jpayne@68: <tool id="csp2-snp" name="CSP2 (SNP Pipeline Mode)" version="0.9.7_Dev21">
jpayne@68:         <description>Run SNP Pipeline analysis on isolates using one or more references.</description>
jpayne@68:         <requirements>
jpayne@68:                 <requirement type="package" version="24.10.1">nextflow</requirement>
jpayne@68:                 <requirement type="package" version="1.5.8">micromamba</requirement>
jpayne@68:         </requirements>
jpayne@68:         <version_command>nextflow -version</version_command>
jpayne@68:         <command detect_errors="aggressive"><![CDATA[
jpayne@68: mkdir -p queries references
jpayne@68: #set readext=""
jpayne@68: #for $reads in $query.coll
jpayne@68: #if $hasattr($reads, "is_of_type") and $reads.is_of_type("fasta")
jpayne@68:          && ln -sf '${reads}' 'queries/${reads.element_identifier}.fasta'
jpayne@68: #else if $reads.forward.is_of_type("fastq.gz","fastqsanger.gz")
jpayne@68:          #set readext="fastq.gz"
jpayne@68:          && ln -sf '${reads.forward}' 'queries/${reads.forward.element_identifier}_1.fastq.gz'
jpayne@68:          && ln -sf '${reads.reverse}' 'queries/${reads.reverse.element_identifier}_2.fastq.gz'
jpayne@68: #else if $reads.forward.is_of_type("fastq.bz2", "fastqsanger.bz2")
jpayne@68:          #set readext="fastq.bz2"
jpayne@68:          && ln -sf '${reads.forward}' 'queries/${reads.forward.element_identifier}_1.fastq.bz2'
jpayne@68:          && ln -sf '${reads.reverse}' 'queries/${reads.reverse.element_identifier}_2.fastq.bz2'
jpayne@68: #else 
jpayne@68:          #set readext="fastq"
jpayne@68:          && ln -sf '${reads.forward}' 'queries/${reads.forward.element_identifier}_1.fastq'
jpayne@68:          && ln -sf '${reads.reverse}' 'queries/${reads.reverse.element_identifier}_2.fastq'
jpayne@68: #end if
jpayne@68: #end for
jpayne@68:         && echo "*** Files in queries directory: ***"
jpayne@68:         && ln -sf '$source.reference' 'references/${source.reference.element_identifier}.fasta'
jpayne@68:         && nextflow run 
jpayne@68:         ${__tool_directory__}/CSP2/CSP2.nf -profile csp2_galaxy 
jpayne@68:         --runmode snp 
jpayne@68:         #if $query.query_select == 'reads'
jpayne@68:         --reads queries
jpayne@68:         --readext $readext 
jpayne@68:         --forward "_1.${readext}"
jpayne@68:         --reverse "_2.${readext}"
jpayne@68:         #else
jpayne@68:         --fasta queries
jpayne@68:         #end if
jpayne@68:         --ref_fasta 'references/${source.reference.element_identifier}.fasta'
jpayne@68:         --min_cov $opt.min_cov 
jpayne@68:         --min_iden $opt.min_iden 
jpayne@68:         --min_len $opt.min_len 
jpayne@68:         --ref_edge $opt.ref_edge 
jpayne@68:         --query_edge $opt.query_edge 
jpayne@68:         --dwin $opt.dwin 
jpayne@68:         --wsnps $opt.wsnps 
jpayne@68:         --out CSP2_SNP_Output
jpayne@68:         && echo "*** Files in output directory: ***"
jpayne@68:         && ls -lah CSP2_SNP_Output
jpayne@68:         && echo "*** Nextflow log follows: ***"
jpayne@68:         && cat .nextflow.log
jpayne@68: ]]>
jpayne@68:         </command>
jpayne@68:         <inputs>
jpayne@68:                 <conditional name="source">
jpayne@68:                         <param name="source_select" type="select" label="Use a curated GalaxyTrakr reference or a reference from your history">
jpayne@68:                                 <option value="curated">Use a GalaxyTrakr reference</option>
jpayne@68:                                 <option value="history">Use a reference from your history</option>
jpayne@68:                         </param>
jpayne@68:                         <when value="curated">
jpayne@68:                                 <param name="reference" type="select" label="Select reference fasta">
jpayne@68:                                         <options from_data_table="all_fasta">
jpayne@68:                                                 <filter type="sort_by" column="2"/>
jpayne@68:                                                 <validator type="no_options" message="No assemblies are available for the selected input dataset"/>
jpayne@68:                                         </options>
jpayne@68:                                 </param>
jpayne@68:                         </when>
jpayne@68:                         <when value="history">
jpayne@68:                                 <param type="data" name="reference" format="fasta" label="Select reference FASTA"/>
jpayne@68:                         </when>
jpayne@68:                 </conditional>
jpayne@68:                 <conditional name="query">
jpayne@68:                         <param name="query_select" type="select" label="Screen a list of paired-reads or a list of assemblies">
jpayne@68:                                 <option value="reads">Screen a list of paired reads</option>
jpayne@68:                                 <option value="assemblies">Screen a list of assemblies</option>
jpayne@68:                         </param>
jpayne@68:                         <when value="reads">
jpayne@68:                                 <param label="Paired reads" name="coll" type="data_collection" format="fastq,fastqsanger,fastq.gz,fastqsanger.gz,fastq.bz2,fastqsanger.bz2" collection_type="list:paired" />
jpayne@68:                         </when>
jpayne@68:                         <when value="assemblies">
jpayne@68:                                 <param label="Assemblies" name="coll" type="data_collection" format="fasta" collection_type="list" />
jpayne@68:                         </when>
jpayne@68:                 </conditional>
jpayne@68:                 <section name="opt" title="Advanced options...">
jpayne@68:                         <param name="min_cov" type="float" value="85" label="Minimum reference genome coverage to proceed with distance estimation" optional="true" />
jpayne@68:                         <param name="min_iden" type="float" value="99" label="Minimum alignment percent identity to detect SNPs" optional="true" />
jpayne@68:                         <param name="min_len" type="integer" value="500" label="Minimum alignment length to detect SNPs" optional="true" />
jpayne@68:                         <param name="ref_edge" type="integer" value="150" label="Prune SNPs within this many bases of reference contig edge" optional="true" />
jpayne@68:                         <param name="query_edge" type="integer" value="150" label="Prune SNPs within this many bases of query contig edge" optional="true" />
jpayne@68:                         <param 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)" optional="true" />
jpayne@68:                         <param name="wsnps" type="text" value="3,2,1" label="Comma-separated list of maximum SNP counts per density window" optional="true" />
jpayne@68:                 </section>
jpayne@68:         </inputs>
jpayne@68:         <outputs>
jpayne@68:                 <!-- <data name="nextflow_log" format="txt" label="Nextflow Log" from_work_dir="Nextflow_Log.txt" /> -->
jpayne@68:                 <data name="isolate_data" format="tabular" label="Isolate Data" from_work_dir="CSP2_SNP_Output/Isolate_Data.tsv" />
jpayne@68:                 <data name="raw_mummer" format="tabular" label="Raw MUMmer Output" from_work_dir="CSP2_SNP_Output/Raw_MUMmer_Summary.tsv" />
jpayne@68:                 <!-- <data name="csp2_zip" format="zip" label="Zipped Output" from_work_dir="CSP2_Output.zip" /> -->
jpayne@68:                 <collection name="reference_results" type="list:list">
jpayne@68:                         <discover_datasets pattern="(?P&lt;name&gt;.+)/CSP2_SNP_Pipeline\.log" format="txt" visible="true" directory="./CSP2_SNP_Output/SNP_Analysis" />
jpayne@68:                         <discover_datasets pattern="(?P&lt;name&gt;.+)/Reference_Screening\.tsv" format="tabular" visible="true" directory="./CSP2_SNP_Output/SNP_Analysis" />
jpayne@68:                         <discover_datasets pattern="(?P&lt;name&gt;.+)/snp_distance_matrix_preserved\.tsv" format="tabular" visible="true" directory="./CSP2_SNP_Output/SNP_Analysis" />
jpayne@68:                         <discover_datasets pattern="(?P&lt;name&gt;.+)/snp_distance_pairwise_preserved\.tsv" format="tabular" visible="true" directory="./CSP2_SNP_Output/SNP_Analysis" />
jpayne@68:                         <discover_datasets pattern="(?P&lt;name&gt;.+)/snpma_preserved\.fasta" format="fasta" visible="true" directory="./CSP2_SNP_Output/SNP_Analysis" />
jpayne@68:                 </collection>
jpayne@68:         </outputs>
jpayne@68:         <tests>
jpayne@68:                 <test>
jpayne@68:                         <param name="source_select" value="history" />
jpayne@68:                         <param name="reference" value="assemblies/Sample_A.fasta" />
jpayne@68:                         <param name="query_select" value="assemblies" />
jpayne@68:                         <param name="coll">
jpayne@68:                                 <collection type="list">
jpayne@68:                                         <!-- <element name="Sample_A" value="assemblies/Sample_A.fasta" /> -->
jpayne@68:                                         <element name="Sample_B" value="assemblies/Sample_B.fasta" />
jpayne@68:                                         <element name="Sample_C" value="assemblies/Sample_C.fasta" />
jpayne@68:                                         <element name="Sample_D" value="assemblies/Sample_D.fasta" />
jpayne@68:                                         <element name="Sample_E" value="assemblies/Sample_E.fasta" />
jpayne@68:                                         <element name="Sample_F" value="assemblies/Sample_F.fasta" />
jpayne@68:                                         <element name="Sample_G" value="assemblies/Sample_G.fasta" />
jpayne@68:                                         <element name="Sample_H" value="assemblies/Sample_H.fasta" />
jpayne@68:                                         <element name="Sample_I" value="assemblies/Sample_I.fasta" />
jpayne@68:                                         <element name="Sample_J" value="assemblies/Sample_J.fasta" />
jpayne@68:                                         <element name="Sample_K" value="assemblies/Sample_K.fasta" />
jpayne@68:                                         <element name="Sample_L" value="assemblies/Sample_L.fasta" />
jpayne@68:                                         <element name="Sample_M" value="assemblies/Sample_M.fasta" />
jpayne@68:                                         <element name="Sample_N" value="assemblies/Sample_N.fasta" />
jpayne@68:                                         <element name="Sample_O" value="assemblies/Sample_O.fasta" />
jpayne@68:                                 </collection>
jpayne@68:                         </param>
jpayne@68: 
jpayne@68:                         <output name="isolate_data" value="Isolate_Data.tsv" />
jpayne@68:                 </test>
jpayne@68:                 <test>
jpayne@68:                         <param name="source_select" value="history" />
jpayne@68:                         <param name="reference" value="assemblies/Sample_A.fasta" ftype="fasta" />
jpayne@68:                         <param name="query_select" value="reads" />
jpayne@68:                         <param name="coll">
jpayne@68:                                 <collection type="list:paired">
jpayne@68:                                         <element name="Sample_A">
jpayne@68:                                                 <collection type="paired">
jpayne@68:                                                         <element name="forward" value="reads/Week_42_Reads_1.fq.gz" ftype="fastqsanger.gz" />
jpayne@68:                                                         <element name="reverse" value="reads/Week_42_Reads_2.fq.gz" ftype="fastqsanger.gz" />
jpayne@68:                                                 </collection>
jpayne@68:                                         </element>
jpayne@68:                                 </collection>
jpayne@68:                         </param>
jpayne@68:                         <output name="screening_results" value="Screening_Results.tsv" />
jpayne@68:                         <output name="isolate_data" value="Isolate_Data.tsv" />
jpayne@68:                 </test>
jpayne@68:         </tests>
jpayne@68:         <help>
jpayne@68:         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.
jpayne@68:         </help>
jpayne@68:         <citations>
jpayne@68:                 <citation type="doi">10.XXXX/placeholder.doi</citation>
jpayne@68:                 <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}}
jpayne@68:                 </citation>
jpayne@68:         </citations>
jpayne@68: </tool>