? last_line[7] : $7)\\n } else {\\n last_line[6] = (last_line[6] > $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] < $7 ? last_line[7] : $7)\\n }\\n if ($4 < last_line[4]) {\\n last_line[3] += ($5 < last_line[4] ? $5 : last_line[4]-1) - $4 + 1\\n last_line[4] = $4\\n }\\n if ($5 > last_line[5]) {\\n last_line[3] += $5 - ($4 > last_line[5] ? $4 : last_line[5]+1) + 1\\n last_line[5] = $5\\n }\\n if (last_line[3]<$3) {\\n last_line[3] = $3\\n }\\n } else {\\n if ((last_line[3] >= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"blast\\\", \\\"genomic_island\\\", (last_line[6]= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"blast\\\", \\\"genomic_island\\\", (last_line[6] $7 ? last_line[7] : $7)\\n } else {\\n last_line[6] = (last_line[6] > $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] < $7 ? last_line[7] : $7)\\n }\\n if ($4 < last_line[4]) {\\n last_line[3] += ($5 < last_line[4] ? $5 : last_line[4]-1) - $4 + 1\\n last_line[4] = $4\\n }\\n if ($5 > last_line[5]) {\\n last_line[3] += $5 - ($4 > last_line[5] ? $4 : last_line[5]+1) + 1\\n last_line[5] = $5\\n }\\n if (last_line[3]<$3) {\\n last_line[3] = $3\\n }\\n } else {\\n if ((last_line[3] >= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"blast\\\", \\\"genomic_island\\\", (last_line[6]= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"blast\\\", \\\"genomic_island\\\", (last_line[6]= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"BEGIN { \\n getline first\\n split(first,last_line)\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($4 <= last_line[5]) && ($5 >= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "63ddf586-e47c-4bf9-a455-f24fbb6f3d3b", + "uuid": "c8bfa0cc-708a-4377-bbf7-d4cd94c3b787", "workflow_outputs": [ { - "label": null, + "label": "BLAST Check", "output_name": "outfile", - "uuid": "0a48ea42-4d76-4fa0-b47a-b2b1308ca890" + "uuid": "f0e2fc33-efac-4699-a9a2-d4632dcf9d1a" } ] } }, "tags": [], - "uuid": "0640d563-c0b6-4f14-af57-6b28e16365f7", - "version": 3 + "uuid": "b8a5c0cd-c86e-4d09-8dfb-78d574822549", + "version": 4 } diff --git a/workflow/workflows/Curated_GI_Detection.ga b/workflow/workflows/Curated_GI_Detection.ga index 17fe134..95e75aa 100644 --- a/workflow/workflows/Curated_GI_Detection.ga +++ b/workflow/workflows/Curated_GI_Detection.ga @@ -20,20 +20,20 @@ "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 445.140625, - "height": 84, - "left": -479, - "right": -279, - "top": 361.140625, - "width": 200, - "x": -479, - "y": 361.140625 + "bottom": 279.03334045410156, + "height": 64.90000915527344, + "left": 138.5, + "right": 288.5, + "top": 214.13333129882812, + "width": 150, + "x": 138.5, + "y": 214.13333129882812 }, "tool_id": null, "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "befa6feb-e83b-4f6f-9c41-2e4cbaa36e2b", + "uuid": "e885de9f-3182-40dd-8191-abf3de0e4894", "workflow_outputs": [] }, "1": { @@ -52,20 +52,20 @@ "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 580.671875, - "height": 84, - "left": -479, - "right": -279, - "top": 496.671875, - "width": 200, - "x": -479, - "y": 496.671875 + "bottom": 414.5666809082031, + "height": 64.89999389648438, + "left": 138.5, + "right": 288.5, + "top": 349.66668701171875, + "width": 150, + "x": 138.5, + "y": 349.66668701171875 }, "tool_id": null, "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "cc9145dc-567f-45e4-87ae-afd2e38699fe", + "uuid": "47361dc5-bb94-40cd-96ea-d7f8847a1380", "workflow_outputs": [] }, "2": { @@ -89,14 +89,14 @@ } ], "position": { - "bottom": 900.6875, - "height": 305, - "left": -206.984375, - "right": -6.984375, - "top": 595.6875, - "width": 200, - "x": -206.984375, - "y": 595.6875 + "bottom": 578.5833435058594, + "height": 228.89999389648438, + "left": 399.51666259765625, + "right": 549.5166625976562, + "top": 349.683349609375, + "width": 150, + "x": 399.51666259765625, + "y": 349.683349609375 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -115,7 +115,7 @@ "tool_state": "{\"bioperl\": \"false\", \"info\": \"false\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"output_type\": \"text\", \"query\": \"[0].features[? type == `\\\"source\\\"`].qualifiers.organism[0]\", \"split\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "1a54b58b-9ad6-4c3f-a4a9-cbd80203c433", + "uuid": "865b8ca1-6d91-469e-b2ed-ba11b8bdc4c3", "workflow_outputs": [] }, "3": { @@ -139,14 +139,14 @@ } ], "position": { - "bottom": 557.15625, - "height": 137, - "left": 89.015625, - "right": 289.015625, - "top": 420.15625, - "width": 200, - "x": 89.015625, - "y": 420.15625 + "bottom": 572.6499938964844, + "height": 106.5, + "left": 639.5166625976562, + "right": 789.5166625976562, + "top": 466.1499938964844, + "width": 150, + "x": 639.5166625976562, + "y": 466.1499938964844 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -165,7 +165,7 @@ "tool_state": "{\"code\": \"tolower($0) ~ \\\"^\\\" ENVIRON[\\\"target\\\"] { print tool_input_id }\", \"envs\": [{\"__index__\": 0, \"name\": \"target\", \"val\": \"salmonella enterica\"}], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"true\", \"__current_case__\": 1, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "c006dcb2-9b58-4107-8479-3a673e0795a0", + "uuid": "3d130f19-b95c-42dd-b816-d063a3b858a7", "workflow_outputs": [] }, "4": { @@ -197,14 +197,14 @@ } ], "position": { - "bottom": 458.171875, - "height": 220, - "left": 327.03125, - "right": 527.03125, - "top": 238.171875, - "width": 200, - "x": 327.03125, - "y": 238.171875 + "bottom": 402.6499938964844, + "height": 172.4833221435547, + "left": 880.5333251953125, + "right": 1030.5333251953125, + "top": 230.1666717529297, + "width": 150, + "x": 880.5333251953125, + "y": 230.1666717529297 }, "post_job_actions": { "HideDatasetActionoutput_discarded": { @@ -222,7 +222,7 @@ "tool_state": "{\"how\": {\"how_filter\": \"remove_if_absent\", \"__current_case__\": 0, \"filter_source\": {\"__class__\": \"ConnectedValue\"}}, \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.0.0", "type": "tool", - "uuid": "828b69c0-5968-4e81-a328-62ab3d450cc6", + "uuid": "2c209d0c-ded3-4480-8e62-c05d71eeb429", "workflow_outputs": [] }, "5": { @@ -240,14 +240,14 @@ "name": "Salmonella Curated GI Detection", "outputs": [], "position": { - "bottom": 413.5, - "height": 137, - "left": 619, - "right": 819, - "top": 276.5, - "width": 200, - "x": 619, - "y": 276.5 + "bottom": 418.43333435058594, + "height": 121.80000305175781, + "left": 1127.5, + "right": 1277.5, + "top": 296.6333312988281, + "width": 150, + "x": 1127.5, + "y": 296.6333312988281 }, "subworkflow": { "a_galaxy_workflow": "true", @@ -262,7 +262,7 @@ "id": 0, "input_connections": {}, "inputs": [], - "label": null, + "label": "Get Curated Salmonella GI Database", "name": "get fasta reference", "outputs": [ { @@ -271,14 +271,14 @@ } ], "position": { - "bottom": 347.5, - "height": 84, - "left": -862, - "right": -662, - "top": 263.5, - "width": 200, - "x": -862, - "y": 263.5 + "bottom": 255.6999969482422, + "height": 80.19999694824219, + "left": 197, + "right": 347, + "top": 175.5, + "width": 150, + "x": 197, + "y": 175.5 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -297,7 +297,7 @@ "tool_state": "{\"pre_installed_fasta\": \"salmonella_gis\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.3.1", "type": "tool", - "uuid": "41c9a13a-229b-4fe1-9cae-07e7b1766a28", + "uuid": "a40a5caf-7f8a-4ddb-a05f-07878cd935a0", "workflow_outputs": [] }, "1": { @@ -316,20 +316,20 @@ "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 494.5, - "height": 84, - "left": -866, - "right": -666, - "top": 410.5, - "width": 200, - "x": -866, - "y": 410.5 + "bottom": 408.6999969482422, + "height": 80.19999694824219, + "left": 189, + "right": 339, + "top": 328.5, + "width": 150, + "x": 189, + "y": 328.5 }, "tool_id": null, "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "86d770e4-8963-4b20-b3cf-8fb77fcc9ce6", + "uuid": "35d0f860-f980-4134-89b2-dd58e6076fd9", "workflow_outputs": [] }, "2": { @@ -344,12 +344,16 @@ } }, "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", + "name": "input_file" + }, { "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", "name": "mask_data_file" } ], - "label": "Genomes_BLAST_Database", + "label": "Build Salmonella BLAST Database", "name": "NCBI BLAST+ makeblastdb", "outputs": [ { @@ -358,14 +362,14 @@ } ], "position": { - "bottom": 600.5, - "height": 189, - "left": -590, - "right": -390, - "top": 411.5, - "width": 200, - "x": -590, - "y": 411.5 + "bottom": 476.6000061035156, + "height": 147.10000610351562, + "left": 465, + "right": 615, + "top": 329.5, + "width": 150, + "x": 465, + "y": 329.5 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -381,10 +385,10 @@ "owner": "devteam", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"ConnectedValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"RuntimeValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.3.3", "type": "tool", - "uuid": "c358e05b-f1a3-4682-9480-7959dbc98e7b", + "uuid": "77439f1c-9ec1-4b0d-84ab-f8618197a0df", "workflow_outputs": [] }, "3": { @@ -402,8 +406,17 @@ "output_name": "output" } }, - "inputs": [], - "label": "GI_blastn", + "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "db_opts" + }, + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "query" + } + ], + "label": "BLASTn Salmonella GI DB", "name": "NCBI BLAST+ blastn", "outputs": [ { @@ -412,14 +425,14 @@ } ], "position": { - "bottom": 523.5, - "height": 189, - "left": -307, - "right": -107, - "top": 334.5, - "width": 200, - "x": -307, - "y": 334.5 + "bottom": 417.8999938964844, + "height": 162.39999389648438, + "left": 749, + "right": 899, + "top": 255.5, + "width": 150, + "x": 749, + "y": 255.5 }, "post_job_actions": { "ChangeDatatypeActionoutput1": { @@ -442,10 +455,10 @@ "owner": "devteam", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"ConnectedValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"RuntimeValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.3.3", "type": "tool", - "uuid": "822ca610-b1b1-492a-b57f-085279f3d1fa", + "uuid": "57e2a49b-27d5-4f8c-8c1e-ed5d0d9dc9e9", "workflow_outputs": [] }, "4": { @@ -460,7 +473,7 @@ } }, "inputs": [], - "label": "BLAST_Hit_Filter", + "label": "BLAST Hit Filter", "name": "AWK Script", "outputs": [ { @@ -469,14 +482,14 @@ } ], "position": { - "bottom": 500.5, - "height": 137, - "left": -4, - "right": 196, - "top": 363.5, - "width": 200, - "x": -4, - "y": 363.5 + "bottom": 359.78334045410156, + "height": 91.28334045410156, + "left": 968, + "right": 1118, + "top": 268.5, + "width": 150, + "x": 968, + "y": 268.5 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -499,10 +512,10 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "dd567ca6-6f19-408e-8167-a7eb34f7d085", + "uuid": "26d4ec13-b9df-418a-8693-786be2b57f70", "workflow_outputs": [] }, "5": { @@ -516,8 +529,13 @@ "output_name": "outfile" } }, - "inputs": [], - "label": "Sort_BLAST_Hits", + "inputs": [ + { + "description": "runtime parameter for tool Sort", + "name": "infile" + } + ], + "label": "Sort BLAST Hits", "name": "Sort", "outputs": [ { @@ -526,14 +544,14 @@ } ], "position": { - "bottom": 491, - "height": 116, - "left": 310, - "right": 510, - "top": 375, - "width": 200, - "x": 310, - "y": 375 + "bottom": 352.98333740234375, + "height": 75.98333740234375, + "left": 1185, + "right": 1335, + "top": 277, + "width": 150, + "x": 1185, + "y": 277 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -549,10 +567,10 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1.1", "type": "tool", - "uuid": "f176af4f-afba-421c-a649-ad4c8dc1783f", + "uuid": "35286835-6249-4ca2-a076-a23c4ae30d04", "workflow_outputs": [] }, "6": { @@ -567,7 +585,7 @@ } }, "inputs": [], - "label": "Grouping_Hits", + "label": "Grouping Hits", "name": "AWK Script", "outputs": [ { @@ -576,14 +594,14 @@ } ], "position": { - "bottom": 488, - "height": 116, - "left": 624, - "right": 824, - "top": 372, - "width": 200, - "x": 624, - "y": 372 + "bottom": 367.28334045410156, + "height": 91.28334045410156, + "left": 1393, + "right": 1543, + "top": 276, + "width": 150, + "x": 1393, + "y": 276 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -609,7 +627,7 @@ "tool_state": "{\"code\": \"BEGIN {\\n getline first\\n split(first,last_line)\\n gap_dist=14000\\n min_length=8000\\n min_cov=0.95\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($2 == last_line[2]) && ($8 == last_line[8]) && (($6 < $7) ? ($6 <= last_line[7]+gap_dist) : ($7 <= last_line[6]+gap_dist))) {\\n if ($8 == \\\"+\\\") {\\n last_line[6] = (last_line[6] < $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] > $7 ? last_line[7] : $7)\\n } else {\\n last_line[6] = (last_line[6] > $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] < $7 ? last_line[7] : $7)\\n }\\n if ($4 < last_line[4]) {\\n last_line[3] += ($5 < last_line[4] ? $5 : last_line[4]-1) - $4 + 1\\n last_line[4] = $4\\n }\\n if ($5 > last_line[5]) {\\n last_line[3] += $5 - ($4 > last_line[5] ? $4 : last_line[5]+1) + 1\\n last_line[5] = $5\\n }\\n if (last_line[3]<$3) {\\n last_line[3] = $3\\n }\\n } else {\\n if ((last_line[3] >= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"curated\\\", \\\"genomic_island\\\", (last_line[6]= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"curated\\\", \\\"genomic_island\\\", (last_line[6]= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"BEGIN { \\n getline first\\n split(first,last_line)\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($4 <= last_line[5]) && ($5 >= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "295c92c0-50fe-437a-9e1d-66ed6ce6507c", + "uuid": "afd56aab-2c50-4fdd-94a2-b2f1ebecaf05", "workflow_outputs": [ { - "label": null, + "label": "Curated Salmonella Islands", "output_name": "outfile", - "uuid": "b4086374-fc71-4248-9c56-988a5697945b" + "uuid": "2ea65090-747b-4b26-96b4-acfb098bfb8a" } ] } }, "tags": "", - "uuid": "c6320762-c3ac-483a-a879-c1af20c2af43" + "uuid": "6beb3a66-9e2c-44d7-86ff-90a84ad754ca" }, - "tool_id": "9c96f7d71c15b79e", + "tool_id": "af038e9879627e7a", "type": "subworkflow", - "uuid": "f087dbf2-c474-4b4c-9a4a-943a43bd42c7", + "uuid": "a556da9f-bac2-433e-b4d3-d38f8b43057a", "workflow_outputs": [ { "label": null, - "output_name": "8:outfile", - "uuid": "d92dd995-f484-4043-a8d5-35903f496045" + "output_name": "Curated Salmonella Islands", + "uuid": "1b7a8463-f187-44b3-a923-72b25342811a" } ] } }, "tags": [], - "uuid": "01f56b28-d564-4b77-bd37-ef48726e9d1f", - "version": 25 + "uuid": "437c6491-9287-4dc2-9284-d9fae9fdd260", + "version": 3 } diff --git a/workflow/workflows/Detect_genomic_islands.ga b/workflow/workflows/Detect_genomic_islands.ga index 3729c04..d2f31c3 100644 --- a/workflow/workflows/Detect_genomic_islands.ga +++ b/workflow/workflows/Detect_genomic_islands.ga @@ -20,17 +20,17 @@ "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 288.8500061035156, + "bottom": 250.85000610351562, "height": 49.68333435058594, - "left": -203, - "right": -53, - "top": 239.1666717529297, + "left": -1191.5, + "right": -1041.5, + "top": 201.1666717529297, "width": 150, - "x": -203, - "y": 239.1666717529297 + "x": -1191.5, + "y": 201.1666717529297 }, "tool_id": null, - "tool_state": "{\"optional\": false, \"collection_type\": \"list\"}", + "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", "uuid": "38c240f7-3fc8-48ab-a01c-74390e0261f8", @@ -61,14 +61,14 @@ } ], "position": { - "bottom": 512.7499847412109, + "bottom": 474.74998474121094, "height": 269.58331298828125, - "left": 140, - "right": 290, - "top": 243.1666717529297, + "left": -848.5, + "right": -698.5, + "top": 205.1666717529297, "width": 150, - "x": 140, - "y": 243.1666717529297 + "x": -848.5, + "y": 205.1666717529297 }, "post_job_actions": { "HideDatasetActioninfo_output": { @@ -116,14 +116,14 @@ } ], "position": { - "bottom": 354.15000915527344, + "bottom": 316.1499938964844, "height": 75.98332214355469, - "left": 490.5, - "right": 640.5, - "top": 278.16668701171875, + "left": -498, + "right": -348, + "top": 240.1666717529297, "width": 150, - "x": 490.5, - "y": 278.16668701171875 + "x": -498, + "y": 240.1666717529297 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -170,14 +170,14 @@ } ], "position": { - "bottom": 792.5500183105469, + "bottom": 754.5500183105469, "height": 141.88333129882812, - "left": 216.1666717529297, - "right": 366.1666717529297, - "top": 650.6666870117188, + "left": -772.3333129882812, + "right": -622.3333129882812, + "top": 612.6666870117188, "width": 150, - "x": 216.1666717529297, - "y": 650.6666870117188 + "x": -772.3333129882812, + "y": 612.6666870117188 }, "post_job_actions": { "HideDatasetActiongff": { @@ -229,14 +229,14 @@ } ], "position": { - "bottom": 593.0666809082031, + "bottom": 555.0666809082031, "height": 162.39999389648438, - "left": 720.1666870117188, - "right": 870.1666870117188, - "top": 430.66668701171875, + "left": -268.33331298828125, + "right": -118.33331298828125, + "top": 392.66668701171875, "width": 150, - "x": 720.1666870117188, - "y": 430.66668701171875 + "x": -268.33331298828125, + "y": 392.66668701171875 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -297,14 +297,14 @@ } ], "position": { - "bottom": 825.7666778564453, + "bottom": 787.7666778564453, "height": 147.09999084472656, - "left": 576.1666870117188, - "right": 726.1666870117188, - "top": 678.6666870117188, + "left": -412.33331298828125, + "right": -262.33331298828125, + "top": 640.6666870117188, "width": 150, - "x": 576.1666870117188, - "y": 678.6666870117188 + "x": -412.33331298828125, + "y": 640.6666870117188 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -334,7 +334,7 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"# Replace all sequence IDs in second tool input with first sequence ID in first tool input\\nBEGIN { FS=OFS=\\\"\\\\t\\\"; }\\ntool_input == 0 && /^[^#]/ { id = $1; nextfile }\\ntool_input == 1 && /^[^#]/ { $1 = id }\\ntool_input == 1 { print }\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}, {\"__index__\": 1, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"# Replace all sequence IDs in second tool input with first sequence ID in first tool input\\nBEGIN { FS=OFS=\\\"\\\\t\\\"; }\\ntool_input == 0 && /^[^#]/ { id = $1; nextfile }\\ntool_input == 1 && /^[^#]/ { $1 = id; $9 = \\\"ID=\\\" tool_input_id \\\"_gi\\\" FNR; }\\ntool_input == 1 { print }\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}, {\"__index__\": 1, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", "uuid": "3ee9e770-3ef5-4e82-a1a3-191ce7a6689a", @@ -353,7 +353,7 @@ }, "inputs": [], "label": null, - "name": "Filter failed", + "name": "Filter failed datasets", "outputs": [ { "name": "output", @@ -361,14 +361,14 @@ } ], "position": { - "bottom": 436.28334045410156, - "height": 91.28334045410156, - "left": 1107, - "right": 1257, - "top": 345, + "bottom": 413.5, + "height": 106.5, + "left": 118.5, + "right": 268.5, + "top": 307, "width": 150, - "x": 1107, - "y": 345 + "x": 118.5, + "y": 307 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -405,14 +405,14 @@ } ], "position": { - "bottom": 800.6666870117188, + "bottom": 762.6666870117188, "height": 106.5, - "left": 873.1666870117188, - "right": 1023.1666870117188, - "top": 694.1666870117188, + "left": -115.33332824707031, + "right": 34.66667175292969, + "top": 656.1666870117188, "width": 150, - "x": 873.1666870117188, - "y": 694.1666870117188 + "x": -115.33332824707031, + "y": 656.1666870117188 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -461,7 +461,7 @@ }, "inputs": [], "label": null, - "name": "Filter failed", + "name": "Filter failed datasets", "outputs": [ { "name": "output", @@ -469,14 +469,14 @@ } ], "position": { - "bottom": 789.9500122070312, - "height": 91.2833251953125, - "left": 1159.1666259765625, - "right": 1309.1666259765625, - "top": 698.6666870117188, + "bottom": 767.1666870117188, + "height": 106.5, + "left": 170.6666717529297, + "right": 320.6666717529297, + "top": 660.6666870117188, "width": 150, - "x": 1159.1666259765625, - "y": 698.6666870117188 + "x": 170.6666717529297, + "y": 660.6666870117188 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -509,7 +509,7 @@ }, "inputs": [], "label": "Aggregate island calls", - "name": "Merge Collections", + "name": "Merge collections", "outputs": [ { "name": "output", @@ -517,14 +517,14 @@ } ], "position": { - "bottom": 429.5666809082031, + "bottom": 391.56666564941406, "height": 162.39999389648438, - "left": 1516, - "right": 1666, - "top": 267.16668701171875, + "left": 527.5, + "right": 677.5, + "top": 229.1666717529297, "width": 150, - "x": 1516, - "y": 267.16668701171875 + "x": 527.5, + "y": 229.1666717529297 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -561,14 +561,14 @@ } ], "position": { - "bottom": 398.95001220703125, + "bottom": 360.95001220703125, "height": 91.2833251953125, - "left": 1835.1666259765625, - "right": 1985.1666259765625, - "top": 307.66668701171875, + "left": 846.6666870117188, + "right": 996.6666870117188, + "top": 269.66668701171875, "width": 150, - "x": 1835.1666259765625, - "y": 307.66668701171875 + "x": 846.6666870117188, + "y": 269.66668701171875 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -593,7 +593,7 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"#!/usr/bin/env gawk -f\\n# Filter islands based on size specified in 'minimum_size' environment variable\\nBEGIN { FS=OFS=\\\"\\\\t\\\"; }\\n/^#/ || ($5-$4 >= ENVIRON[\\\"minimum_size\\\"]) { print }\", \"envs\": [{\"__index__\": 0, \"name\": \"minimum_size\", \"val\": \"8000\"}], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"#!/usr/bin/env gawk -f\\n# Filter islands based on size specified in 'minimum_size' environment variable\\nBEGIN { FS=OFS=\\\"\\\\t\\\"; }\\n/^#/ || ($5-$4 >= ENVIRON[\\\"minimum_size\\\"]) { print }\", \"envs\": [{\"__index__\": 0, \"name\": \"minimum_size\", \"val\": \"8000\"}], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", "uuid": "4cfdad5f-ccab-4a0e-84b8-9d75651a6a4d", @@ -601,70 +601,12 @@ { "label": "Unmerged Islands", "output_name": "outfile", - "uuid": "cb85093d-b0a0-4965-9808-67e37b3586aa" - } - ] - }, - "11": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/feature_merge/feature-merge/1.3", - "errors": null, - "id": 11, - "input_connections": { - "inputs": { - "id": 10, - "output_name": "outfile" - } - }, - "inputs": [], - "label": "Merge Islands", - "name": "GFF/GTF Feature Merge", - "outputs": [ - { - "name": "output", - "type": "gff" - } - ], - "position": { - "bottom": 395.45001220703125, - "height": 91.2833251953125, - "left": 2132.5, - "right": 2282.5, - "top": 304.16668701171875, - "width": 150, - "x": 2132.5, - "y": 304.16668701171875 - }, - "post_job_actions": { - "RenameDatasetActionoutput": { - "action_arguments": { - "newname": "Predicted Genomic Islands" - }, - "action_type": "RenameDatasetAction", - "output_name": "output" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/feature_merge/feature-merge/1.3", - "tool_shed_repository": { - "changeset_revision": "cc4bffe1d9b5", - "name": "feature_merge", - "owner": "brinkmanlab", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"exact_only\": \"false\", \"exclude_components\": \"true\", \"features\": [], \"ignore_strand\": \"true\", \"inputs\": {\"__class__\": \"ConnectedValue\"}, \"merge_strategy\": \"append\", \"threshold\": \"500\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.3", - "type": "tool", - "uuid": "364fa9e4-ba42-453a-a1fb-035b09c1893f", - "workflow_outputs": [ - { - "label": "Predicted Genomic Islands", - "output_name": "output", - "uuid": "80accc8a-2295-4fbe-a77b-a9f5044479ef" + "uuid": "c95002d7-e650-45d9-8c18-92990b8bc92a" } ] } }, "tags": [], - "uuid": "6e00813e-1749-467e-94b7-b7221dcf1470", - "version": 3 + "uuid": "79a72918-57f0-4639-b2c3-a050a80bb486", + "version": 6 } diff --git a/workflow/workflows/Island_Check.ga b/workflow/workflows/Island_Check.ga index 7a6322a..bf8e13a 100644 --- a/workflow/workflows/Island_Check.ga +++ b/workflow/workflows/Island_Check.ga @@ -2,7 +2,7 @@ "a_galaxy_workflow": "true", "annotation": "", "format-version": "0.1", - "name": "Island Check (imported from uploaded file)", + "name": "Island Check", "steps": { "0": { "annotation": "", @@ -13,27 +13,27 @@ "inputs": [ { "description": "", - "name": "Stitched genomes GBK" + "name": "Additional annotations" } ], - "label": "Stitched genomes GBK", + "label": "Additional annotations", "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 502.96875, - "height": 42, - "left": 140.4375, - "right": 240.4375, - "top": 460.96875, - "width": 100, - "x": 140.4375, - "y": 460.96875 + "bottom": 157.46665954589844, + "height": 64.89999389648438, + "left": -398.4666748046875, + "right": -248.4666748046875, + "top": 92.56666564941406, + "width": 150, + "x": -398.4666748046875, + "y": 92.56666564941406 }, "tool_id": null, "tool_state": "{\"optional\": false, \"tag\": \"\", \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "b4d62a94-a99d-43b0-a708-fd08adc1fb6d", + "uuid": "ee71d98f-be7a-4e92-89dd-a203a42b13c0", "workflow_outputs": [] }, "1": { @@ -45,27 +45,27 @@ "inputs": [ { "description": "", - "name": "Detected Islands" + "name": "Predicted Islands" } ], - "label": "Detected Islands", + "label": "Predicted Islands", "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 652.03125, - "height": 31.5, - "left": 135.1875, - "right": 235.1875, - "top": 620.53125, - "width": 100, - "x": 135.1875, - "y": 620.53125 + "bottom": 287.21665954589844, + "height": 49.683319091796875, + "left": -399.5, + "right": -249.5, + "top": 237.53334045410156, + "width": 150, + "x": -399.5, + "y": 237.53334045410156 }, "tool_id": null, "tool_state": "{\"optional\": false, \"tag\": \"\", \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "956f8be6-45da-4bd4-b810-d2bee4bfe86d", + "uuid": "1fb2255e-49f4-4435-8e6f-47d41c492131", "workflow_outputs": [] }, "2": { @@ -77,27 +77,27 @@ "inputs": [ { "description": "", - "name": "Additional annotations" + "name": "FASTA" } ], - "label": "Additional annotations", + "label": "FASTA", "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 773.5625, - "height": 42, - "left": 133.21875, - "right": 233.21875, - "top": 731.5625, - "width": 100, - "x": 133.21875, - "y": 731.5625 + "bottom": 516.6500091552734, + "height": 49.68333435058594, + "left": -396.5333251953125, + "right": -246.5333251953125, + "top": 466.9666748046875, + "width": 150, + "x": -396.5333251953125, + "y": 466.9666748046875 }, "tool_id": null, "tool_state": "{\"optional\": false, \"tag\": \"\", \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "61ee0aa3-6fb0-412a-afce-f2e912ca3439", + "uuid": "13709d0a-4f5b-413d-a408-9fb2532c61d1", "workflow_outputs": [] }, "3": { @@ -109,125 +109,63 @@ "inputs": [ { "description": "", - "name": "FASTA" + "name": "Stitched genomes GBK" } ], - "label": "FASTA", + "label": "Stitched genomes GBK", "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 912.46875, - "height": 31.5, - "left": 144.15625, - "right": 244.15625, - "top": 880.96875, - "width": 100, - "x": 144.15625, - "y": 880.96875 + "bottom": 656.8666839599609, + "height": 64.90000915527344, + "left": -394.25, + "right": -244.25, + "top": 591.9666748046875, + "width": 150, + "x": -394.25, + "y": 591.9666748046875 }, "tool_id": null, "tool_state": "{\"optional\": false, \"tag\": \"\", \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "c81200e5-65e9-49f0-934e-4de5937213bd", + "uuid": "710fff66-644b-401a-bfa6-bd71a5bde5e6", "workflow_outputs": [] }, "4": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/1.1.1", - "errors": null, "id": 4, "input_connections": { - "infile": { - "id": 1, - "output_name": "output" - } - }, - "inputs": [ - { - "description": "runtime parameter for tool Text reformatting", - "name": "infile" - } - ], - "label": null, - "name": "Text reformatting", - "outputs": [ - { - "name": "outfile", - "type": "input" - } - ], - "position": { - "bottom": 651.671875, - "height": 58, - "left": 390.171875, - "right": 490.171875, - "top": 593.671875, - "width": 100, - "x": 390.171875, - "y": 593.671875 - }, - "post_job_actions": { - "ChangeDatatypeActionoutfile": { - "action_arguments": { - "newtype": "gff3" - }, - "action_type": "ChangeDatatypeAction", - "output_name": "outfile" - }, - "HideDatasetActionoutfile": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/1.1.1", - "tool_shed_repository": { - "changeset_revision": "a6f147a050a2", - "name": "text_processing", - "owner": "bgruening", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"code\": \"$3==\\\"genomic_island\\\" { \\nif ($9 == \\\"\\\") $9 = \\\".\\\";\\nprint\\n}\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1.1", - "type": "tool", - "uuid": "ca65ccd8-2494-48b5-ba9b-a6977d4f278f", - "workflow_outputs": [] - }, - "5": { - "annotation": "", - "id": 5, - "input_connections": { - "FASTA Formatted Genomes": { - "id": 3, + "Fasta": { + "id": 2, "input_subworkflow_step_id": 0, "output_name": "output" }, - "Genbank Formatted Genomes": { - "id": 0, + "Predicted GIs": { + "id": 1, "input_subworkflow_step_id": 1, "output_name": "output" } }, "inputs": [], "label": null, - "name": "Curated GI Detection", + "name": "BLAST Check", "outputs": [], "position": { - "bottom": 492.171875, - "height": 84, - "left": 541.171875, - "right": 641.171875, - "top": 408.171875, - "width": 100, - "x": 541.171875, - "y": 408.171875 + "bottom": 448.78334045410156, + "height": 101.28334045410156, + "left": -178, + "right": -28, + "top": 347.5, + "width": 150, + "x": -178, + "y": 347.5 }, "subworkflow": { "a_galaxy_workflow": "true", "annotation": "", "format-version": "0.1", - "name": "Curated GI Detection", + "name": "BLAST Check", "steps": { "0": { "annotation": "", @@ -238,27 +176,27 @@ "inputs": [ { "description": "", - "name": "FASTA Formatted Genomes" + "name": "Fasta" } ], - "label": "FASTA Formatted Genomes", + "label": "Fasta", "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 445.140625, - "height": 84, - "left": -479, - "right": -279, - "top": 361.140625, - "width": 200, - "x": -479, - "y": 361.140625 + "bottom": 425.9699014692164, + "height": 33.2833251953125, + "left": 429.95022446361935, + "right": 530.4502244636194, + "top": 392.6865762739039, + "width": 100.5, + "x": 429.95022446361935, + "y": 392.6865762739039 }, "tool_id": null, - "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", + "tool_state": "{\"optional\": false, \"tag\": \"\", \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "befa6feb-e83b-4f6f-9c41-2e4cbaa36e2b", + "uuid": "2e74c941-fce6-431e-9aa4-825301ff8be1", "workflow_outputs": [] }, "1": { @@ -270,77 +208,86 @@ "inputs": [ { "description": "", - "name": "Genbank Formatted Genomes" + "name": "Predicted GIs" } ], - "label": "Genbank Formatted Genomes", + "label": "Predicted GIs", "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 580.671875, - "height": 84, - "left": -479, - "right": -279, - "top": 496.671875, - "width": 200, - "x": -479, - "y": 496.671875 + "bottom": 623.3828307194495, + "height": 33.2833251953125, + "left": 423.9552056611474, + "right": 524.4552056611474, + "top": 590.099505524137, + "width": 100.5, + "x": 423.9552056611474, + "y": 590.099505524137 }, "tool_id": null, - "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", + "tool_state": "{\"optional\": false, \"tag\": \"\", \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "cc9145dc-567f-45e4-87ae-afd2e38699fe", + "uuid": "2ff5d6fe-a5fc-4e41-8658-53c9df028c8a", "workflow_outputs": [] }, "2": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/biopython_convert/biopython-convert/1.1", + "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", "errors": null, "id": 2, "input_connections": { - "input": { - "id": 1, + "input_file": { + "id": 0, "output_name": "output" } }, - "inputs": [], - "label": "Extract organism name", - "name": "BioPython SeqIO Converter", + "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", + "name": "input_file" + }, + { + "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", + "name": "mask_data_file" + } + ], + "label": "Genomes BLAST Database", + "name": "NCBI BLAST+ makeblastdb", "outputs": [ { - "name": "output", - "type": "txt" + "name": "outfile", + "type": "data" } ], "position": { - "bottom": 900.6875, - "height": 305, - "left": -206.984375, - "right": -6.984375, - "top": 595.6875, - "width": 200, - "x": -206.984375, - "y": 595.6875 + "bottom": 597.9529819773204, + "height": 98.54998779296875, + "left": 1207.014943592584, + "right": 1307.514943592584, + "top": 499.40299418435166, + "width": 100.5, + "x": 1207.014943592584, + "y": 499.40299418435166 }, "post_job_actions": { - "HideDatasetActionoutput": { + "HideDatasetActionoutfile": { "action_arguments": {}, "action_type": "HideDatasetAction", - "output_name": "output" + "output_name": "outfile" } }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/biopython_convert/biopython-convert/1.1", + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", "tool_shed_repository": { - "changeset_revision": "1c71c770bfd4", - "name": "biopython_convert", - "owner": "brinkmanlab", + "changeset_revision": "2889433c7ae1", + "name": "ncbi_blast_plus", + "owner": "devteam", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"bioperl\": \"false\", \"info\": \"false\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"output_type\": \"text\", \"query\": \"[0].features[? type == `\\\"source\\\"`].qualifiers.organism[0]\", \"split\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1", + "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"RuntimeValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "0.3.3", "type": "tool", - "uuid": "1a54b58b-9ad6-4c3f-a4a9-cbd80203c433", + "uuid": "1f450a2d-4ef2-47da-a17d-3a1016925d7c", "workflow_outputs": [] }, "3": { @@ -350,12 +297,12 @@ "id": 3, "input_connections": { "infiles_0|input_type_selection|infile": { - "id": 2, + "id": 1, "output_name": "output" } }, "inputs": [], - "label": "List Salmonella Files", + "label": "Remove Comments", "name": "AWK Script", "outputs": [ { @@ -364,14 +311,14 @@ } ], "position": { - "bottom": 557.15625, - "height": 137, - "left": 89.015625, - "right": 289.015625, - "top": 420.15625, - "width": 200, - "x": 89.015625, - "y": 420.15625 + "bottom": 648.3151559402693, + "height": 71.35000610351562, + "left": 631.7661570079291, + "right": 732.2661570079291, + "top": 576.9651498367537, + "width": 100.5, + "x": 631.7661570079291, + "y": 576.9651498367537 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -387,889 +334,113 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"tolower($0) ~ \\\"^\\\" ENVIRON[\\\"target\\\"] { print tool_input_id }\", \"envs\": [{\"__index__\": 0, \"name\": \"target\", \"val\": \"salmonella enterica\"}], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"true\", \"__current_case__\": 1, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"$0 !~ /^#/ {print}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "c006dcb2-9b58-4107-8479-3a673e0795a0", + "uuid": "a6f43e4d-1ff1-416d-b499-025bf6bf4efc", "workflow_outputs": [] }, "4": { "annotation": "", - "content_id": "__FILTER_FROM_FILE__", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", "errors": null, "id": 4, "input_connections": { - "how|filter_source": { + "inputs": { "id": 3, "output_name": "outfile" - }, - "input": { - "id": 0, - "output_name": "output" } }, - "inputs": [], - "label": "Filter Salmonella", - "name": "Filter collection", - "outputs": [ + "inputs": [ { - "name": "output_filtered", - "type": "input" - }, + "description": "runtime parameter for tool Concatenate datasets", + "name": "inputs" + } + ], + "label": null, + "name": "Concatenate datasets", + "outputs": [ { - "name": "output_discarded", + "name": "out_file1", "type": "input" } ], "position": { - "bottom": 458.171875, - "height": 220, - "left": 327.03125, - "right": 527.03125, - "top": 238.171875, - "width": 200, - "x": 327.03125, - "y": 238.171875 + "bottom": 678.6621890850921, + "height": 61.100006103515625, + "left": 815.4726854011194, + "right": 915.9726854011194, + "top": 617.5621829815765, + "width": 100.5, + "x": 815.4726854011194, + "y": 617.5621829815765 }, "post_job_actions": { - "HideDatasetActionoutput_discarded": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "output_discarded" + "ChangeDatatypeActionout_file1": { + "action_arguments": { + "newtype": "gff" + }, + "action_type": "ChangeDatatypeAction", + "output_name": "out_file1" }, - "HideDatasetActionoutput_filtered": { + "HideDatasetActionout_file1": { "action_arguments": {}, "action_type": "HideDatasetAction", - "output_name": "output_filtered" + "output_name": "out_file1" } }, - "tool_id": "__FILTER_FROM_FILE__", - "tool_state": "{\"how\": {\"how_filter\": \"remove_if_absent\", \"__current_case__\": 0, \"filter_source\": {\"__class__\": \"ConnectedValue\"}}, \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.0.0", + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", + "tool_shed_repository": { + "changeset_revision": "a6f147a050a2", + "name": "text_processing", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"inputs\": {\"__class__\": \"RuntimeValue\"}, \"queries\": [], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "0.1.0", "type": "tool", - "uuid": "828b69c0-5968-4e81-a328-62ab3d450cc6", + "uuid": "2708c62b-91ea-4b71-9f16-0fce9a9737f0", "workflow_outputs": [] }, "5": { "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cut_tool/1.1.0", + "errors": null, "id": 5, "input_connections": { - "FASTA Salmonella Genomes": { + "input": { "id": 4, - "input_subworkflow_step_id": 1, - "output_name": "output_filtered" + "output_name": "out_file1" } }, - "inputs": [], - "label": null, - "name": "Salmonella Curated GI Detection", - "outputs": [], + "inputs": [ + { + "description": "runtime parameter for tool Cut", + "name": "input" + } + ], + "label": "To BED", + "name": "Cut", + "outputs": [ + { + "name": "output", + "type": "tabular" + } + ], "position": { - "bottom": 413.5, - "height": 137, - "left": 619, - "right": 819, - "top": 276.5, - "width": 200, - "x": 619, - "y": 276.5 + "bottom": 710.2947538005772, + "height": 50.916656494140625, + "left": 997.5622376399253, + "right": 1098.0622376399253, + "top": 659.3780973064365, + "width": 100.5, + "x": 997.5622376399253, + "y": 659.3780973064365 }, - "subworkflow": { - "a_galaxy_workflow": "true", - "annotation": "", - "format-version": "0.1", - "name": "Salmonella Curated GI Detection", - "steps": { - "0": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/artbio/get_reference_fasta/get_fasta_reference/0.3.1", - "errors": null, - "id": 0, - "input_connections": {}, - "inputs": [], - "label": null, - "name": "get fasta reference", - "outputs": [ - { - "name": "output", - "type": "fasta" - } - ], - "position": { - "bottom": 347.5, - "height": 84, - "left": -862, - "right": -662, - "top": 263.5, - "width": 200, - "x": -862, - "y": 263.5 - }, - "post_job_actions": { - "HideDatasetActionoutput": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "output" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/artbio/get_reference_fasta/get_fasta_reference/0.3.1", - "tool_shed_repository": { - "changeset_revision": "f3c9cbd84728", - "name": "get_reference_fasta", - "owner": "artbio", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"pre_installed_fasta\": \"salmonella_gis\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "0.3.1", - "type": "tool", - "uuid": "41c9a13a-229b-4fe1-9cae-07e7b1766a28", - "workflow_outputs": [] - }, - "1": { - "annotation": "", - "content_id": null, - "errors": null, - "id": 1, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "FASTA Salmonella Genomes" - } - ], - "label": "FASTA Salmonella Genomes", - "name": "Input dataset collection", - "outputs": [], - "position": { - "bottom": 494.5, - "height": 84, - "left": -866, - "right": -666, - "top": 410.5, - "width": 200, - "x": -866, - "y": 410.5 - }, - "tool_id": null, - "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", - "tool_version": null, - "type": "data_collection_input", - "uuid": "86d770e4-8963-4b20-b3cf-8fb77fcc9ce6", - "workflow_outputs": [] - }, - "2": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", - "errors": null, - "id": 2, - "input_connections": { - "input_file": { - "id": 1, - "output_name": "output" - } - }, - "inputs": [ - { - "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", - "name": "mask_data_file" - } - ], - "label": "Genomes_BLAST_Database", - "name": "NCBI BLAST+ makeblastdb", - "outputs": [ - { - "name": "outfile", - "type": "data" - } - ], - "position": { - "bottom": 600.5, - "height": 189, - "left": -590, - "right": -390, - "top": 411.5, - "width": 200, - "x": -590, - "y": 411.5 - }, - "post_job_actions": { - "HideDatasetActionoutfile": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", - "tool_shed_repository": { - "changeset_revision": "2889433c7ae1", - "name": "ncbi_blast_plus", - "owner": "devteam", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"ConnectedValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "0.3.3", - "type": "tool", - "uuid": "c358e05b-f1a3-4682-9480-7959dbc98e7b", - "workflow_outputs": [] - }, - "3": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.3.3", - "errors": null, - "id": 3, - "input_connections": { - "db_opts|histdb": { - "id": 2, - "output_name": "outfile" - }, - "query": { - "id": 0, - "output_name": "output" - } - }, - "inputs": [], - "label": "GI_blastn", - "name": "NCBI BLAST+ blastn", - "outputs": [ - { - "name": "output1", - "type": "tabular" - } - ], - "position": { - "bottom": 523.5, - "height": 189, - "left": -307, - "right": -107, - "top": 334.5, - "width": 200, - "x": -307, - "y": 334.5 - }, - "post_job_actions": { - "ChangeDatatypeActionoutput1": { - "action_arguments": { - "newtype": "tabular" - }, - "action_type": "ChangeDatatypeAction", - "output_name": "output1" - }, - "HideDatasetActionoutput1": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "output1" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.3.3", - "tool_shed_repository": { - "changeset_revision": "2889433c7ae1", - "name": "ncbi_blast_plus", - "owner": "devteam", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"ConnectedValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "0.3.3", - "type": "tool", - "uuid": "822ca610-b1b1-492a-b57f-085279f3d1fa", - "workflow_outputs": [] - }, - "4": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", - "errors": null, - "id": 4, - "input_connections": { - "infiles_0|input_type_selection|infile": { - "id": 3, - "output_name": "output1" - } - }, - "inputs": [], - "label": "BLAST_Hit_Filter", - "name": "AWK Script", - "outputs": [ - { - "name": "outfile", - "type": "txt" - } - ], - "position": { - "bottom": 500.5, - "height": 137, - "left": -4, - "right": 196, - "top": 363.5, - "width": 200, - "x": -4, - "y": 363.5 - }, - "post_job_actions": { - "ChangeDatatypeActionoutfile": { - "action_arguments": { - "newtype": "tabular" - }, - "action_type": "ChangeDatatypeAction", - "output_name": "outfile" - }, - "HideDatasetActionoutfile": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", - "tool_shed_repository": { - "changeset_revision": "df0c8f33c8ea", - "name": "awkscript", - "owner": "brinkmanlab", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1", - "type": "tool", - "uuid": "dd567ca6-6f19-408e-8167-a7eb34f7d085", - "workflow_outputs": [] - }, - "5": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sort_header_tool/1.1.1", - "errors": null, - "id": 5, - "input_connections": { - "infile": { - "id": 4, - "output_name": "outfile" - } - }, - "inputs": [], - "label": "Sort_BLAST_Hits", - "name": "Sort", - "outputs": [ - { - "name": "outfile", - "type": "input" - } - ], - "position": { - "bottom": 491, - "height": 116, - "left": 310, - "right": 510, - "top": 375, - "width": 200, - "x": 310, - "y": 375 - }, - "post_job_actions": { - "HideDatasetActionoutfile": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sort_header_tool/1.1.1", - "tool_shed_repository": { - "changeset_revision": "a6f147a050a2", - "name": "text_processing", - "owner": "bgruening", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1.1", - "type": "tool", - "uuid": "f176af4f-afba-421c-a649-ad4c8dc1783f", - "workflow_outputs": [] - }, - "6": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", - "errors": null, - "id": 6, - "input_connections": { - "infiles_0|input_type_selection|infile": { - "id": 5, - "output_name": "outfile" - } - }, - "inputs": [], - "label": "Grouping_Hits", - "name": "AWK Script", - "outputs": [ - { - "name": "outfile", - "type": "txt" - } - ], - "position": { - "bottom": 488, - "height": 116, - "left": 624, - "right": 824, - "top": 372, - "width": 200, - "x": 624, - "y": 372 - }, - "post_job_actions": { - "ChangeDatatypeActionoutfile": { - "action_arguments": { - "newtype": "gff" - }, - "action_type": "ChangeDatatypeAction", - "output_name": "outfile" - }, - "HideDatasetActionoutfile": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", - "tool_shed_repository": { - "changeset_revision": "df0c8f33c8ea", - "name": "awkscript", - "owner": "brinkmanlab", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"code\": \"BEGIN {\\n getline first\\n split(first,last_line)\\n gap_dist=14000\\n min_length=8000\\n min_cov=0.95\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($2 == last_line[2]) && ($8 == last_line[8]) && (($6 < $7) ? ($6 <= last_line[7]+gap_dist) : ($7 <= last_line[6]+gap_dist))) {\\n if ($8 == \\\"+\\\") {\\n last_line[6] = (last_line[6] < $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] > $7 ? last_line[7] : $7)\\n } else {\\n last_line[6] = (last_line[6] > $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] < $7 ? last_line[7] : $7)\\n }\\n if ($4 < last_line[4]) {\\n last_line[3] += ($5 < last_line[4] ? $5 : last_line[4]-1) - $4 + 1\\n last_line[4] = $4\\n }\\n if ($5 > last_line[5]) {\\n last_line[3] += $5 - ($4 > last_line[5] ? $4 : last_line[5]+1) + 1\\n last_line[5] = $5\\n }\\n if (last_line[3]<$3) {\\n last_line[3] = $3\\n }\\n } else {\\n if ((last_line[3] >= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"curated\\\", \\\"genomic_island\\\", (last_line[6]= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"curated\\\", \\\"genomic_island\\\", (last_line[6]= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1", - "type": "tool", - "uuid": "295c92c0-50fe-437a-9e1d-66ed6ce6507c", - "workflow_outputs": [ - { - "label": null, - "output_name": "outfile", - "uuid": "b4086374-fc71-4248-9c56-988a5697945b" - } - ] - } - }, - "tags": "", - "uuid": "c6320762-c3ac-483a-a879-c1af20c2af43" - }, - "tool_id": "9c96f7d71c15b79e", - "type": "subworkflow", - "uuid": "f087dbf2-c474-4b4c-9a4a-943a43bd42c7", - "workflow_outputs": [ - { - "label": null, - "output_name": "8:outfile", - "uuid": "d92dd995-f484-4043-a8d5-35903f496045" - } - ] - } - }, - "tags": "", - "uuid": "01f56b28-d564-4b77-bd37-ef48726e9d1f" - }, - "tool_id": "46a6f5d3fe38c112", - "type": "subworkflow", - "uuid": "3374ba25-2d91-4d35-afb1-40e0018c0da1", - "workflow_outputs": [ - { - "label": null, - "output_name": "5:8:outfile", - "uuid": "a068f06d-ee09-42ed-a28f-6e96641a5191" - } - ] - }, - "6": { - "annotation": "", - "id": 6, - "input_connections": { - "Fasta": { - "id": 3, - "input_subworkflow_step_id": 1, - "output_name": "output" - }, - "Predicted GIs": { - "id": 1, - "input_subworkflow_step_id": 0, - "output_name": "output" - } - }, - "inputs": [], - "label": null, - "name": "BLAST Check", - "outputs": [], - "position": { - "bottom": 977.171875, - "height": 89, - "left": 446.21875, - "right": 546.21875, - "top": 888.171875, - "width": 100, - "x": 446.21875, - "y": 888.171875 - }, - "subworkflow": { - "a_galaxy_workflow": "true", - "annotation": "", - "format-version": "0.1", - "name": "BLAST Check", - "steps": { - "0": { - "annotation": "", - "content_id": null, - "errors": null, - "id": 0, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "Predicted GIs" - } - ], - "label": "Predicted GIs", - "name": "Input dataset collection", - "outputs": [], - "position": { - "bottom": 297, - "height": 63, - "left": -2154, - "right": -1954, - "top": 234, - "width": 200, - "x": -2154, - "y": 234 - }, - "tool_id": null, - "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", - "tool_version": null, - "type": "data_collection_input", - "uuid": "df998bc1-c517-438b-a3b9-2b1433409960", - "workflow_outputs": [] - }, - "1": { - "annotation": "", - "content_id": null, - "errors": null, - "id": 1, - "input_connections": {}, - "inputs": [ - { - "description": "", - "name": "Fasta" - } - ], - "label": "Fasta", - "name": "Input dataset collection", - "outputs": [], - "position": { - "bottom": 483.140625, - "height": 63, - "left": -2145, - "right": -1945, - "top": 420.140625, - "width": 200, - "x": -2145, - "y": 420.140625 - }, - "tool_id": null, - "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", - "tool_version": null, - "type": "data_collection_input", - "uuid": "0cc3432d-180c-453e-98f4-bed53a1e164d", - "workflow_outputs": [] - }, - "2": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", - "errors": null, - "id": 2, - "input_connections": { - "infiles_0|input_type_selection|infile": { - "id": 0, - "output_name": "output" - } - }, - "inputs": [], - "label": "Remove Comments", - "name": "AWK Script", - "outputs": [ - { - "name": "outfile", - "type": "txt" - } - ], - "position": { - "bottom": 414.5, - "height": 137, - "left": -1882, - "right": -1682, - "top": 277.5, - "width": 200, - "x": -1882, - "y": 277.5 - }, - "post_job_actions": { - "HideDatasetActionoutfile": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", - "tool_shed_repository": { - "changeset_revision": "df0c8f33c8ea", - "name": "awkscript", - "owner": "brinkmanlab", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"code\": \"$0 !~ /^#/ {print}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1", - "type": "tool", - "uuid": "128d8539-0229-4a00-be44-4a4c3b5ad6aa", - "workflow_outputs": [] - }, - "3": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", - "errors": null, - "id": 3, - "input_connections": { - "input_file": { - "id": 1, - "output_name": "output" - } - }, - "inputs": [ - { - "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", - "name": "mask_data_file" - } - ], - "label": "Genomes_BLAST_Database", - "name": "NCBI BLAST+ makeblastdb", - "outputs": [ - { - "name": "outfile", - "type": "data" - } - ], - "position": { - "bottom": 698, - "height": 189, - "left": -1368, - "right": -1168, - "top": 509, - "width": 200, - "x": -1368, - "y": 509 - }, - "post_job_actions": { - "HideDatasetActionoutfile": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", - "tool_shed_repository": { - "changeset_revision": "2889433c7ae1", - "name": "ncbi_blast_plus", - "owner": "devteam", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"ConnectedValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "0.3.3", - "type": "tool", - "uuid": "f2eeae19-7311-4e37-81ce-ca3eab907ed7", - "workflow_outputs": [] - }, - "4": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", - "errors": null, - "id": 4, - "input_connections": { - "inputs": { - "id": 2, - "output_name": "outfile" - } - }, - "inputs": [], - "label": null, - "name": "Concatenate datasets", - "outputs": [ - { - "name": "out_file1", - "type": "input" - } - ], - "position": { - "bottom": 258, - "height": 116, - "left": -1864, - "right": -1664, - "top": 142, - "width": 200, - "x": -1864, - "y": 142 - }, - "post_job_actions": { - "ChangeDatatypeActionout_file1": { - "action_arguments": { - "newtype": "gff" - }, - "action_type": "ChangeDatatypeAction", - "output_name": "out_file1" - }, - "HideDatasetActionout_file1": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "out_file1" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", - "tool_shed_repository": { - "changeset_revision": "a6f147a050a2", - "name": "text_processing", - "owner": "bgruening", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"inputs\": {\"__class__\": \"ConnectedValue\"}, \"queries\": [], \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "0.1.0", - "type": "tool", - "uuid": "9743034e-6142-44fb-9589-818673298fdb", - "workflow_outputs": [] - }, - "5": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cut_tool/1.1.0", - "errors": null, - "id": 5, - "input_connections": { - "input": { - "id": 4, - "output_name": "out_file1" - } - }, - "inputs": [], - "label": "To BED", - "name": "Cut", - "outputs": [ - { - "name": "output", - "type": "tabular" - } - ], - "position": { - "bottom": 314.640625, - "height": 95, - "left": -1587.859375, - "right": -1387.859375, - "top": 219.640625, - "width": 200, - "x": -1587.859375, - "y": 219.640625 - }, - "post_job_actions": { - "ChangeDatatypeActionoutput": { - "action_arguments": { - "newtype": "bed" + "post_job_actions": { + "ChangeDatatypeActionoutput": { + "action_arguments": { + "newtype": "bed" }, "action_type": "ChangeDatatypeAction", "output_name": "output" @@ -1294,10 +465,10 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"complement\": \"\", \"cut_type_options\": {\"cut_element\": \"-f\", \"__current_case__\": 0, \"list\": \"1,4,5\"}, \"delimiter\": \"\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"complement\": \"\", \"cut_type_options\": {\"cut_element\": \"-f\", \"__current_case__\": 0, \"list\": \"1\\n4\\n5\\n\"}, \"delimiter\": \"\", \"input\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1.0", "type": "tool", - "uuid": "430ecfe4-eba3-4d4b-9b48-91256419c5b4", + "uuid": "a99850a0-dace-407e-bbe0-0e9f28dc2a95", "workflow_outputs": [] }, "6": { @@ -1307,7 +478,7 @@ "id": 6, "input_connections": { "fasta_source|fasta": { - "id": 1, + "id": 0, "output_name": "output" }, "input": { @@ -1315,7 +486,16 @@ "output_name": "output" } }, - "inputs": [], + "inputs": [ + { + "description": "runtime parameter for tool bedtools GetFastaBed", + "name": "fasta_source" + }, + { + "description": "runtime parameter for tool bedtools GetFastaBed", + "name": "input" + } + ], "label": null, "name": "bedtools GetFastaBed", "outputs": [ @@ -1325,14 +505,14 @@ } ], "position": { - "bottom": 417.140625, - "height": 147, - "left": -1334.5, - "right": -1134.5, - "top": 270.140625, - "width": 200, - "x": -1334.5, - "y": 270.140625 + "bottom": 434.1693998877682, + "height": 78.04998779296875, + "left": 1209.1294018190297, + "right": 1309.6294018190297, + "top": 356.11941209479943, + "width": 100.5, + "x": 1209.1294018190297, + "y": 356.11941209479943 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -1355,10 +535,10 @@ "owner": "iuc", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"fasta_source\": {\"fasta_source_selector\": \"history\", \"__current_case__\": 0, \"fasta\": {\"__class__\": \"ConnectedValue\"}}, \"input\": {\"__class__\": \"ConnectedValue\"}, \"name\": \"false\", \"split\": \"false\", \"strand\": \"false\", \"tab\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"fasta_source\": {\"fasta_source_selector\": \"history\", \"__current_case__\": 0, \"fasta\": {\"__class__\": \"RuntimeValue\"}}, \"input\": {\"__class__\": \"RuntimeValue\"}, \"name\": \"false\", \"split\": \"false\", \"strand\": \"false\", \"tab\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "2.27.1", "type": "tool", - "uuid": "d1cb4922-9789-4979-9c1b-77effc532ce9", + "uuid": "51610527-c75b-46dd-b6b3-536afdc6ff09", "workflow_outputs": [] }, "7": { @@ -1368,7 +548,7 @@ "id": 7, "input_connections": { "db_opts|histdb": { - "id": 3, + "id": 2, "output_name": "outfile" }, "query": { @@ -1376,8 +556,17 @@ "output_name": "output" } }, - "inputs": [], - "label": "GI_blastn", + "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "db_opts" + }, + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "query" + } + ], + "label": "GI blastn", "name": "NCBI BLAST+ blastn", "outputs": [ { @@ -1386,14 +575,14 @@ } ], "position": { - "bottom": 624, - "height": 189, - "left": -1045, - "right": -845, - "top": 435, - "width": 200, - "x": -1045, - "y": 435 + "bottom": 521.1290761463678, + "height": 98.61666870117188, + "left": 1415.0497208780316, + "right": 1515.5497208780316, + "top": 422.5124074451959, + "width": 100.5, + "x": 1415.0497208780316, + "y": 422.5124074451959 }, "post_job_actions": { "ChangeDatatypeActionoutput1": { @@ -1416,10 +605,10 @@ "owner": "devteam", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"ConnectedValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"RuntimeValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.3.3", "type": "tool", - "uuid": "326c1904-2584-482f-a6ab-411ad5c83618", + "uuid": "b8110196-7807-4841-935d-6dbe31f02bef", "workflow_outputs": [] }, "8": { @@ -1434,7 +623,7 @@ } }, "inputs": [], - "label": "BLAST_Hit_Filter", + "label": "BLAST Hit Filter", "name": "AWK Script", "outputs": [ { @@ -1443,14 +632,14 @@ } ], "position": { - "bottom": 596, - "height": 137, - "left": -688, - "right": -488, - "top": 459, - "width": 200, - "x": -688, - "y": 459 + "bottom": 579.2512519039325, + "height": 61.16667175292969, + "left": 1604.900485366138, + "right": 1705.400485366138, + "top": 518.0845801510028, + "width": 100.5, + "x": 1604.900485366138, + "y": 518.0845801510028 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -1473,10 +662,10 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "519adf18-77ea-4431-823a-3825f8b91315", + "uuid": "7c326117-af60-42cd-a9ed-8a528b05cdab", "workflow_outputs": [] }, "9": { @@ -1490,7 +679,12 @@ "output_name": "outfile" } }, - "inputs": [], + "inputs": [ + { + "description": "runtime parameter for tool Filter empty datasets", + "name": "input" + } + ], "label": null, "name": "Filter empty datasets", "outputs": [ @@ -1500,28 +694,28 @@ } ], "position": { - "bottom": 768.5, - "height": 137, - "left": -579, - "right": -379, - "top": 631.5, - "width": 200, - "x": -579, - "y": 631.5 - }, - "post_job_actions": {}, + "bottom": 614.1858452184877, + "height": 71.35000610351562, + "left": 1791.4926101912313, + "right": 1891.9926101912313, + "top": 542.835839114972, + "width": 100.5, + "x": 1791.4926101912313, + "y": 542.835839114972 + }, + "post_job_actions": { + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" + } + }, "tool_id": "__FILTER_EMPTY_DATASETS__", - "tool_state": "{\"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"input\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.0.0", "type": "tool", - "uuid": "2cfc0170-d2a0-4951-b2d9-ae3cbd34c16b", - "workflow_outputs": [ - { - "label": null, - "output_name": "output", - "uuid": "5a4a3708-2879-423a-84f0-aa756b7b60df" - } - ] + "uuid": "5f94f81f-3d8d-4f00-875e-8b4728d21d3c", + "workflow_outputs": [] }, "10": { "annotation": "", @@ -1534,8 +728,13 @@ "output_name": "output" } }, - "inputs": [], - "label": "Sort_BLAST_Hits", + "inputs": [ + { + "description": "runtime parameter for tool Sort", + "name": "infile" + } + ], + "label": "Sort BLAST Hits", "name": "Sort", "outputs": [ { @@ -1544,14 +743,14 @@ } ], "position": { - "bottom": 616.5, - "height": 116, - "left": -420, - "right": -220, - "top": 500.5, - "width": 200, - "x": -420, - "y": 500.5 + "bottom": 650.8669326554483, + "height": 50.91667175292969, + "left": 1981.840856751399, + "right": 2082.340856751399, + "top": 599.9502609025186, + "width": 100.5, + "x": 1981.840856751399, + "y": 599.9502609025186 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -1567,10 +766,10 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1.1", "type": "tool", - "uuid": "fe51dedc-8b91-4c57-9d5d-eb063c9107f3", + "uuid": "de530266-f665-4458-aa0d-e2a4712a5959", "workflow_outputs": [] }, "11": { @@ -1585,7 +784,7 @@ } }, "inputs": [], - "label": "Grouping_Hits", + "label": "Grouping Hits", "name": "AWK Script", "outputs": [ { @@ -1594,14 +793,14 @@ } ], "position": { - "bottom": 529.5, - "height": 116, - "left": -178, - "right": 22, - "top": 413.5, - "width": 200, - "x": -178, - "y": 413.5 + "bottom": 686.0174089118616, + "height": 61.16667175292969, + "left": 2170.1244752798507, + "right": 2270.6244752798507, + "top": 624.8507371589319, + "width": 100.5, + "x": 2170.1244752798507, + "y": 624.8507371589319 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -1624,10 +823,10 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"BEGIN {\\n getline first\\n split(first,last_line)\\n gap_dist=14000\\n min_length=8000\\n min_cov=0.95\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($2 == last_line[2]) && ($8 == last_line[8]) && (($6 < $7) ? ($6 <= last_line[7]+gap_dist) : ($7 <= last_line[6]+gap_dist))) {\\n if ($8 == \\\"+\\\") {\\n last_line[6] = (last_line[6] < $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] > $7 ? last_line[7] : $7)\\n } else {\\n last_line[6] = (last_line[6] > $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] < $7 ? last_line[7] : $7)\\n }\\n if ($4 < last_line[4]) {\\n last_line[3] += ($5 < last_line[4] ? $5 : last_line[4]-1) - $4 + 1\\n last_line[4] = $4\\n }\\n if ($5 > last_line[5]) {\\n last_line[3] += $5 - ($4 > last_line[5] ? $4 : last_line[5]+1) + 1\\n last_line[5] = $5\\n }\\n if (last_line[3]<$3) {\\n last_line[3] = $3\\n }\\n } else {\\n if ((last_line[3] >= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"blast\\\", \\\"genomic_island\\\", (last_line[6]= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"blast\\\", \\\"genomic_island\\\", (last_line[6] $7 ? last_line[7] : $7)\\n } else {\\n last_line[6] = (last_line[6] > $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] < $7 ? last_line[7] : $7)\\n }\\n if ($4 < last_line[4]) {\\n last_line[3] += ($5 < last_line[4] ? $5 : last_line[4]-1) - $4 + 1\\n last_line[4] = $4\\n }\\n if ($5 > last_line[5]) {\\n last_line[3] += $5 - ($4 > last_line[5] ? $4 : last_line[5]+1) + 1\\n last_line[5] = $5\\n }\\n if (last_line[3]<$3) {\\n last_line[3] = $3\\n }\\n } else {\\n if ((last_line[3] >= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"blast\\\", \\\"genomic_island\\\", (last_line[6]= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"blast\\\", \\\"genomic_island\\\", (last_line[6]= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1", + "type": "tool", + "uuid": "c8bfa0cc-708a-4377-bbf7-d4cd94c3b787", + "workflow_outputs": [ + { + "label": "BLAST Check", + "output_name": "outfile", + "uuid": "f0e2fc33-efac-4699-a9a2-d4632dcf9d1a" + } + ] + } + }, + "tags": "", + "uuid": "b8a5c0cd-c86e-4d09-8dfb-78d574822549" + }, + "tool_id": "5858972939ce3b5b", + "type": "subworkflow", + "uuid": "e04067f8-c1b4-4669-ba7f-0f981f5a1ba5", + "workflow_outputs": [ + { + "label": null, + "output_name": "BLAST Check", + "uuid": "9d0a3d39-7164-4fc0-86ab-15328b3d1367" + } + ] + }, + "5": { + "annotation": "", + "id": 5, + "input_connections": { + "FASTA Formatted Genomes": { + "id": 2, + "input_subworkflow_step_id": 0, + "output_name": "output" + }, + "Genbank Formatted Genomes": { + "id": 3, + "input_subworkflow_step_id": 1, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "Curated GI Detection", + "outputs": [], + "position": { + "bottom": 713.6833343505859, + "height": 147.18333435058594, + "left": -181, + "right": -31, + "top": 566.5, + "width": 150, + "x": -181, + "y": 566.5 + }, + "subworkflow": { + "a_galaxy_workflow": "true", + "annotation": "", + "format-version": "0.1", + "name": "Curated GI Detection", + "steps": { + "0": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 0, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "FASTA Formatted Genomes" + } + ], + "label": "FASTA Formatted Genomes", + "name": "Input dataset collection", + "outputs": [], + "position": { + "bottom": 279.03334045410156, + "height": 64.90000915527344, + "left": 138.5, + "right": 288.5, + "top": 214.13333129882812, + "width": 150, + "x": 138.5, + "y": 214.13333129882812 + }, + "tool_id": null, + "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", + "tool_version": null, + "type": "data_collection_input", + "uuid": "e885de9f-3182-40dd-8191-abf3de0e4894", + "workflow_outputs": [] + }, + "1": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 1, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "Genbank Formatted Genomes" + } + ], + "label": "Genbank Formatted Genomes", + "name": "Input dataset collection", + "outputs": [], + "position": { + "bottom": 414.5666809082031, + "height": 64.89999389648438, + "left": 138.5, + "right": 288.5, + "top": 349.66668701171875, + "width": 150, + "x": 138.5, + "y": 349.66668701171875 + }, + "tool_id": null, + "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", + "tool_version": null, + "type": "data_collection_input", + "uuid": "47361dc5-bb94-40cd-96ea-d7f8847a1380", + "workflow_outputs": [] + }, + "2": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/biopython_convert/biopython-convert/1.1", + "errors": null, + "id": 2, + "input_connections": { + "input": { + "id": 1, + "output_name": "output" + } + }, + "inputs": [], + "label": "Extract organism name", + "name": "BioPython SeqIO Converter", + "outputs": [ + { + "name": "output", + "type": "txt" + } + ], + "position": { + "bottom": 578.5833435058594, + "height": 228.89999389648438, + "left": 399.51666259765625, + "right": 549.5166625976562, + "top": 349.683349609375, + "width": 150, + "x": 399.51666259765625, + "y": 349.683349609375 + }, + "post_job_actions": { + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/biopython_convert/biopython-convert/1.1", + "tool_shed_repository": { + "changeset_revision": "1c71c770bfd4", + "name": "biopython_convert", + "owner": "brinkmanlab", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"bioperl\": \"false\", \"info\": \"false\", \"input\": {\"__class__\": \"ConnectedValue\"}, \"output_type\": \"text\", \"query\": \"[0].features[? type == `\\\"source\\\"`].qualifiers.organism[0]\", \"split\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1", "type": "tool", - "uuid": "7324e427-c6c1-4102-83f4-97ca856665e3", + "uuid": "865b8ca1-6d91-469e-b2ed-ba11b8bdc4c3", "workflow_outputs": [] }, - "14": { + "3": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sort_header_tool/1.1.1", + "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", "errors": null, - "id": 14, + "id": 3, "input_connections": { - "infile": { - "id": 13, - "output_name": "out_file1" + "infiles_0|input_type_selection|infile": { + "id": 2, + "output_name": "output" } }, "inputs": [], - "label": "Sort_All_BLAST_Groups", - "name": "Sort", + "label": "List Salmonella Files", + "name": "AWK Script", "outputs": [ { "name": "outfile", - "type": "input" + "type": "txt" } ], "position": { - "bottom": 391, - "height": 116, - "left": 794, - "right": 994, - "top": 275, - "width": 200, - "x": 794, - "y": 275 + "bottom": 572.6499938964844, + "height": 106.5, + "left": 639.5166625976562, + "right": 789.5166625976562, + "top": 466.1499938964844, + "width": 150, + "x": 639.5166625976562, + "y": 466.1499938964844 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -1785,114 +1268,647 @@ "output_name": "outfile" } }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sort_header_tool/1.1.1", + "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", "tool_shed_repository": { - "changeset_revision": "a6f147a050a2", - "name": "text_processing", - "owner": "bgruening", + "changeset_revision": "df0c8f33c8ea", + "name": "awkscript", + "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"4\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1.1", + "tool_state": "{\"code\": \"tolower($0) ~ \\\"^\\\" ENVIRON[\\\"target\\\"] { print tool_input_id }\", \"envs\": [{\"__index__\": 0, \"name\": \"target\", \"val\": \"salmonella enterica\"}], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"true\", \"__current_case__\": 1, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1", "type": "tool", - "uuid": "e19536bc-ef33-418e-a38f-888ba4b51d6f", + "uuid": "3d130f19-b95c-42dd-b816-d063a3b858a7", "workflow_outputs": [] }, - "15": { + "4": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", + "content_id": "__FILTER_FROM_FILE__", "errors": null, - "id": 15, + "id": 4, "input_connections": { - "infiles_0|input_type_selection|infile": { - "id": 14, + "how|filter_source": { + "id": 3, "output_name": "outfile" + }, + "input": { + "id": 0, + "output_name": "output" } }, "inputs": [], - "label": "Filter_Redundant", - "name": "AWK Script", + "label": "Filter Salmonella", + "name": "Filter collection", "outputs": [ { - "name": "outfile", - "type": "txt" + "name": "output_filtered", + "type": "input" + }, + { + "name": "output_discarded", + "type": "input" } ], "position": { - "bottom": 439, - "height": 137, - "left": 1088, - "right": 1288, - "top": 302, - "width": 200, - "x": 1088, - "y": 302 + "bottom": 402.6499938964844, + "height": 172.4833221435547, + "left": 880.5333251953125, + "right": 1030.5333251953125, + "top": 230.1666717529297, + "width": 150, + "x": 880.5333251953125, + "y": 230.1666717529297 }, "post_job_actions": { - "ChangeDatatypeActionoutfile": { - "action_arguments": { - "newtype": "gff" + "HideDatasetActionoutput_discarded": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_discarded" + }, + "HideDatasetActionoutput_filtered": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output_filtered" + } + }, + "tool_id": "__FILTER_FROM_FILE__", + "tool_state": "{\"how\": {\"how_filter\": \"remove_if_absent\", \"__current_case__\": 0, \"filter_source\": {\"__class__\": \"ConnectedValue\"}}, \"input\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.0.0", + "type": "tool", + "uuid": "2c209d0c-ded3-4480-8e62-c05d71eeb429", + "workflow_outputs": [] + }, + "5": { + "annotation": "", + "id": 5, + "input_connections": { + "FASTA Salmonella Genomes": { + "id": 4, + "input_subworkflow_step_id": 1, + "output_name": "output_filtered" + } + }, + "inputs": [], + "label": null, + "name": "Salmonella Curated GI Detection", + "outputs": [], + "position": { + "bottom": 418.43333435058594, + "height": 121.80000305175781, + "left": 1127.5, + "right": 1277.5, + "top": 296.6333312988281, + "width": 150, + "x": 1127.5, + "y": 296.6333312988281 + }, + "subworkflow": { + "a_galaxy_workflow": "true", + "annotation": "", + "format-version": "0.1", + "name": "Salmonella Curated GI Detection", + "steps": { + "0": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/artbio/get_reference_fasta/get_fasta_reference/0.3.1", + "errors": null, + "id": 0, + "input_connections": {}, + "inputs": [], + "label": "Get Curated Salmonella GI Database", + "name": "get fasta reference", + "outputs": [ + { + "name": "output", + "type": "fasta" + } + ], + "position": { + "bottom": 255.6999969482422, + "height": 80.19999694824219, + "left": 197, + "right": 347, + "top": 175.5, + "width": 150, + "x": 197, + "y": 175.5 + }, + "post_job_actions": { + "HideDatasetActionoutput": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/artbio/get_reference_fasta/get_fasta_reference/0.3.1", + "tool_shed_repository": { + "changeset_revision": "f3c9cbd84728", + "name": "get_reference_fasta", + "owner": "artbio", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"pre_installed_fasta\": \"salmonella_gis\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "0.3.1", + "type": "tool", + "uuid": "a40a5caf-7f8a-4ddb-a05f-07878cd935a0", + "workflow_outputs": [] + }, + "1": { + "annotation": "", + "content_id": null, + "errors": null, + "id": 1, + "input_connections": {}, + "inputs": [ + { + "description": "", + "name": "FASTA Salmonella Genomes" + } + ], + "label": "FASTA Salmonella Genomes", + "name": "Input dataset collection", + "outputs": [], + "position": { + "bottom": 408.6999969482422, + "height": 80.19999694824219, + "left": 189, + "right": 339, + "top": 328.5, + "width": 150, + "x": 189, + "y": 328.5 + }, + "tool_id": null, + "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", + "tool_version": null, + "type": "data_collection_input", + "uuid": "35d0f860-f980-4134-89b2-dd58e6076fd9", + "workflow_outputs": [] + }, + "2": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", + "errors": null, + "id": 2, + "input_connections": { + "input_file": { + "id": 1, + "output_name": "output" + } + }, + "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", + "name": "input_file" + }, + { + "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", + "name": "mask_data_file" + } + ], + "label": "Build Salmonella BLAST Database", + "name": "NCBI BLAST+ makeblastdb", + "outputs": [ + { + "name": "outfile", + "type": "data" + } + ], + "position": { + "bottom": 476.6000061035156, + "height": 147.10000610351562, + "left": 465, + "right": 615, + "top": 329.5, + "width": 150, + "x": 465, + "y": 329.5 + }, + "post_job_actions": { + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_makeblastdb/0.3.3", + "tool_shed_repository": { + "changeset_revision": "2889433c7ae1", + "name": "ncbi_blast_plus", + "owner": "devteam", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"RuntimeValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "0.3.3", + "type": "tool", + "uuid": "77439f1c-9ec1-4b0d-84ab-f8618197a0df", + "workflow_outputs": [] + }, + "3": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.3.3", + "errors": null, + "id": 3, + "input_connections": { + "db_opts|histdb": { + "id": 2, + "output_name": "outfile" + }, + "query": { + "id": 0, + "output_name": "output" + } + }, + "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "db_opts" + }, + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "query" + } + ], + "label": "BLASTn Salmonella GI DB", + "name": "NCBI BLAST+ blastn", + "outputs": [ + { + "name": "output1", + "type": "tabular" + } + ], + "position": { + "bottom": 417.8999938964844, + "height": 162.39999389648438, + "left": 749, + "right": 899, + "top": 255.5, + "width": 150, + "x": 749, + "y": 255.5 + }, + "post_job_actions": { + "ChangeDatatypeActionoutput1": { + "action_arguments": { + "newtype": "tabular" + }, + "action_type": "ChangeDatatypeAction", + "output_name": "output1" + }, + "HideDatasetActionoutput1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "output1" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/devteam/ncbi_blast_plus/ncbi_blastn_wrapper/0.3.3", + "tool_shed_repository": { + "changeset_revision": "2889433c7ae1", + "name": "ncbi_blast_plus", + "owner": "devteam", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"RuntimeValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "0.3.3", + "type": "tool", + "uuid": "57e2a49b-27d5-4f8c-8c1e-ed5d0d9dc9e9", + "workflow_outputs": [] }, - "action_type": "ChangeDatatypeAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", - "tool_shed_repository": { - "changeset_revision": "df0c8f33c8ea", - "name": "awkscript", - "owner": "brinkmanlab", - "tool_shed": "toolshed.g2.bx.psu.edu" + "4": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", + "errors": null, + "id": 4, + "input_connections": { + "infiles_0|input_type_selection|infile": { + "id": 3, + "output_name": "output1" + } + }, + "inputs": [], + "label": "BLAST Hit Filter", + "name": "AWK Script", + "outputs": [ + { + "name": "outfile", + "type": "txt" + } + ], + "position": { + "bottom": 359.78334045410156, + "height": 91.28334045410156, + "left": 968, + "right": 1118, + "top": 268.5, + "width": 150, + "x": 968, + "y": 268.5 + }, + "post_job_actions": { + "ChangeDatatypeActionoutfile": { + "action_arguments": { + "newtype": "tabular" + }, + "action_type": "ChangeDatatypeAction", + "output_name": "outfile" + }, + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", + "tool_shed_repository": { + "changeset_revision": "df0c8f33c8ea", + "name": "awkscript", + "owner": "brinkmanlab", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1", + "type": "tool", + "uuid": "26d4ec13-b9df-418a-8693-786be2b57f70", + "workflow_outputs": [] + }, + "5": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sort_header_tool/1.1.1", + "errors": null, + "id": 5, + "input_connections": { + "infile": { + "id": 4, + "output_name": "outfile" + } + }, + "inputs": [ + { + "description": "runtime parameter for tool Sort", + "name": "infile" + } + ], + "label": "Sort BLAST Hits", + "name": "Sort", + "outputs": [ + { + "name": "outfile", + "type": "input" + } + ], + "position": { + "bottom": 352.98333740234375, + "height": 75.98333740234375, + "left": 1185, + "right": 1335, + "top": 277, + "width": 150, + "x": 1185, + "y": 277 + }, + "post_job_actions": { + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sort_header_tool/1.1.1", + "tool_shed_repository": { + "changeset_revision": "a6f147a050a2", + "name": "text_processing", + "owner": "bgruening", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1.1", + "type": "tool", + "uuid": "35286835-6249-4ca2-a076-a23c4ae30d04", + "workflow_outputs": [] + }, + "6": { + "annotation": "", + "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", + "errors": null, + "id": 6, + "input_connections": { + "infiles_0|input_type_selection|infile": { + "id": 5, + "output_name": "outfile" + } + }, + "inputs": [], + "label": "Grouping Hits", + "name": "AWK Script", + "outputs": [ + { + "name": "outfile", + "type": "txt" + } + ], + "position": { + "bottom": 367.28334045410156, + "height": 91.28334045410156, + "left": 1393, + "right": 1543, + "top": 276, + "width": 150, + "x": 1393, + "y": 276 + }, + "post_job_actions": { + "ChangeDatatypeActionoutfile": { + "action_arguments": { + "newtype": "gff" + }, + "action_type": "ChangeDatatypeAction", + "output_name": "outfile" + }, + "HideDatasetActionoutfile": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile" + } + }, + "tool_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/awkscript/awkscript/1.1", + "tool_shed_repository": { + "changeset_revision": "df0c8f33c8ea", + "name": "awkscript", + "owner": "brinkmanlab", + "tool_shed": "toolshed.g2.bx.psu.edu" + }, + "tool_state": "{\"code\": \"BEGIN {\\n getline first\\n split(first,last_line)\\n gap_dist=14000\\n min_length=8000\\n min_cov=0.95\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($2 == last_line[2]) && ($8 == last_line[8]) && (($6 < $7) ? ($6 <= last_line[7]+gap_dist) : ($7 <= last_line[6]+gap_dist))) {\\n if ($8 == \\\"+\\\") {\\n last_line[6] = (last_line[6] < $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] > $7 ? last_line[7] : $7)\\n } else {\\n last_line[6] = (last_line[6] > $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] < $7 ? last_line[7] : $7)\\n }\\n if ($4 < last_line[4]) {\\n last_line[3] += ($5 < last_line[4] ? $5 : last_line[4]-1) - $4 + 1\\n last_line[4] = $4\\n }\\n if ($5 > last_line[5]) {\\n last_line[3] += $5 - ($4 > last_line[5] ? $4 : last_line[5]+1) + 1\\n last_line[5] = $5\\n }\\n if (last_line[3]<$3) {\\n last_line[3] = $3\\n }\\n } else {\\n if ((last_line[3] >= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"curated\\\", \\\"genomic_island\\\", (last_line[6]= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"curated\\\", \\\"genomic_island\\\", (last_line[6]= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1", + "type": "tool", + "uuid": "afd56aab-2c50-4fdd-94a2-b2f1ebecaf05", + "workflow_outputs": [ + { + "label": "Curated Salmonella Islands", + "output_name": "outfile", + "uuid": "2ea65090-747b-4b26-96b4-acfb098bfb8a" + } + ] + } + }, + "tags": "", + "uuid": "6beb3a66-9e2c-44d7-86ff-90a84ad754ca" }, - "tool_state": "{\"code\": \"BEGIN { \\n getline first\\n split(first,last_line)\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($4 <= last_line[5]) && ($5 >= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1", - "type": "tool", - "uuid": "63ddf586-e47c-4bf9-a455-f24fbb6f3d3b", + "tool_id": "af038e9879627e7a", + "type": "subworkflow", + "uuid": "a556da9f-bac2-433e-b4d3-d38f8b43057a", "workflow_outputs": [ { "label": null, - "output_name": "outfile", - "uuid": "0a48ea42-4d76-4fa0-b47a-b2b1308ca890" + "output_name": "Curated Salmonella Islands", + "uuid": "1b7a8463-f187-44b3-a923-72b25342811a" } ] } }, "tags": "", - "uuid": "0640d563-c0b6-4f14-af57-6b28e16365f7" + "uuid": "437c6491-9287-4dc2-9284-d9fae9fdd260" }, - "tool_id": "c5fef218dc723258", + "tool_id": "3f1254ab68c6af36", "type": "subworkflow", - "uuid": "7c9b70fb-538c-45b5-92d5-66bc1312a22b", + "uuid": "d7562afb-005d-48fe-b6b2-b23efeb3db83", "workflow_outputs": [ { "label": null, - "output_name": "15:outfile", - "uuid": "4c699612-d4d9-4f84-a355-1e2a8c5c9e51" - }, - { - "label": "input dataset(s) (filtered empty datasets)", - "output_name": "9:output", - "uuid": "0a9bd815-da7b-40d1-aaa9-abb6693d80f9" + "output_name": "5:Curated Salmonella Islands", + "uuid": "891398b5-cda3-4f8b-8ae1-52ba523f326b" } ] }, - "7": { + "6": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", "errors": null, - "id": 7, + "id": 6, "input_connections": { "inputs": { + "id": 1, + "output_name": "output" + }, + "queries_0|inputs2": { "id": 4, - "output_name": "outfile" + "output_name": "BLAST Check" } }, - "inputs": [ - { - "description": "runtime parameter for tool Concatenate datasets", - "name": "inputs" - } - ], + "inputs": [], "label": null, "name": "Concatenate datasets", "outputs": [ @@ -1902,14 +1918,14 @@ } ], "position": { - "bottom": 652.671875, - "height": 58, - "left": 630.21875, - "right": 730.21875, - "top": 594.671875, - "width": 100, - "x": 630.21875, - "y": 594.671875 + "bottom": 338.5, + "height": 116.5, + "left": 34.31666564941406, + "right": 184.31666564941406, + "top": 222, + "width": 150, + "x": 34.31666564941406, + "y": 222 }, "post_job_actions": { "HideDatasetActionout_file1": { @@ -1925,21 +1941,21 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"inputs\": {\"__class__\": \"RuntimeValue\"}, \"queries\": [], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"inputs\": {\"__class__\": \"ConnectedValue\"}, \"queries\": [{\"__index__\": 0, \"inputs2\": {\"__class__\": \"ConnectedValue\"}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.1.0", "type": "tool", - "uuid": "dd746f71-92ce-4f03-9a2c-69f9660b3bfd", + "uuid": "b9e69461-db1b-47bc-800e-47254de96b70", "workflow_outputs": [] }, - "8": { + "7": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/1.1.1", "errors": null, - "id": 8, + "id": 7, "input_connections": { "infile": { "id": 5, - "output_name": "5:8:outfile" + "output_name": "5:Curated Salmonella Islands" } }, "inputs": [ @@ -1957,14 +1973,14 @@ } ], "position": { - "bottom": 574.84375, - "height": 58, - "left": 911.5, - "right": 1011.5, - "top": 516.84375, - "width": 100, - "x": 911.5, - "y": 516.84375 + "bottom": 595.8332977294922, + "height": 75.98332214355469, + "left": 37.81666564941406, + "right": 187.81666564941406, + "top": 519.8499755859375, + "width": 150, + "x": 37.81666564941406, + "y": 519.8499755859375 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -1990,93 +2006,25 @@ "tool_state": "{\"code\": \"$3==\\\"genomic_island\\\" { print }\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1.1", "type": "tool", - "uuid": "075ee654-0a60-4138-bd6c-b8462f1e4031", - "workflow_outputs": [] - }, - "9": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", - "errors": null, - "id": 9, - "input_connections": { - "inputs": { - "id": 7, - "output_name": "out_file1" - }, - "queries_0|inputs2": { - "id": 6, - "output_name": "15:outfile" - } - }, - "inputs": [ - { - "description": "runtime parameter for tool Concatenate datasets", - "name": "inputs" - } - ], - "label": null, - "name": "Concatenate datasets", - "outputs": [ - { - "name": "out_file1", - "type": "input" - } - ], - "position": { - "bottom": 870.5, - "height": 73.5, - "left": 828, - "right": 928, - "top": 797, - "width": 100, - "x": 828, - "y": 797 - }, - "post_job_actions": { - "HideDatasetActionout_file1": { - "action_arguments": {}, - "action_type": "HideDatasetAction", - "output_name": "out_file1" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", - "tool_shed_repository": { - "changeset_revision": "a6f147a050a2", - "name": "text_processing", - "owner": "bgruening", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"inputs\": {\"__class__\": \"RuntimeValue\"}, \"queries\": [{\"__index__\": 0, \"inputs2\": {\"__class__\": \"RuntimeValue\"}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "0.1.0", - "type": "tool", - "uuid": "35f0b3e8-f79a-4edf-954e-9309db83544d", + "uuid": "d0ec6484-78a1-4520-bf98-a43ac631a0a2", "workflow_outputs": [] }, - "10": { + "8": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_intersectbed/2.27.1", "errors": null, - "id": 10, + "id": 8, "input_connections": { "inputA": { - "id": 9, + "id": 6, "output_name": "out_file1" }, "reduce_or_iterate|inputB": { - "id": 8, + "id": 7, "output_name": "outfile" } }, - "inputs": [ - { - "description": "runtime parameter for tool bedtools Intersect intervals", - "name": "inputA" - }, - { - "description": "runtime parameter for tool bedtools Intersect intervals", - "name": "reduce_or_iterate" - } - ], + "inputs": [], "label": "Filtered GIs - Do Not Overlap Curated", "name": "bedtools Intersect intervals", "outputs": [ @@ -2086,16 +2034,23 @@ } ], "position": { - "bottom": 787.71875, - "height": 73.5, - "left": 1175.140625, - "right": 1275.140625, - "top": 714.21875, - "width": 100, - "x": 1175.140625, - "y": 714.21875 + "bottom": 389.7166748046875, + "height": 116.5, + "left": 321.45001220703125, + "right": 471.45001220703125, + "top": 273.2166748046875, + "width": 150, + "x": 321.45001220703125, + "y": 273.2166748046875 }, "post_job_actions": { + "ChangeDatatypeActionoutput": { + "action_arguments": { + "newtype": "gff3" + }, + "action_type": "ChangeDatatypeAction", + "output_name": "output" + }, "HideDatasetActionoutput": { "action_arguments": {}, "action_type": "HideDatasetAction", @@ -2109,37 +2064,28 @@ "owner": "iuc", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"bed\": \"false\", \"count\": \"false\", \"fraction_cond\": {\"fraction_select\": \"default\", \"__current_case__\": 0}, \"header\": \"false\", \"inputA\": {\"__class__\": \"RuntimeValue\"}, \"invert\": \"true\", \"once\": \"false\", \"overlap_mode\": null, \"reduce_or_iterate\": {\"reduce_or_iterate_selector\": \"iterate\", \"__current_case__\": 0, \"inputB\": {\"__class__\": \"RuntimeValue\"}}, \"sorted\": \"false\", \"split\": \"false\", \"strand\": \"\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"bed\": \"false\", \"count\": \"false\", \"fraction_cond\": {\"fraction_select\": \"default\", \"__current_case__\": 0}, \"header\": \"false\", \"inputA\": {\"__class__\": \"ConnectedValue\"}, \"invert\": \"true\", \"once\": \"false\", \"overlap_mode\": null, \"reduce_or_iterate\": {\"reduce_or_iterate_selector\": \"iterate\", \"__current_case__\": 0, \"inputB\": {\"__class__\": \"ConnectedValue\"}}, \"sorted\": \"false\", \"split\": \"false\", \"strand\": \"\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "2.27.1", "type": "tool", - "uuid": "160e7a5e-76e6-49a5-b7c1-5ca4efd17bcd", + "uuid": "acd4b201-fc17-4716-b6a6-d7d554296cfe", "workflow_outputs": [] }, - "11": { + "9": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/iuc/bedtools/bedtools_intersectbed/2.27.1", "errors": null, - "id": 11, + "id": 9, "input_connections": { "inputA": { - "id": 9, + "id": 6, "output_name": "out_file1" }, "reduce_or_iterate|inputB": { - "id": 8, + "id": 7, "output_name": "outfile" } }, - "inputs": [ - { - "description": "runtime parameter for tool bedtools Intersect intervals", - "name": "inputA" - }, - { - "description": "runtime parameter for tool bedtools Intersect intervals", - "name": "reduce_or_iterate" - } - ], + "inputs": [], "label": "Removed Predictions - Overlap Curated", "name": "bedtools Intersect intervals", "outputs": [ @@ -2149,19 +2095,21 @@ } ], "position": { - "bottom": 1023.125, - "height": 84, - "left": 1147.109375, - "right": 1247.109375, - "top": 939.125, - "width": 100, - "x": 1147.109375, - "y": 939.125 + "bottom": 623.9333343505859, + "height": 131.8000030517578, + "left": 319.433349609375, + "right": 469.433349609375, + "top": 492.1333312988281, + "width": 150, + "x": 319.433349609375, + "y": 492.1333312988281 }, "post_job_actions": { - "HideDatasetActionoutput": { - "action_arguments": {}, - "action_type": "HideDatasetAction", + "ChangeDatatypeActionoutput": { + "action_arguments": { + "newtype": "gff3" + }, + "action_type": "ChangeDatatypeAction", "output_name": "output" } }, @@ -2172,213 +2120,122 @@ "owner": "iuc", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"bed\": \"false\", \"count\": \"false\", \"fraction_cond\": {\"fraction_select\": \"default\", \"__current_case__\": 0}, \"header\": \"false\", \"inputA\": {\"__class__\": \"RuntimeValue\"}, \"invert\": \"false\", \"once\": \"false\", \"overlap_mode\": [\"-wa\"], \"reduce_or_iterate\": {\"reduce_or_iterate_selector\": \"iterate\", \"__current_case__\": 0, \"inputB\": {\"__class__\": \"RuntimeValue\"}}, \"sorted\": \"false\", \"split\": \"false\", \"strand\": \"\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"bed\": \"false\", \"count\": \"false\", \"fraction_cond\": {\"fraction_select\": \"default\", \"__current_case__\": 0}, \"header\": \"false\", \"inputA\": {\"__class__\": \"ConnectedValue\"}, \"invert\": \"false\", \"once\": \"false\", \"overlap_mode\": [\"-wa\"], \"reduce_or_iterate\": {\"reduce_or_iterate_selector\": \"iterate\", \"__current_case__\": 0, \"inputB\": {\"__class__\": \"ConnectedValue\"}}, \"sorted\": \"false\", \"split\": \"false\", \"strand\": \"\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "2.27.1", "type": "tool", - "uuid": "b1785d2f-5c88-48cc-bbe1-5b43e9ffcd3c", - "workflow_outputs": [] + "uuid": "ae28a385-26ef-46e2-b860-8d3f1c51b519", + "workflow_outputs": [ + { + "label": null, + "output_name": "output", + "uuid": "9a1574c6-dd7c-4cb9-9912-e58f3a2b8b0f" + } + ] }, - "12": { + "10": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/1.1.1", + "content_id": "__BUILD_LIST__", "errors": null, - "id": 12, + "id": 10, "input_connections": { - "infile": { - "id": 10, + "datasets_0|input": { + "id": 8, "output_name": "output" } }, - "inputs": [ - { - "description": "runtime parameter for tool Text reformatting", - "name": "infile" - } - ], + "inputs": [], "label": null, - "name": "Text reformatting", + "name": "Build list", "outputs": [ { - "name": "outfile", + "name": "output", "type": "input" } ], "position": { - "bottom": 838.171875, - "height": 58, - "left": 1474, - "right": 1574, - "top": 780.171875, - "width": 100, - "x": 1474, - "y": 780.171875 + "bottom": 322.28334045410156, + "height": 91.28334045410156, + "left": 526.5333251953125, + "right": 676.5333251953125, + "top": 231, + "width": 150, + "x": 526.5333251953125, + "y": 231 }, "post_job_actions": { - "ChangeDatatypeActionoutfile": { - "action_arguments": { - "newtype": "gff3" - }, - "action_type": "ChangeDatatypeAction", - "output_name": "outfile" - }, - "HideDatasetActionoutfile": { + "HideDatasetActionoutput": { "action_arguments": {}, "action_type": "HideDatasetAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/1.1.1", - "tool_shed_repository": { - "changeset_revision": "a6f147a050a2", - "name": "text_processing", - "owner": "bgruening", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"code\": \"{ \\nif ($9 == \\\".\\\") $9 = \\\"\\\";\\nprint\\n}\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1.1", - "type": "tool", - "uuid": "a9cc5c7e-82d2-4b87-92f2-4978909c2198", - "workflow_outputs": [] - }, - "13": { - "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/1.1.1", - "errors": null, - "id": 13, - "input_connections": { - "infile": { - "id": 11, "output_name": "output" } }, - "inputs": [ - { - "description": "runtime parameter for tool Text reformatting", - "name": "infile" - } - ], - "label": null, - "name": "Text reformatting", - "outputs": [ - { - "name": "outfile", - "type": "input" - } - ], - "position": { - "bottom": 1019, - "height": 58, - "left": 1452, - "right": 1552, - "top": 961, - "width": 100, - "x": 1452, - "y": 961 - }, - "post_job_actions": { - "ChangeDatatypeActionoutfile": { - "action_arguments": { - "newtype": "gff3" - }, - "action_type": "ChangeDatatypeAction", - "output_name": "outfile" - } - }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_awk_tool/1.1.1", - "tool_shed_repository": { - "changeset_revision": "a6f147a050a2", - "name": "text_processing", - "owner": "bgruening", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"code\": \"{ \\nif ($9 == \\\".\\\") $9 = \\\"\\\";\\nprint\\n}\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "1.1.1", + "tool_id": "__BUILD_LIST__", + "tool_state": "{\"datasets\": [{\"__index__\": 0, \"input\": {\"__class__\": \"ConnectedValue\"}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.1.0", "type": "tool", - "uuid": "959dba15-591d-41dc-b5ba-76505da6c432", - "workflow_outputs": [ - { - "label": null, - "output_name": "outfile", - "uuid": "7b346f81-daf6-4554-8100-4605e6b3aab2" - } - ] + "uuid": "8fa7d9fa-16bd-4f89-9f12-347dbb682645", + "workflow_outputs": [] }, - "14": { + "11": { "annotation": "", - "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", + "content_id": "__MERGE_COLLECTION__", "errors": null, - "id": 14, + "id": 11, "input_connections": { - "inputs": { - "id": 2, + "inputs_0|input": { + "id": 0, "output_name": "output" }, - "queries_0|inputs2": { - "id": 12, - "output_name": "outfile" + "inputs_1|input": { + "id": 10, + "output_name": "output" } }, - "inputs": [ - { - "description": "runtime parameter for tool Concatenate datasets", - "name": "inputs" - } - ], + "inputs": [], "label": null, - "name": "Concatenate datasets", + "name": "Merge collections", "outputs": [ { - "name": "out_file1", + "name": "output", "type": "input" } ], "position": { - "bottom": 618.5, - "height": 73.5, - "left": 1604, - "right": 1704, - "top": 545, - "width": 100, - "x": 1604, - "y": 545 + "bottom": 254.89999389648438, + "height": 162.39999389648438, + "left": 720.5333251953125, + "right": 870.5333251953125, + "top": 92.5, + "width": 150, + "x": 720.5333251953125, + "y": 92.5 }, "post_job_actions": { - "HideDatasetActionout_file1": { + "HideDatasetActionoutput": { "action_arguments": {}, "action_type": "HideDatasetAction", - "output_name": "out_file1" + "output_name": "output" } }, - "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", - "tool_shed_repository": { - "changeset_revision": "a6f147a050a2", - "name": "text_processing", - "owner": "bgruening", - "tool_shed": "toolshed.g2.bx.psu.edu" - }, - "tool_state": "{\"inputs\": {\"__class__\": \"RuntimeValue\"}, \"queries\": [{\"__index__\": 0, \"inputs2\": {\"__class__\": \"RuntimeValue\"}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", - "tool_version": "0.1.0", + "tool_id": "__MERGE_COLLECTION__", + "tool_state": "{\"advanced\": {\"conflict\": {\"duplicate_options\": \"suffix_conflict_rest\", \"__current_case__\": 1, \"suffix_pattern\": \"_#\"}}, \"inputs\": [{\"__index__\": 0, \"input\": {\"__class__\": \"ConnectedValue\"}}, {\"__index__\": 1, \"input\": {\"__class__\": \"ConnectedValue\"}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_version": "1.0.0", "type": "tool", - "uuid": "d140007c-3314-4ac1-9759-ead2488d4be0", + "uuid": "e8e469ad-8f3b-4ffb-a198-9eb565cb21ff", "workflow_outputs": [] }, - "15": { + "12": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/brinkmanlab/feature_merge/feature-merge/1.3", "errors": null, - "id": 15, + "id": 12, "input_connections": { "inputs": { - "id": 14, - "output_name": "out_file1" + "id": 11, + "output_name": "output" } }, - "inputs": [ - { - "description": "runtime parameter for tool GFF/GTF Feature Merge", - "name": "inputs" - } - ], + "inputs": [], "label": "Merge user annotations", "name": "GFF/GTF Feature Merge", "outputs": [ @@ -2388,14 +2245,14 @@ } ], "position": { - "bottom": 651.6875, - "height": 58, - "left": 1873.6875, - "right": 1973.6875, - "top": 593.6875, - "width": 100, - "x": 1873.6875, - "y": 593.6875 + "bottom": 490.8833465576172, + "height": 91.19999694824219, + "left": 913, + "right": 1063, + "top": 399.683349609375, + "width": 150, + "x": 913, + "y": 399.683349609375 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -2411,33 +2268,28 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"exact_only\": \"false\", \"exclude_components\": \"true\", \"features\": [{\"__index__\": 0, \"csv\": \"genomic_island\"}], \"ignore_strand\": \"true\", \"inputs\": {\"__class__\": \"RuntimeValue\"}, \"merge_strategy\": \"append\", \"threshold\": \"500\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"exact_only\": \"false\", \"exclude_components\": \"true\", \"features\": [{\"__index__\": 0, \"csv\": \"genomic_island\"}], \"ignore_strand\": \"true\", \"inputs\": {\"__class__\": \"ConnectedValue\"}, \"merge_strategy\": \"append\", \"threshold\": \"500\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.3", "type": "tool", - "uuid": "a1024fc0-8d78-4667-996f-d619591a5db1", + "uuid": "24c516eb-e3f7-4c20-a957-3da93ee8e140", "workflow_outputs": [] }, - "16": { + "13": { "annotation": "", "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", "errors": null, - "id": 16, + "id": 13, "input_connections": { "inputs": { - "id": 5, - "output_name": "5:8:outfile" + "id": 12, + "output_name": "output" }, "queries_0|inputs2": { - "id": 15, - "output_name": "output" + "id": 5, + "output_name": "5:Curated Salmonella Islands" } }, - "inputs": [ - { - "description": "runtime parameter for tool Concatenate datasets", - "name": "inputs" - } - ], + "inputs": [], "label": null, "name": "Concatenate datasets", "outputs": [ @@ -2447,16 +2299,31 @@ } ], "position": { - "bottom": 552.5, - "height": 73.5, - "left": 2132, - "right": 2232, - "top": 479, - "width": 100, - "x": 2132, - "y": 479 + "bottom": 738.5, + "height": 116.5, + "left": 1111.316650390625, + "right": 1261.316650390625, + "top": 622, + "width": 150, + "x": 1111.316650390625, + "y": 622 + }, + "post_job_actions": { + "ChangeDatatypeActionout_file1": { + "action_arguments": { + "newtype": "gff3" + }, + "action_type": "ChangeDatatypeAction", + "output_name": "out_file1" + }, + "RenameDatasetActionout_file1": { + "action_arguments": { + "newname": "Final Annotations" + }, + "action_type": "RenameDatasetAction", + "output_name": "out_file1" + } }, - "post_job_actions": {}, "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_cat/0.1.0", "tool_shed_repository": { "changeset_revision": "a6f147a050a2", @@ -2464,20 +2331,20 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"inputs\": {\"__class__\": \"RuntimeValue\"}, \"queries\": [{\"__index__\": 0, \"inputs2\": {\"__class__\": \"RuntimeValue\"}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"inputs\": {\"__class__\": \"ConnectedValue\"}, \"queries\": [{\"__index__\": 0, \"inputs2\": {\"__class__\": \"ConnectedValue\"}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.1.0", "type": "tool", - "uuid": "96475057-a2bb-4343-a3e7-80d0ea0f1242", + "uuid": "b22ca800-232a-453f-9d85-77aa50fc62d9", "workflow_outputs": [ { - "label": null, + "label": "Final Annotations", "output_name": "out_file1", - "uuid": "96723179-b960-4240-abcc-b5c577b922d8" + "uuid": "aef47f82-aaf3-446d-b474-1fc30595460e" } ] } }, "tags": [], - "uuid": "f38dc5fe-09bb-4711-9e40-bf74904a9079", - "version": 25 + "uuid": "47930c10-5cf3-4621-83c6-a28a7edaf97c", + "version": 7 } diff --git a/workflow/workflows/Salmonella_Curated_GI_Detection.ga b/workflow/workflows/Salmonella_Curated_GI_Detection.ga index a045bee..82c63c4 100644 --- a/workflow/workflows/Salmonella_Curated_GI_Detection.ga +++ b/workflow/workflows/Salmonella_Curated_GI_Detection.ga @@ -11,7 +11,7 @@ "id": 0, "input_connections": {}, "inputs": [], - "label": null, + "label": "Get Curated Salmonella GI Database", "name": "get fasta reference", "outputs": [ { @@ -20,14 +20,14 @@ } ], "position": { - "bottom": 347.5, - "height": 84, - "left": -862, - "right": -662, - "top": 263.5, - "width": 200, - "x": -862, - "y": 263.5 + "bottom": 255.6999969482422, + "height": 80.19999694824219, + "left": 197, + "right": 347, + "top": 175.5, + "width": 150, + "x": 197, + "y": 175.5 }, "post_job_actions": { "HideDatasetActionoutput": { @@ -46,7 +46,7 @@ "tool_state": "{\"pre_installed_fasta\": \"salmonella_gis\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.3.1", "type": "tool", - "uuid": "41c9a13a-229b-4fe1-9cae-07e7b1766a28", + "uuid": "a40a5caf-7f8a-4ddb-a05f-07878cd935a0", "workflow_outputs": [] }, "1": { @@ -65,20 +65,20 @@ "name": "Input dataset collection", "outputs": [], "position": { - "bottom": 494.5, - "height": 84, - "left": -866, - "right": -666, - "top": 410.5, - "width": 200, - "x": -866, - "y": 410.5 + "bottom": 408.6999969482422, + "height": 80.19999694824219, + "left": 189, + "right": 339, + "top": 328.5, + "width": 150, + "x": 189, + "y": 328.5 }, "tool_id": null, "tool_state": "{\"optional\": false, \"tag\": null, \"collection_type\": \"list\"}", "tool_version": null, "type": "data_collection_input", - "uuid": "86d770e4-8963-4b20-b3cf-8fb77fcc9ce6", + "uuid": "35d0f860-f980-4134-89b2-dd58e6076fd9", "workflow_outputs": [] }, "2": { @@ -93,12 +93,16 @@ } }, "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", + "name": "input_file" + }, { "description": "runtime parameter for tool NCBI BLAST+ makeblastdb", "name": "mask_data_file" } ], - "label": "Genomes_BLAST_Database", + "label": "Build Salmonella BLAST Database", "name": "NCBI BLAST+ makeblastdb", "outputs": [ { @@ -107,14 +111,14 @@ } ], "position": { - "bottom": 600.5, - "height": 189, - "left": -590, - "right": -390, - "top": 411.5, - "width": 200, - "x": -590, - "y": 411.5 + "bottom": 476.6000061035156, + "height": 147.10000610351562, + "left": 465, + "right": 615, + "top": 329.5, + "width": 150, + "x": 465, + "y": 329.5 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -130,10 +134,10 @@ "owner": "devteam", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"ConnectedValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"dbtype\": \"nucl\", \"hash_index\": \"false\", \"input_file\": {\"__class__\": \"RuntimeValue\"}, \"mask_data_file\": {\"__class__\": \"RuntimeValue\"}, \"parse_seqids\": \"false\", \"tax\": {\"taxselect\": \"\", \"__current_case__\": 0}, \"title\": \"Genomes\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.3.3", "type": "tool", - "uuid": "c358e05b-f1a3-4682-9480-7959dbc98e7b", + "uuid": "77439f1c-9ec1-4b0d-84ab-f8618197a0df", "workflow_outputs": [] }, "3": { @@ -151,8 +155,17 @@ "output_name": "output" } }, - "inputs": [], - "label": "GI_blastn", + "inputs": [ + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "db_opts" + }, + { + "description": "runtime parameter for tool NCBI BLAST+ blastn", + "name": "query" + } + ], + "label": "BLASTn Salmonella GI DB", "name": "NCBI BLAST+ blastn", "outputs": [ { @@ -161,14 +174,14 @@ } ], "position": { - "bottom": 523.5, - "height": 189, - "left": -307, - "right": -107, - "top": 334.5, - "width": 200, - "x": -307, - "y": 334.5 + "bottom": 417.8999938964844, + "height": 162.39999389648438, + "left": 749, + "right": 899, + "top": 255.5, + "width": 150, + "x": 749, + "y": 255.5 }, "post_job_actions": { "ChangeDatatypeActionoutput1": { @@ -191,10 +204,10 @@ "owner": "devteam", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"ConnectedValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"ConnectedValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"adv_opts\": {\"adv_opts_selector\": \"advanced\", \"__current_case__\": 1, \"filter_query\": \"true\", \"strand\": \"-strand both\", \"max_hits\": \"0\", \"max_hsps\": null, \"identity_cutoff\": \"0.0\", \"word_size\": null, \"ungapped\": \"false\", \"parse_deflines\": \"false\", \"adv_optional_id_files_opts\": {\"adv_optional_id_files_opts_selector\": \"none\", \"__current_case__\": 0}, \"qcov_hsp_perc\": \"0.0\", \"window_size\": null, \"gapopen\": null, \"gapextend\": null}, \"blast_type\": \"blastn\", \"db_opts\": {\"db_opts_selector\": \"histdb\", \"__current_case__\": 1, \"database\": \"\", \"histdb\": {\"__class__\": \"RuntimeValue\"}, \"subject\": \"\"}, \"evalue_cutoff\": \"0.001\", \"output\": {\"out_format\": \"cols\", \"__current_case__\": 2, \"std_cols\": [\"qseqid\", \"sseqid\", \"pident\", \"length\", \"mismatch\", \"gapopen\", \"qstart\", \"qend\", \"sstart\", \"send\", \"evalue\", \"bitscore\"], \"ext_cols\": [\"qlen\"], \"ids_cols\": null, \"misc_cols\": null, \"tax_cols\": null}, \"query\": {\"__class__\": \"RuntimeValue\"}, \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "0.3.3", "type": "tool", - "uuid": "822ca610-b1b1-492a-b57f-085279f3d1fa", + "uuid": "57e2a49b-27d5-4f8c-8c1e-ed5d0d9dc9e9", "workflow_outputs": [] }, "4": { @@ -209,7 +222,7 @@ } }, "inputs": [], - "label": "BLAST_Hit_Filter", + "label": "BLAST Hit Filter", "name": "AWK Script", "outputs": [ { @@ -218,14 +231,14 @@ } ], "position": { - "bottom": 500.5, - "height": 137, - "left": -4, - "right": 196, - "top": 363.5, - "width": 200, - "x": -4, - "y": 363.5 + "bottom": 359.78334045410156, + "height": 91.28334045410156, + "left": 968, + "right": 1118, + "top": 268.5, + "width": 150, + "x": 968, + "y": 268.5 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -248,10 +261,10 @@ "owner": "brinkmanlab", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"$3>=90 && $4>400 && $11<1.63e-7 {print $1, $2, $4, $7, $8, $9, $10, ($9<$10 ? \\\"+\\\" : \\\"-\\\"), $13}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "dd567ca6-6f19-408e-8167-a7eb34f7d085", + "uuid": "26d4ec13-b9df-418a-8693-786be2b57f70", "workflow_outputs": [] }, "5": { @@ -265,8 +278,13 @@ "output_name": "outfile" } }, - "inputs": [], - "label": "Sort_BLAST_Hits", + "inputs": [ + { + "description": "runtime parameter for tool Sort", + "name": "infile" + } + ], + "label": "Sort BLAST Hits", "name": "Sort", "outputs": [ { @@ -275,14 +293,14 @@ } ], "position": { - "bottom": 491, - "height": 116, - "left": 310, - "right": 510, - "top": 375, - "width": 200, - "x": 310, - "y": 375 + "bottom": 352.98333740234375, + "height": 75.98333740234375, + "left": 1185, + "right": 1335, + "top": 277, + "width": 150, + "x": 1185, + "y": 277 }, "post_job_actions": { "HideDatasetActionoutfile": { @@ -298,10 +316,10 @@ "owner": "bgruening", "tool_shed": "toolshed.g2.bx.psu.edu" }, - "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"ConnectedValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"header\": \"0\", \"ignore_case\": \"false\", \"infile\": {\"__class__\": \"RuntimeValue\"}, \"sortkeys\": [{\"__index__\": 0, \"column\": \"1\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 1, \"column\": \"2\", \"order\": \"\", \"style\": \"\"}, {\"__index__\": 2, \"column\": \"8\", \"order\": \"r\", \"style\": \"\"}, {\"__index__\": 3, \"column\": \"6\", \"order\": \"\", \"style\": \"n\"}], \"unique\": \"false\", \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1.1", "type": "tool", - "uuid": "f176af4f-afba-421c-a649-ad4c8dc1783f", + "uuid": "35286835-6249-4ca2-a076-a23c4ae30d04", "workflow_outputs": [] }, "6": { @@ -316,7 +334,7 @@ } }, "inputs": [], - "label": "Grouping_Hits", + "label": "Grouping Hits", "name": "AWK Script", "outputs": [ { @@ -325,14 +343,14 @@ } ], "position": { - "bottom": 488, - "height": 116, - "left": 624, - "right": 824, - "top": 372, - "width": 200, - "x": 624, - "y": 372 + "bottom": 367.28334045410156, + "height": 91.28334045410156, + "left": 1393, + "right": 1543, + "top": 276, + "width": 150, + "x": 1393, + "y": 276 }, "post_job_actions": { "ChangeDatatypeActionoutfile": { @@ -358,7 +376,7 @@ "tool_state": "{\"code\": \"BEGIN {\\n getline first\\n split(first,last_line)\\n gap_dist=14000\\n min_length=8000\\n min_cov=0.95\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($2 == last_line[2]) && ($8 == last_line[8]) && (($6 < $7) ? ($6 <= last_line[7]+gap_dist) : ($7 <= last_line[6]+gap_dist))) {\\n if ($8 == \\\"+\\\") {\\n last_line[6] = (last_line[6] < $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] > $7 ? last_line[7] : $7)\\n } else {\\n last_line[6] = (last_line[6] > $6 ? last_line[6] : $6)\\n last_line[7] = (last_line[7] < $7 ? last_line[7] : $7)\\n }\\n if ($4 < last_line[4]) {\\n last_line[3] += ($5 < last_line[4] ? $5 : last_line[4]-1) - $4 + 1\\n last_line[4] = $4\\n }\\n if ($5 > last_line[5]) {\\n last_line[3] += $5 - ($4 > last_line[5] ? $4 : last_line[5]+1) + 1\\n last_line[5] = $5\\n }\\n if (last_line[3]<$3) {\\n last_line[3] = $3\\n }\\n } else {\\n if ((last_line[3] >= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"curated\\\", \\\"genomic_island\\\", (last_line[6]= min_length) && (last_line[3]/last_line[9]>=min_cov)) {\\n print last_line[2], \\\"curated\\\", \\\"genomic_island\\\", (last_line[6]= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"ConnectedValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", + "tool_state": "{\"code\": \"BEGIN { \\n getline first\\n split(first,last_line)\\n}\\n\\n{\\n if (($1 == last_line[1]) && ($4 <= last_line[5]) && ($5 >= last_line[4])) {\\n if (last_line[5]-last_line[4] < $5-$4) {\\n split($0,last_line)\\n }\\n } else {\\n print join(last_line, \\\"\\\\t\\\")\\n split($0,last_line)\\n }\\n}\\n\\nEND {\\n print join(last_line, \\\"\\\\t\\\")\\n}\\n\\nfunction join(array, sep)\\n{\\n if (sep == \\\"\\\")\\n sep = \\\" \\\"\\n result = \\\"\\\"\\n for (i=1; i<=length(array); i++) {\\n if (i==1) {\\n result = array[i]\\n } else {\\n result = result sep array[i]\\n }\\n }\\n return result\\n}\", \"envs\": [], \"infiles\": [{\"__index__\": 0, \"input_type_selection\": {\"multiple\": \"false\", \"__current_case__\": 0, \"infile\": {\"__class__\": \"RuntimeValue\"}}}], \"__page__\": null, \"__rerun_remap_job_id__\": null}", "tool_version": "1.1", "type": "tool", - "uuid": "295c92c0-50fe-437a-9e1d-66ed6ce6507c", + "uuid": "afd56aab-2c50-4fdd-94a2-b2f1ebecaf05", "workflow_outputs": [ { - "label": null, + "label": "Curated Salmonella Islands", "output_name": "outfile", - "uuid": "b4086374-fc71-4248-9c56-988a5697945b" + "uuid": "2ea65090-747b-4b26-96b4-acfb098bfb8a" } ] } }, "tags": [], - "uuid": "c6320762-c3ac-483a-a879-c1af20c2af43", + "uuid": "6beb3a66-9e2c-44d7-86ff-90a84ad754ca", "version": 9 }