Mercurial > repos > iuc > ncbi_fcs_gx

comparison ncbi_fcs_gx.xml @ 0:3cdb96f2855d draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/ncbi_fcs_gx commit 4a6561ed00e004260be3f3c29d81e814c60e20af
author iuc
date Fri, 12 Jan 2024 22:11:39 +0000
parents
children 49f8eae39606
comparison
equal deleted inserted replaced
-1:000000000000 0:3cdb96f2855d
1 <tool id="ncbi_fcs_gx" name="NCBI FCS GX" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@">
2 <description>detects contamination from foreign organisms in genome sequences</description>
3 <macros>
4 <import>macros.xml</import>
5 </macros>
6 <expand macro="edam_ontology"/>
7 <expand macro="requirements"/>
8 <command detect_errors="exit_code"><![CDATA[
9 GX_NUM_CORES=\${GALAXY_SLOTS:-2}
10 #if $mode.mode_selector == "screen"
11 ## copy data to local storage
12 #set manifest_pathname = $mode.screen_adv.database.fields.name
13 #if $mode.config_tag.fields.use_source_manifest == "1"
14 #set manifest_pathname = $mode.screen_adv.database.fields.source_manifest
15 #end if
16 mkdir -p '$mode.config_tag.fields.node_cache_dir' &&
17 sync_files.py get --mft '$manifest_pathname' --dir '$mode.config_tag.fields.node_cache_dir' > /dev/null 2>&1 &&
18 ## run gx
19 run_gx.py
20 #if $mode.config_tag.fields.phone_home == "1"
21 --phone-home-label '$mode.config_tag.fields.phone_home_label'
22 #end if
23 --fasta '$mode.fasta'
24 #if $mode.id.id_selector == "gx_div"
25 --tax-id '1'
26 --div '$mode.id.div'
27 #else
28 --tax-id '$mode.id.tax_id'
29 #end if
30 #if $mode.species != ""
31 --species '$mode.species'
32 #end if
33 --split-fasta '$mode.screen_adv.split_fasta'
34 #if $mode.screen_adv.div:
35 --div '$mode.screen_adv.div'
36 #end if
37 --gx-db '$mode.config_tag.fields.node_cache_dir'
38 --out-basename output
39 --action-report true
40 --generate-logfile false
41 #elif $mode.mode_selector == "clean"
42 ## run gx
43 gx clean-genome
44 --input '$mode.input'
45 --action-report '$mode.action_report'
46 --contam-fasta-out 'contam.fa'
47 --min-seq-len '$mode.min_seq_len'
48 --output 'clean.fa'
49 #end if
50 ]]></command>
51 <environment_variables>
52 <environment_variable name="GX_ALIGN_EXCLUDE_TAXA">$getVar('mode.screen_adv.gx_align_exclude_taxa', '')</environment_variable>
53 <environment_variable name="GX_EXTRA_CONTAM_DIVS"><![CDATA[#echo ','.join($getVar('mode.screen_adv.gx_extra_contam_divs', []))]]></environment_variable>
54 </environment_variables>
55 <inputs>
56 <conditional name="mode">
57 <param name="mode_selector" type="select" label="Choose the mode">
58 <option value="screen" selected="true">Screen genome</option>
59 <option value="clean">Clean genome</option>
60 </param>
61 <when value="screen">
62 <!-- value, name, use_source_manifest, phone_home, phone_home_label, node_cache_dir -->
63 <param name="config_tag" type="select" label="Database">
64 <options from_data_table="ncbi_fcs_gx_config">
65 <filter type="sort_by" name="sorted_description" column="1"/>
66 </options>
67 <validator message="No database is available" type="no_options"/>
68 </param>
69 <param argument="--fasta" type="data" format="fasta" label="Input file (Fasta file)" help="To detect contamination from foreign organisms, a genome assembly in a fasta file."/>
70 <conditional name="id">
71 <param name="id_selector" type="select" label="Taxonomy entry">
72 <option value="gx_div" selected="true">GX Division</option>
73 <option value="ncbi_tax">NCBI Taxonomic identifier</option>
74 </param>
75 <when value="gx_div">
76 <param argument="--div" type="select">
77 <options from_data_table="ncbi_fcs_gx_divisions">
78 <filter type="param_value" ref="config_tag" column="1" />
79 <filter type="sort_by" name="sorted_description" column="2" />
80 </options>
81 <validator message="No GX Divisions are available" type="no_options"/>
82 </param>
83 </when>
84 <when value="ncbi_tax">
85 <!-- https://www.ncbi.nlm.nih.gov/taxonomy -->
86 <param argument="--tax-id" type="text" label="Taxonomic identifier" help="The appropriate tax-id for your genome assembly. The appropriate tax-id for an organism can be retrieved from the NCBI Taxonomy website."/>
87 </when>
88 </conditional>
89 <param argument="--species" type="text" optional="true" label="Species binomial name"/>
90 <section name="screen_adv" title="Advanced options">
91 <!-- comma separated list of taxa to ignore in GX_ALIGN_EXCLUDE_TAXA environment variable -->
92 <param name="gx_align_exclude_taxa" type="text" value="" optional="true" label="Taxonomic identifier(s) to exclude" help="Multiple tax-ids may be provided as a comma-separated list.">
93 <validator type="regex" message="comma separated integers">^\s*\d+\s*(,\s*\d+\s*)*$</validator>
94 <sanitizer invalid_char="">
95 <valid initial="string.digits">
96 <add value=","/>
97 </valid>
98 </sanitizer>
99 </param>
100 <param name="gx_extra_contam_divs" type="select" multiple="true" optional="true" label="Additional contaminants to identify" help="Multiple gx-divisions may be selected.">
101 <options from_data_table="ncbi_fcs_gx_divisions">
102 <filter type="param_value" ref="config_tag" column="1" />
103 <filter type="sort_by" name="sorted_description" column="2" />
104 </options>
105 <validator message="No GX Divisions are available" type="no_options"/>
106 </param>
107 <param argument="--split-fasta" type="boolean" checked="true" optional="true" label="Split fasta sequences on N-runs of length at least 10"/>
108 <param argument="--div" type="text" value="" optional="true" label="BLAST-div of the tax-id" help="from 'NCBI BLAST name' on taxon Info page"/>
109 <param name="database" type="select" label="Database location">
110 <options from_data_table="ncbi_fcs_gx_databases">
111 <filter type="param_value" ref="config_tag" column="0"/>
112 </options>
113 <validator message="No database location is available" type="no_options"/>
114 </param>
115 </section>
116 </when>
117 <when value="clean">
118 <param argument="--input" type="data" format="fasta" label="Input file (Fasta file)" help="To detect contamination from foreign organisms, a genome assembly in a fasta file."/>
119 <param argument="--action-report" type="data" format="tabular" label="Select Action report"/>
120 <param argument="--min-seq-len" type="integer" value="200" label="Minimumm sequence length to keep"/>
121 </when>
122 </conditional>
123 </inputs>
124 <outputs>
125 <!-- mode == screen -->
126 <data name="taxonomy_report" format="tabular" from_work_dir="output.taxonomy.rpt" label="${tool.name} on ${on_string}: Taxonomy report">
127 <filter>mode['mode_selector'] == 'screen'</filter>
128 </data>
129 <data name="action_report" format="tabular" from_work_dir="output.fcs_gx_report.txt" label="${tool.name} on ${on_string}: Action report">
130 <filter>mode['mode_selector'] == 'screen'</filter>
131 </data>
132 <!-- mode == clean -->
133 <data name="contam_fasta" format="fasta" from_work_dir="contam.fa" label="${tool.name} on ${on_string}: Fasta for EXCLUDE entries">
134 <filter>mode['mode_selector'] == 'clean'</filter>
135 </data>
136 <data name="clean_fasta" format="fasta" from_work_dir="clean.fa" metadata_source="mode.input" label="${tool.name} on ${on_string}: Cleaned Fasta">
137 <filter>mode['mode_selector'] == 'clean'</filter>
138 </data>
139 </outputs>
140 <tests>
141 <test expect_num_outputs="2">
142 <param name="mode_selector" value="screen"/>
143 <param name="config_tag" value="test-only" />
144 <param name="id_selector" value="ncbi_tax"/>
145 <param name="fasta" value="fcsgx_test.fa.gz" ftype="fasta"/>
146 <param name="tax_id" value="6973"/>
147 <output name="taxonomy_report" file="output.taxonomy.rpt" compare="diff" lines_diff="2" />
148 <output name="action_report" file="output.fcs_gx_report.txt" compare="diff" lines_diff="2" />
149 </test>
150 <test expect_num_outputs="2">
151 <param name="mode_selector" value="clean"/>
152 <param name="id_selector" value="ncbi_tax"/>
153 <param name="input" value="fcsgx_test.fa.gz" ftype="fasta"/>
154 <param name="action_report" value="output.fcs_gx_report.txt" ftype="tabular"/>
155 <output name="contam_fasta" decompress="true" file="output.contam.fa.gz" ftype="fasta" />
156 <output name="clean_fasta" decompress="true" file="output.clean.fa.gz" ftype="fasta" />
157 </test>
158 </tests>
159 <help><![CDATA[
160 FCS-GX detects contamination from foreign organisms in genome sequences using the genome cross-species aligner (GX). The FCS-GX executable retrieves a Docker or Singularity container and runs a pipeline to align sequences to a large database of NCBI genomes through modified k-mer seeds and assign a most likely taxonomic division.
161
162 FCS-GX classifies sequences as contaminant when their taxonomic assignment is different from the user provided taxonomic identifier. A contamination summary provides an overview of observed contaminant divisions, counts, and total sizes, and an action report provides details and recommended actions for each problematic sequence.
163
164 https://github.com/ncbi/fcs/wiki/FCS-GX
165 ]]></help>
166 <expand macro="citations"/>
167 </tool>