Mercurial > repos > nml > mob_suite
changeset 2:2c7f721a4905 draft
planemo upload for repository https://github.com/phac-nml/mob-suite commit eecf57ea6130c483487087554c53918fe2b556dd
author | nml |
---|---|
date | Fri, 17 Aug 2018 11:12:08 -0400 |
parents | 2c419df5cdaa |
children | 066b2a1c6a55 |
files | README.md mob_recon.xml mob_typer.xml workflows/AMRworkflow_STARAMR.ga |
diffstat | 4 files changed, 223 insertions(+), 70 deletions(-) [+] |
line wrap: on
line diff
--- a/README.md Thu Jul 26 14:50:59 2018 -0400 +++ b/README.md Fri Aug 17 11:12:08 2018 -0400 @@ -1,8 +1,7 @@ # MOB-suite: Software tools for clustering, reconstruction and typing of plasmids from draft assemblies -These are Galaxy wrappers for [MOB-Suite](https://github.com/phac-nml/mob-suite): a set of tools designed to be a modular set of tools for the typing and reconstruction of plasmid sequences from WGS assemblies. +These are Galaxy wrappers for [MOB-Suite](https://github.com/phac-nml/mob-suite): a set of tools designed to be a modular set of tools for the typing and reconstruction of plasmid sequences from WGS assemblies. This wrapper allows to conveniently run all these tools on a Galaxy instance. Also coupling MOB-Suite tools with other AMR tools via workflows allows to obtain both mobilome and resistome valuable information. For more information please refer to https://github.com/phac-nml/mob-suite -
--- a/mob_recon.xml Thu Jul 26 14:50:59 2018 -0400 +++ b/mob_recon.xml Fri Aug 17 11:12:08 2018 -0400 @@ -1,13 +1,17 @@ -<tool id="mob_recon" name="MOB-Recon" version="1.4.5"> +<tool id="mob_recon" name="MOB-Recon" version="1.4.8"> <description>Type contigs and extract plasmid sequences</description> <requirements> - <requirement type="package" version="1.4.5">mob_suite</requirement> + <requirement type="package" version="1.4.8">mob_suite</requirement> </requirements> <command detect_errors="exit_code"> <![CDATA[ - ln -s "${input}" "${input.name}"; + #import re + #import os.path - mob_recon --num_threads \${GALAXY_SLOTS:-4} --infile "${input.name}" + #set $named_input = re.sub(r'(\s|\(|\)|:|!)', '_', str($input.element_identifier)+".fasta") + ln -s "$input" $named_input && + + mob_recon --num_threads \${GALAXY_SLOTS:-4} --infile "${named_input}" #if str($adv_param.unicycler_contigs) == "True": --unicycler_contigs #end if @@ -17,7 +21,16 @@ #if str($adv_param.min_length_condition.min_length_param) == "True": --min_length ${adv_param.min_length_condition.min_length_value} #end if - --run_typer --outdir '.'; + --run_typer --min_rep_evalue '${adv_param.min_rep_evalue}' + --min_rep_evalue '${adv_param.min_rep_evalue}' + --min_mob_evalue '${adv_param.min_mob_evalue}' + --min_con_evalue '${adv_param.min_con_evalue}' + --min_rep_ident '${adv_param.min_rep_ident}' + --min_mob_ident '${adv_param.min_mob_ident}' + --min_con_ident '${adv_param.min_con_ident}' + --min_rpp_ident '${adv_param.min_rpp_ident}' + --outdir '.' && + mkdir ./sequences && (cp plasmid*.fasta chromosome.fasta ./sequences 2> /dev/null || true) ]]> </command> <inputs> @@ -32,7 +45,7 @@ <option value="False">No</option> </param> <conditional name="min_length_condition"> - <param name="min_length_param" label="Minimum length of contigs to classify" type="select" value="False"> + <param name="min_length_param" label="Minimum length of contigs to process" type="select" value="False"> <option value="False">No</option> <option value="True">Yes</option> </param> @@ -41,22 +54,28 @@ </when> <when value="False"/> </conditional> + <param name="min_rep_evalue" label="Minimum evalue threshold for replicon blastn" type="float" min="0.00001" max="1" value="0.00001"/> + <param name="min_mob_evalue" label="Minimum evalue threshold for relaxase tblastn" type="float" min="0.00001" max="1" value="0.00001"/> + <param name="min_con_evalue" label="Minimum evalue threshold for contig blastn" type="float" min="0.00001" max="1" value="0.00001"/> + <param name="min_rpp_evalue" label="Minimum evalue threshold for repetitve elements blastn" type="float" min="0.00001" max="1" value="0.00001"/> + <param name="min_rep_ident" label="Minimum sequence identity for replicons" type="integer" min="0" max="100" value="80"/> + <param name="min_mob_ident" label="Minimum sequence identity for relaxases" type="integer" min="0" max="100" value="80"/> + <param name="min_con_ident" label="Minimum sequence identity for contigs" type="integer" min="0" max="100" value="80"/> + <param name="min_rpp_ident" label="Minimum sequence identity for repetitive elements" type="integer" min="0" max="100" value="80"/> </section> </inputs> <outputs> - <data name="outfile1" format="tabular" from_work_dir="contig_report.txt" label="${tool.name}: Overall contig MOB-recon report"/> - <data name="outfile2" format="tabular" from_work_dir="repetitive_blast_report.txt" label="${tool.name}: Repetitive elements BLAST report"/> - <data name="outfile3" format="fasta" from_work_dir="chromosome.fasta" label="${tool.name}: Chromosomal sequences"/> - <data name="outfile4" format="tabular" from_work_dir="mobtyper_aggregate_report.txt" label="${tool.name}: Aggregate MOB-typer report for all contigs"/> - <data name="outfile5" format="fasta" label="${tool.name}: Plasmid sequence" hidden="true"> - <discover_datasets pattern="plasmid_.+\.(?P<ext>.+)" ext="fasta" visible="true" assign_primary_output="false"/> - </data> + <data name="outfile1" format="tabular" from_work_dir="contig_report.txt" label="${tool.name} on ${on_string}: Overall contig MOB-recon report"/> + <data name="outfile2" format="tabular" from_work_dir="repetitive_blast_report.txt" label="${tool.name} on ${on_string}: Repetitive elements BLAST report"/> + <data name="outfile3" format="tabular" from_work_dir="mobtyper_aggregate_report.txt" label="${tool.name} on ${on_string}: Aggregate MOB-typer report for all contigs"/> + <collection name="seqhits" type="list" label="${tool.name} on ${on_string}: Extracted sequences (plasmids,chromosome(s))"> + <discover_datasets pattern="__name_and_ext__" directory="sequences" /> + </collection> </outputs> <tests> <test> <param name="input" value="plasmid_476.fasta" ftype="fasta"/> <section name="adv_param"> - <param name="evalue" value="0.00001"/> <param name="unicycler_contigs" value="True"/> <param name="run_circlator" value="True"/> </section> @@ -75,6 +94,10 @@ For more information please visit https://github.com/phac-nml/mob-suite/. +**Workflow** + + There is a preliminary Mobilome and Resistome Analysis Workflow linking mob_recon with staramr providing reports on mobilome and resistome https://github.com/phac-nml/mob-suite/. + ----- **Input:**
--- a/mob_typer.xml Thu Jul 26 14:50:59 2018 -0400 +++ b/mob_typer.xml Fri Aug 17 11:12:08 2018 -0400 @@ -1,70 +1,36 @@ -<tool id="mob_typer" name="MOB-Typer" version="1.4.5"> +<tool id="mob_typer" name="MOB-Typer" version="1.4.8"> <description>Get the plasmid type and mobility given its sequence</description> <requirements> - <requirement type="package" version="1.4.5">mob_suite</requirement> + <requirement type="package" version="1.4.8">mob_suite</requirement> </requirements> <command detect_errors="exit_code"> <![CDATA[ ln -s "${input}" "${input.name}"; mob_typer --num_threads \${GALAXY_SLOTS:-4} --infile "${input.name}" - #if str($adv_param.min_rep_evalue_condition.min_rep_evalue_param) == "True": - --min_rep_evalue $adv_param.min_rep_evalue_condition.min_rep_evalue_value - #end if - #if str($adv_param.min_mob_evalue_condition.min_mob_evalue_param) == "True": - --min_mob_evalue $adv_param.min_mob_evalue_condition.min_mob_evalue_value - #end if - #if str($adv_param.min_con_evalue_condition.min_con_evalue_param) == "True": - --min_con_evalue $adv_param.min_con_evalue_condition.min_con_evalue_value - #end if - #if str($adv_param.min_ori_evalue_condition.min_ori_evalue_param) == "True": - --min_ori_evalue $adv_param.min_ori_evalue_condition.min_ori_evalue_value - #end if + --min_rep_evalue '$adv_param.min_rep_evalue_value' + --min_mob_evalue '$adv_param.min_mob_evalue_value' + --min_con_evalue '$adv_param.min_con_evalue_value' + --min_ori_evalue '$adv_param.min_ori_evalue_value' + --min_mpf_evalue '$adv_param.min_mpf_evalue' + --min_rep_ident '$adv_param.min_rep_ident' + --min_mob_ident '$adv_param.min_mob_ident' + --min_ori_ident '$adv_param.min_ori_ident' + --min_mpf_ident '$adv_param.min_mpf_ident' --outdir '.'; ]]> </command> <inputs> <param name="input" type="data" format="fasta" label="Input" help="FASTA file with contig(s)"/> <section name="adv_param" title="Advanced parameters" expanded="False"> - <conditional name="min_rep_evalue_condition"> - <param name="min_rep_evalue_param" label="Minimum evalue threshold for replicon blastn" type="select" value="False"> - <option value="False">No</option> - <option value="True">Yes</option> - </param> - <when value="True"> - <param name="min_rep_evalue_value" type="float" value="0.00001" min="0" max="1" label="e-value (default: 0.00001)"/> - </when> - <when value="False"/> - </conditional> - <conditional name="min_mob_evalue_condition"> - <param name="min_mob_evalue_param" label="Minimum evalue threshold for relaxase tblastn" type="select" value="False"> - <option value="False">No</option> - <option value="True">Yes</option> - </param> - <when value="True"> - <param name="min_mob_evalue_value" type="float" value="0.00001" min="0.00001" max="1" label="e-value (default: 0.00001)"/> - </when> - <when value="False"/> - </conditional> - <conditional name="min_con_evalue_condition"> - <param name="min_con_evalue_param" label="Minimum evalue threshold for contig blastn" type="select" value="False"> - <option value="False">No</option> - <option value="True">Yes</option> - </param> - <when value="True"> - <param name="min_con_evalue_value" type="float" value="0.00001" min="0" max="1" label="e-value (default: 0.00001)"/> - </when> - <when value="False"/> - </conditional> - <conditional name="min_ori_evalue_condition"> - <param name="min_ori_evalue_param" label="Minimum evalue threshold for oriT elements blastn" type="select" value="False"> - <option value="False">No</option> - <option value="True">Yes</option> - </param> - <when value="True"> - <param name="min_ori_evalue_value" type="float" value="0.00001" min="0" max="1" label="e-value (default: 0.00001)"/> - </when> - <when value="False"/> - </conditional> + <param name="min_rep_evalue_value" type="float" value="0.00001" min="0.00001" max="1" label="Minimum evalue threshold for replicon blastn"/> + <param name="min_mob_evalue_value" type="float" value="0.00001" min="0.00001" max="1" label="Minimum evalue threshold for relaxase tblastn"/> + <param name="min_con_evalue_value" type="float" value="0.00001" min="0.00001" max="1" label="Minimum evalue threshold for contig blastn"/> + <param name="min_ori_evalue_value" type="float" value="0.00001" min="0.00001" max="1" label="Minimum evalue threshold for oriT elements blastn"/> + <param name="min_mpf_evalue" type="float" value="0.00001" min="0.00001" max="1" label="Minimum evalue threshold for mpf elements blastn"/> + <param name="min_rep_ident" label="Minimum sequence identity for replicons" type="integer" min="0" max="100" value="80"/> + <param name="min_mob_ident" label="Minimum sequence identity for relaxases" type="integer" min="0" max="100" value="80"/> + <param name="min_ori_ident" label="Minimum sequence identity for oriT elements" type="integer" min="0" max="100" value="90"/> + <param name="min_mpf_ident" label="Minimum sequence identity for mpf elements" type="integer" min="0" max="100" value="80"/> </section> </inputs> <outputs>
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/workflows/AMRworkflow_STARAMR.ga Fri Aug 17 11:12:08 2018 -0400 @@ -0,0 +1,165 @@ +{ + "a_galaxy_workflow": "true", + "annotation": "", + "format-version": "0.1", + "name": "AMRworkflow", + "steps": { + "0": { + "annotation": "", + "content_id": "mob_recon", + "errors": null, + "id": 0, + "input_connections": {}, + "inputs": [ + { + "description": "runtime parameter for tool MOB-Recon", + "name": "input" + } + ], + "label": null, + "name": "MOB-Recon", + "outputs": [ + { + "name": "seqhits", + "type": "input" + }, + { + "name": "outfile1", + "type": "tabular" + }, + { + "name": "outfile2", + "type": "tabular" + }, + { + "name": "outfile3", + "type": "tabular" + } + ], + "position": { + "left": 200, + "top": 200 + }, + "post_job_actions": { + "HideDatasetActionoutfile1": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile1" + }, + "HideDatasetActionoutfile2": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile2" + }, + "HideDatasetActionoutfile3": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "outfile3" + } + }, + "tool_id": "mob_recon", + "tool_state": "{\"adv_param\": \"{\\\"min_length_condition\\\": {\\\"min_length_param\\\": \\\"False\\\", \\\"__current_case__\\\": 1}, \\\"min_con_ident\\\": \\\"80\\\", \\\"min_mob_ident\\\": \\\"80\\\", \\\"min_mob_evalue\\\": \\\"1e-05\\\", \\\"min_rep_evalue\\\": \\\"1e-05\\\", \\\"unicycler_contigs\\\": \\\"True\\\", \\\"min_rep_ident\\\": \\\"80\\\", \\\"min_rpp_evalue\\\": \\\"1e-05\\\", \\\"min_rpp_ident\\\": \\\"80\\\", \\\"run_circlator\\\": \\\"True\\\", \\\"min_con_evalue\\\": \\\"1e-05\\\"}\", \"input\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\", \"__rerun_remap_job_id__\": null, \"chromInfo\": \"\\\"/galaxy/tool-data/shared/ucsc/chrom/?.len\\\"\", \"__page__\": null}", + "tool_version": "1.4.8", + "type": "tool", + "uuid": "017d5861-8836-46a8-8444-0dcaca60eda1", + "workflow_outputs": [ + { + "label": null, + "output_name": "seqhits", + "uuid": "6304bc2b-bcc9-47c4-8e3b-9d2ee5a80433" + } + ] + }, + "1": { + "annotation": "", + "content_id": "staramr_search", + "errors": null, + "id": 1, + "input_connections": { + "genomes": { + "id": 0, + "output_name": "seqhits" + } + }, + "inputs": [ + { + "description": "runtime parameter for tool staramr", + "name": "genomes" + } + ], + "label": null, + "name": "staramr", + "outputs": [ + { + "name": "blast_hits", + "type": "input" + }, + { + "name": "summary", + "type": "tabular" + }, + { + "name": "resfinder", + "type": "tabular" + }, + { + "name": "pointfinder", + "type": "tabular" + }, + { + "name": "settings", + "type": "txt" + }, + { + "name": "excel", + "type": "xlsx" + } + ], + "position": { + "left": 420, + "top": 200 + }, + "post_job_actions": { + "HideDatasetActionpointfinder": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "pointfinder" + }, + "HideDatasetActionsettings": { + "action_arguments": {}, + "action_type": "HideDatasetAction", + "output_name": "settings" + } + }, + "tool_id": "staramr_search", + "tool_state": "{\"__page__\": null, \"use_pointfinder\": \"{\\\"enable\\\": \\\"false\\\", \\\"__current_case__\\\": 1}\", \"__rerun_remap_job_id__\": null, \"genomes\": \"{\\\"__class__\\\": \\\"RuntimeValue\\\"}\", \"chromInfo\": \"\\\"/galaxy/tool-data/shared/ucsc/chrom/?.len\\\"\", \"advanced\": \"{\\\"pid_threshold\\\": \\\"98.0\\\", \\\"exclude_resistance_phenotypes\\\": \\\"false\\\", \\\"exclude_negatives\\\": \\\"false\\\", \\\"plength_resfinder\\\": \\\"60.0\\\", \\\"plength_pointfinder\\\": \\\"95.0\\\", \\\"report_all_blast\\\": \\\"false\\\"}\"}", + "tool_version": "0.2.2", + "type": "tool", + "uuid": "41f8718f-87a0-45d2-b79d-e5afddea1532", + "workflow_outputs": [ + { + "label": null, + "output_name": "blast_hits", + "uuid": "9e8592ad-71c0-4528-8e8b-efcd3bb9183a" + }, + { + "label": null, + "output_name": "resfinder", + "uuid": "2f27a46b-29bf-4abf-9d26-fce4e9fc9dd5" + }, + { + "label": null, + "output_name": "excel", + "uuid": "67549d0b-0480-4c25-8bfd-1fa78cfdb066" + }, + { + "label": null, + "output_name": "summary", + "uuid": "17c9e48c-234a-4cb7-986d-f88166bfe0a7" + } + ] + } + }, + "tags": [], + "uuid": "4c695847-5fc7-4020-a125-a500700ab320" +}