comparison csp2-screen.xml @ 69:33d812a61356

planemo upload commit 2e9511a184a1ca667c7be0c6321a36dc4e3d116d
author jpayne
date Tue, 18 Mar 2025 17:55:14 -0400
parents
children
comparison
equal deleted inserted replaced
67:0e9998148a16 69:33d812a61356
1 <tool id="csp2-screen" name="CSP2 (Screening Mode)" version="0.9.7_Dev21">
2 <description>Screen query assemblies against reference assemblies</description>
3 <requirements>
4 <!-- <requirement type="package" version="24.10.1">nextflow</requirement>
5 <requirement type="package" version="1.5.8">micromamba</requirement> -->
6 <container type="docker">cfsanbiostatistics/csp2:v.0.9.7.4</container>
7 </requirements>
8 <version_command>nextflow -version</version_command>
9 <command detect_errors="exit_code"><![CDATA[
10 mkdir -p queries references
11 #set readext=""
12 #for $reads in $query.coll
13 #if $hasattr($reads, "is_of_type") and $reads.is_of_type("fasta")
14 && ln -sf ${reads} 'queries/${reads.element_identifier}.fasta'
15 #else if $reads.forward.is_of_type("fastq.gz","fastqsanger.gz")
16 #set readext="fastq.gz"
17 && ln -sf '${reads.forward}' 'queries/${reads.forward.element_identifier}_1.fastq.gz'
18 && ln -sf '${reads.reverse}' 'queries/${reads.reverse.element_identifier}_2.fastq.gz'
19 #else if $reads.forward.is_of_type("fastq.bz2", "fastqsanger.bz2")
20 #set readext="fastq.bz2"
21 && ln -sf '${reads.forward}' 'queries/${reads.forward.element_identifier}_1.fastq.bz2'
22 && ln -sf '${reads.reverse}' 'queries/${reads.reverse.element_identifier}_2.fastq.bz2'
23 #else
24 #set readext="fastq"
25 && ln -sf '${reads.forward}' 'queries/${reads.forward.element_identifier}_1.fastq'
26 && ln -sf '${reads.reverse}' 'queries/${reads.reverse.element_identifier}_2.fastq'
27 #end if
28 #end for
29 && echo "*** Files in queries directory: ***"
30 && ls -lah queries/
31 && ln -sf '$source.reference' 'references/${source.reference.element_identifier}.fasta'
32 && nextflow run
33 ${__tool_directory__}/CSP2/CSP2.nf -profile csp2_galaxy
34 --runmode screen
35 #if $query.query_select == 'reads'
36 --reads queries
37 --readext $readext
38 --forward "_1.${readext}"
39 --reverse "_2.${readext}"
40 #else
41 --fasta queries
42 #end if
43 --ref_fasta 'references/${source.reference.element_identifier}.fasta'
44 --min_cov $opt.min_cov
45 --min_iden $opt.min_iden
46 --min_len $opt.min_len
47 --ref_edge $opt.ref_edge
48 --query_edge $opt.query_edge
49 --dwin $opt.dwin
50 --wsnps $opt.wsnps
51 --out CSP2_Screen_Output
52 && echo "*** Files in output directory: ***"
53 && ls -lah CSP2_Screen_Output
54 && echo "*** Nextflow log follows: ***"
55 && cat .nextflow.log
56 ]]>
57 </command>
58 <inputs>
59 <conditional name="source">
60 <param name="source_select" type="select" label="Use a curated GalaxyTrakr reference or a reference from your history">
61 <option value="curated">Use a GalaxyTrakr reference</option>
62 <option value="history">Use a reference from your history</option>
63 </param>
64 <when value="curated">
65 <param name="reference" type="select" label="Select reference fasta">
66 <options from_data_table="all_fasta">
67 <filter type="sort_by" column="2"/>
68 <validator type="no_options" message="No assemblies are available for the selected input dataset"/>
69 </options>
70 </param>
71 </when>
72 <when value="history">
73 <param type="data" name="reference" format="fasta" label="Select reference FASTA"/>
74 </when>
75 </conditional>
76 <conditional name="query">
77 <param name="query_select" type="select" label="Screen a list of paired-end reads or a list of assemblies">
78 <option value="reads">Screen a list of paired reads</option>
79 <option value="assemblies">Screen a list of assemblies</option>
80 </param>
81 <when value="reads">
82 <param label="Paired reads" name="coll" type="data_collection" format="fastq,fastqsanger,fastq.gz,fastqsanger.gz,fastq.bz2,fastqsanger.bz2" collection_type="list:paired" />
83 </when>
84 <when value="assemblies">
85 <param label="Assemblies" name="coll" type="data_collection" format="fasta" collection_type="list" />
86 </when>
87 </conditional>
88 <section name="opt" title="Advanced options...">
89 <param argument="--min_cov" name="min_cov" type="float" value="85" label="Minimum reference genome coverage to proceed with distance estimation" />
90 <param argument="--min_eden" name="min_iden" type="float" value="99" label="Minimum alignment percent identity to detect SNPs" />
91 <param argument="--min_len" name="min_len" type="integer" value="500" label="Minimum alignment length to detect SNPs" />
92 <param argument="--ref_edge" name="ref_edge" type="integer" value="150" label="Prune SNPs within this many bases of reference contig edge" />
93 <param argument="--query_edge" name="query_edge" type="integer" value="150" label="Prune SNPs within this many bases of query contig edge" />
94 <param argument="--dwin" name="dwin" type="text" value="1000,125,15" label="Comma-separated set of window sizes for SNP density filtration (Set to 0 to disable density filtration)" />
95 <param argument="--wsnips" name="wsnps" type="text" value="3,2,1" label="Comma-separated list of maximum SNP counts per density window" />
96 </section>
97 </inputs>
98 <outputs>
99 <data name="raw_mummer" format="tabular" label="Raw MUMmer Output" from_work_dir="CSP2_Screen_Output/Raw_MUMmer_Summary.tsv" />
100 <data name="isolate_data" format="tabular" label="Isolate Data" from_work_dir="CSP2_Screen_Output/Isolate_Data.tsv" />
101 <data name="screening_results" format="tabular" label="Screening Results" from_work_dir="CSP2_Screen_Output/Screening_Results.tsv" />
102 <!-- <data name="nextflow_log" format="txt" label="Nextflow Log" from_work_dir="Nextflow_Log.txt" /> -->
103 </outputs>
104 <tests>
105 <test>
106 <param name="source_select" value="history" />
107 <param name="reference" value="assemblies/Sample_A.fasta" ftype="fasta" />
108 <param name="query_select" value="assemblies" />
109 <param name="coll">
110 <collection type="list">
111 <!-- <element name="Sample_A" value="assemblies/Sample_A.fasta" /> -->
112 <element name="Sample_B" value="assemblies/Sample_B.fasta" />
113 <element name="Sample_C" value="assemblies/Sample_C.fasta" />
114 <element name="Sample_D" value="assemblies/Sample_D.fasta" />
115 <element name="Sample_E" value="assemblies/Sample_E.fasta" />
116 <element name="Sample_F" value="assemblies/Sample_F.fasta" />
117 <element name="Sample_G" value="assemblies/Sample_G.fasta" />
118 <element name="Sample_H" value="assemblies/Sample_H.fasta" />
119 <element name="Sample_I" value="assemblies/Sample_I.fasta" />
120 <element name="Sample_J" value="assemblies/Sample_J.fasta" />
121 <element name="Sample_K" value="assemblies/Sample_K.fasta" />
122 <element name="Sample_L" value="assemblies/Sample_L.fasta" />
123 <element name="Sample_M" value="assemblies/Sample_M.fasta" />
124 <element name="Sample_N" value="assemblies/Sample_N.fasta" />
125 <element name="Sample_O" value="assemblies/Sample_O.fasta" />
126 </collection>
127 </param>
128
129 <output name="screening_results" value="Screening_Results.tsv" />
130 <output name="isolate_data" value="Isolate_Data.tsv" />
131 </test>
132 <test>
133 <param name="source_select" value="history" />
134 <param name="reference" value="assemblies/Sample_A.fasta" ftype="fasta" />
135 <param name="query_select" value="reads" />
136 <param name="coll">
137 <collection type="list:paired">
138 <element name="Sample_A" >
139 <collection type="paired">
140 <element name="forward" value="reads/Week_42_Reads_1.fq.gz" ftype="fastqsanger.gz" />
141 <element name="reverse" value="reads/Week_42_Reads_2.fq.gz" ftype="fastqsanger.gz" />
142 </collection>
143 </element>
144 </collection>
145 </param>
146 <output name="screening_results" value="Screening_Results.tsv" />
147 <output name="isolate_data" value="Isolate_Data.tsv" />
148 </test>
149 </tests>
150 <help>
151 This tool takes query assemblies and reference assemblies and calculates the pairwise distance between each query/reference combination. If no reference is provided, all queries are compared to all other queries.
152 </help>
153 <citations>
154 <citation type="doi">10.XXXX/placeholder.doi</citation>
155 <citation type="bibtex">@article{example2024,title={CFSAN SNP Pipeline 2 (CSP2): a pipeline for fast and accurate SNP distance estimation from bacterial genome assemblies.},author={Doe, John and Smith, Jane},journal={Submitted},year={2024}}
156 </citation>
157 </citations>
158 </tool>