Mercurial > repos > mikel-egana-aranguren > sadi_docker_workflow
changeset 1:47db0b303a50 default tip
Added annotations to workflow
| author | Mikel Egana <mikel.egana.aranguren@gmail.com> |
|---|---|
| date | Mon, 07 Sep 2015 19:18:27 +0200 |
| parents | 22be3a551998 |
| children | |
| files | Galaxy-Workflow-SADI-Docker_use_case.ga |
| diffstat | 1 files changed, 520 insertions(+), 0 deletions(-) [+] |
line wrap: on
line diff
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/Galaxy-Workflow-SADI-Docker_use_case.ga Mon Sep 07 19:18:27 2015 +0200 @@ -0,0 +1,520 @@ +{ + "a_galaxy_workflow": "true", + "annotation": "The workflow answers the following question: Given a set of UniProt proteins, which ones are related to PubMed abstracts containing the term ``brain'', and what are they KEGG entries? The workflow starts from a simple list of UniProt identifiers, and retrieves different datasets from a regular SADI service (to obtain KEGG entries) and a set of 3 OpenLifeData2SADI services (to obtain PubMed abstracts). The results are then merged and queried to obtain the KEGG entries of proteins that are related to PubMed abstracts that contain the term", + "format-version": "0.1", + "name": "SADI-Docker use case", + "steps": { + "0": { + "annotation": "Convert white spaces to tabs, so that Galaxy recognises the list of UniProt IDs as a column", + "id": 0, + "input_connections": {}, + "inputs": [], + "label": null, + "name": "Convert", + "outputs": [ + { + "name": "out_file1", + "type": "tabular" + } + ], + "position": { + "left": 204.9895977973938, + "top": 202.90626764297485 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "Convert characters1", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"condense\": \"\\\"True\\\"\", \"strip\": \"\\\"True\\\"\", \"input\": \"null\", \"convert_from\": \"\\\"s\\\"\"}", + "tool_version": "1.0.0", + "type": "tool", + "user_outputs": [], + "uuid": "658762bb-9052-4dd2-bb6b-e77613a6dcfe" + }, + "1": { + "annotation": "Convert UniProt identifiers to URIs: Add a column with the first part of the URI", + "id": 1, + "input_connections": { + "input": { + "id": 0, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Add column", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 199.9895977973938, + "top": 335.9097466468811 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "addValue", + "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"__rerun_remap_job_id__\": null, \"exp\": \"\\\"http://openlifedata.org/uniprot:\\\"\", \"iterate\": \"\\\"no\\\"\"}", + "tool_version": "1.0.0", + "type": "tool", + "user_outputs": [], + "uuid": "c2421757-1d5d-435c-ac9e-fb036ccd4448" + }, + "2": { + "annotation": "Convert UniProt identifiers to URIs", + "id": 2, + "input_connections": { + "input1": { + "id": 1, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Merge Columns", + "outputs": [ + { + "name": "out_file1", + "type": "tabular" + } + ], + "position": { + "left": 201.96875429153442, + "top": 473.91322565078735 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "mergeCols1", + "tool_state": "{\"__page__\": 0, \"input1\": \"null\", \"__rerun_remap_job_id__\": null, \"col2\": \"\\\"\\\"\", \"col1\": \"\\\"\\\"\", \"columns\": \"[]\"}", + "tool_version": "1.0.1", + "type": "tool", + "user_outputs": [], + "uuid": "63b4d2ff-0e1d-449d-aad5-a2ccd033d36a" + }, + "3": { + "annotation": "Obtain the UniProt URIs", + "id": 3, + "input_connections": { + "input": { + "id": 2, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Cut", + "outputs": [ + { + "name": "out_file1", + "type": "tabular" + } + ], + "position": { + "left": 202.99307680130005, + "top": 592.906298160553 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "Cut1", + "tool_state": "{\"columnList\": \"\\\"c3\\\"\", \"input\": \"null\", \"delimiter\": \"\\\"T\\\"\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "1.0.2", + "type": "tool", + "user_outputs": [], + "uuid": "f3912f69-b561-4dd5-91f0-1e492e19bb9c" + }, + "4": { + "annotation": "Add a column with the URI of the SADI service input OWL class", + "id": 4, + "input_connections": { + "input": { + "id": 3, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Add column", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 448.9826703071594, + "top": 480.9097466468811 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "addValue", + "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"__rerun_remap_job_id__\": null, \"exp\": \"\\\"http://purl.oclc.org/SADI/LSRN/UniProt_Record\\\"\", \"iterate\": \"\\\"no\\\"\"}", + "tool_version": "1.0.0", + "type": "tool", + "user_outputs": [], + "uuid": "c01e5509-2a60-4ad4-848c-b31f9ccfc841" + }, + "5": { + "annotation": "Add a column with the URI of the SADI service input OWL class", + "id": 5, + "input_connections": { + "input": { + "id": 3, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "Add column", + "outputs": [ + { + "name": "out_file1", + "type": "input" + } + ], + "position": { + "left": 449.9896283149719, + "top": 636.9166741371155 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "addValue", + "tool_state": "{\"__page__\": 0, \"input\": \"null\", \"__rerun_remap_job_id__\": null, \"exp\": \"\\\"http://openlifedata.org/uniprot_vocabulary:Resource\\\"\", \"iterate\": \"\\\"no\\\"\"}", + "tool_version": "1.0.0", + "type": "tool", + "user_outputs": [], + "uuid": "98b1f088-3de5-4dbd-a655-63a630517e1a" + }, + "6": { + "annotation": "Convert to RDF, adding the triple rdf:type input OWL class", + "id": 6, + "input_connections": { + "input_file": { + "id": 4, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "SADI - Tab-to-RDF", + "outputs": [ + { + "name": "output_file", + "type": "rdf" + } + ], + "position": { + "left": 449.9896283149719, + "top": 340.9097466468811 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-tab2rdf", + "tool_state": "{\"__page__\": 0, \"triples\": \"[{\\\"__index__\\\": 0, \\\"o_col\\\": \\\"\\\", \\\"o_type\\\": \\\"False\\\", \\\"s_col\\\": \\\"\\\", \\\"p_val\\\": \\\"http://www.w3.org/1999/02/22-rdf-syntax-ns#type\\\"}]\", \"input_file\": \"null\", \"output_format\": \"\\\"ntriples\\\"\", \"namespace\": \"{\\\"choose_namespace\\\": \\\"none\\\", \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "a05b8d4f-1b5f-4930-bd77-dfb5163b4556" + }, + "7": { + "annotation": "Convert to RDF, adding the triple rdf:type input OWL class", + "id": 7, + "input_connections": { + "input_file": { + "id": 5, + "output_name": "out_file1" + } + }, + "inputs": [], + "label": null, + "name": "SADI - Tab-to-RDF", + "outputs": [ + { + "name": "output_file", + "type": "rdf" + } + ], + "position": { + "left": 447.9757122993469, + "top": 766.9166741371155 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-tab2rdf", + "tool_state": "{\"__page__\": 0, \"triples\": \"[{\\\"__index__\\\": 0, \\\"o_col\\\": \\\"\\\", \\\"o_type\\\": \\\"False\\\", \\\"s_col\\\": \\\"\\\", \\\"p_val\\\": \\\"http://www.w3.org/1999/02/22-rdf-syntax-ns#type\\\"}]\", \"input_file\": \"null\", \"output_format\": \"\\\"ntriples\\\"\", \"namespace\": \"{\\\"choose_namespace\\\": \\\"none\\\", \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "217c654d-5353-4d93-a6de-c5c5bb205e64" + }, + "8": { + "annotation": "Convert to RDF/XML syntax", + "id": 8, + "input_connections": { + "input_file": { + "id": 6, + "output_name": "output_file" + } + }, + "inputs": [], + "label": null, + "name": "SADI - RDF Format (Rapper)", + "outputs": [ + { + "name": "output_file", + "type": "rdf" + } + ], + "position": { + "left": 438.9826703071594, + "top": 200.90973138809204 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-rapper", + "tool_state": "{\"__page__\": 0, \"output_format\": \"\\\"rdfxml\\\"\", \"__rerun_remap_job_id__\": null, \"input_format\": \"\\\"ntriples\\\"\", \"input_file\": \"null\"}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "b5e1600e-a4ef-4ed6-a839-46bd1bcc59b3" + }, + "9": { + "annotation": "Convert to RDF/XML syntax", + "id": 9, + "input_connections": { + "input_file": { + "id": 7, + "output_name": "output_file" + } + }, + "inputs": [], + "label": null, + "name": "SADI - RDF Format (Rapper)", + "outputs": [ + { + "name": "output_file", + "type": "rdf" + } + ], + "position": { + "left": 446.98611879348755, + "top": 913.9132561683655 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-rapper", + "tool_state": "{\"__page__\": 0, \"output_format\": \"\\\"rdfxml\\\"\", \"__rerun_remap_job_id__\": null, \"input_format\": \"\\\"ntriples\\\"\", \"input_file\": \"null\"}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "995db5b3-4e12-48c8-bda0-f961861f008c" + }, + "10": { + "annotation": "Execute SADI service to obtain KEGG entries", + "id": 10, + "input_connections": { + "input": { + "id": 8, + "output_name": "output_file" + } + }, + "inputs": [], + "label": null, + "name": "SADI client", + "outputs": [ + { + "name": "output", + "type": "rdf" + } + ], + "position": { + "left": 775.9965863227844, + "top": 202.90626764297485 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-sadi_client", + "tool_state": "{\"url\": \"\\\"http://sadiframework.org/services/getKEGGIDFromUniProt\\\"\", \"input\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "01c2bbe7-a5e1-4bad-8c93-b01c6d0a5c8f" + }, + "11": { + "annotation": "Execute SADI service to obtain HGNC entries", + "id": 11, + "input_connections": { + "input": { + "id": 9, + "output_name": "output_file" + } + }, + "inputs": [], + "label": null, + "name": "SADI client", + "outputs": [ + { + "name": "output", + "type": "rdf" + } + ], + "position": { + "left": 788.9826703071594, + "top": 579.9201531410217 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-sadi_client", + "tool_state": "{\"url\": \"\\\"http://biordf.org/cgi-bin/SADI/OpenLifeData2SADI/SADI/hgnc/uniprot_vocabulary_Resource_hgnc_vocabulary_x-uniprot-inverse_hgnc_vocabulary_Resource\\\"\", \"input\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "ccdd413c-36be-48c3-96fc-4f233208a6b8" + }, + "12": { + "annotation": "Execute SADI service to obtain OMIM entries", + "id": 12, + "input_connections": { + "input": { + "id": 11, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "SADI client", + "outputs": [ + { + "name": "output", + "type": "rdf" + } + ], + "position": { + "left": 793.9826703071594, + "top": 753.9132561683655 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-sadi_client", + "tool_state": "{\"url\": \"\\\"http://biordf.org/cgi-bin/SADI/OpenLifeData2SADI/SADI/hgnc/hgnc_vocabulary_Resource_hgnc_vocabulary_x-omim_omim_vocabulary_Gene\\\"\", \"input\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "b7af50a8-a92a-44d9-bf1e-ca90cb69905d" + }, + "13": { + "annotation": "Execute SADI service to obtain PubMed entries", + "id": 13, + "input_connections": { + "input": { + "id": 12, + "output_name": "output" + } + }, + "inputs": [], + "label": null, + "name": "SADI client", + "outputs": [ + { + "name": "output", + "type": "rdf" + } + ], + "position": { + "left": 799.9896283149719, + "top": 914.9201531410217 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-sadi_client", + "tool_state": "{\"url\": \"\\\"http://biordf.org/cgi-bin/SADI/OpenLifeData2SADI/SADI/omim/omim_vocabulary_Gene_omim_vocabulary_article_pubmed_vocabulary_PubMedRecord\\\"\", \"input\": \"null\", \"__rerun_remap_job_id__\": null, \"__page__\": 0}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "3dcb6fc4-4021-4ea6-8e88-31c71d68699d" + }, + "14": { + "annotation": "Merge all the inputs and outputs in single RDF graph: SADI only decorates inputs with new triples, so we can keep track of the whole process via URIs", + "id": 14, + "input_connections": { + "input_files_0|input_file": { + "id": 13, + "output_name": "output" + }, + "input_files_1|input_file": { + "id": 12, + "output_name": "output" + }, + "input_files_2|input_file": { + "id": 11, + "output_name": "output" + }, + "input_files_3|input_file": { + "id": 10, + "output_name": "output" + }, + "input_files_4|input_file": { + "id": 9, + "output_name": "output_file" + }, + "input_files_5|input_file": { + "id": 8, + "output_name": "output_file" + } + }, + "inputs": [], + "label": null, + "name": "SADI - Merge RDF Graphs", + "outputs": [ + { + "name": "output_file", + "type": "rdf" + } + ], + "position": { + "left": 1125.9965252876282, + "top": 318.91319513320923 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-mergeRDFgraphs", + "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"input_files\": \"[{\\\"__index__\\\": 0, \\\"input_file\\\": null}, {\\\"__index__\\\": 1, \\\"input_file\\\": null}, {\\\"__index__\\\": 2, \\\"input_file\\\": null}, {\\\"__index__\\\": 3, \\\"input_file\\\": null}, {\\\"__index__\\\": 4, \\\"input_file\\\": null}, {\\\"__index__\\\": 5, \\\"input_file\\\": null}]\"}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "7c24dd55-3032-4329-9fa3-734c963d8ed2" + }, + "15": { + "annotation": "Query the merged RDF graph with the actual question", + "id": 15, + "input_connections": { + "ontology": { + "id": 14, + "output_name": "output_file" + } + }, + "inputs": [], + "label": null, + "name": "SADI - Execute an SPARQL query against an RDF file", + "outputs": [ + { + "name": "output", + "type": "html" + } + ], + "position": { + "left": 1439.993106842041, + "top": 461.92015266418457 + }, + "post_job_actions": {}, + "tool_errors": null, + "tool_id": "SADI-Docker-SPARQLGalaxy", + "tool_state": "{\"out_format\": \"\\\"html\\\"\", \"__page__\": 0, \"pasted_query\": \"\\\"PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> \\\\nPREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\\\\nPREFIX sadi: <http://sadiframework.org/ontologies/predicates.owl#>\\\\nPREFIX lsrn: <http://purl.oclc.org/SADI/LSRN/>\\\\n\\\\nSELECT ?protein ?label ?KEGG\\\\nWHERE { \\\\n?protein rdf:type lsrn:UniProt_Record . \\\\n?protein sadi:isEncodedBy ?KEGG . \\\\n?protein ?prot2hgnc ?hgnc . \\\\n?hgnc ?hgnc2omim ?omim . \\\\n?omim ?omim2pubmed ?pubmed . \\\\n?pubmed rdfs:label ?label . \\\\nFILTER (regex (?label, 'brain'))\\\\n}\\\"\", \"ontology\": \"null\", \"__rerun_remap_job_id__\": null}", + "tool_version": "0.1", + "type": "tool", + "user_outputs": [], + "uuid": "9a453f8c-370e-4ab7-8e45-73cbba721db9" + } + }, + "uuid": "8796e5be-1c91-4f91-9946-02a09a79401d" +} \ No newline at end of file