rliterman@41: <tool id="csp2-screen" name="CSP2 (Screening Mode)" version="0.9.7_Dev21">
rliterman@29:         <description>Screen query assemblies against reference assemblies</description>
rliterman@29:         <requirements>
rliterman@29:                 <requirement type="package" version="24.10.1">nextflow</requirement>
rliterman@29:                 <requirement type="package" version="1.5.8">micromamba</requirement>
rliterman@29:         </requirements>
rliterman@29:         <version_command>nextflow -version</version_command>
rliterman@29:         <command detect_errors="aggressive"><![CDATA[
rliterman@42: 
rliterman@46: mkdir ./queries ./references ./output;
rliterman@29: 
rliterman@29: if [ -n "$query_fasta" ] && [ "$query_fasta" != "None" ]; then
rliterman@29:         #for query in $query_fasta:
rliterman@45:                 ln -sf ${query} ./queries/${query.element_identifier};
rliterman@29:         #end for
rliterman@45:         export QUERY_FASTA_ARG="--fasta ./queries";
rliterman@29: else
rliterman@45:         export QUERY_FASTA_ARG="";
rliterman@43: fi;
rliterman@29: 
rliterman@29: if [ -n "$query_reads" ] && [ "$query_reads" != "None" ]; then
rliterman@29:         #for query in $query_reads:
rliterman@45:                 ln -sf ${query} ./queries/${query.element_identifier};
rliterman@29:         #end for
rliterman@45:         export QUERY_READS_ARG="--reads ./queries";
rliterman@29: else
rliterman@45:         export QUERY_READS_ARG="";
rliterman@43: fi;
rliterman@29: 
rliterman@29: if [ -n "$ref_fasta" ] && [ "$ref_fasta" != "None" ]; then
rliterman@29:         #for ref in $ref_fasta:
rliterman@45:                 ln -sf ${ref} ./references/${ref.element_identifier};
rliterman@29:         #end for
rliterman@45:         export REF_FASTA_ARG="--ref_fasta ./references";
rliterman@29: else
rliterman@45:         export REF_FASTA_ARG="";
rliterman@43: fi;
rliterman@29: 
rliterman@29: if [ -n "$ref_reads" ] && [ "$ref_reads" != "None" ]; then
rliterman@29:         #for ref in $ref_reads:
rliterman@45:                 ln -sf ${ref} ./references/${ref.element_identifier};
rliterman@29:         #end for
rliterman@45:         export REF_READS_ARG="--ref_reads ./references";
rliterman@29: else
rliterman@45:         export REF_READS_ARG="";
rliterman@43: fi;
rliterman@29: 
rliterman@29: if [ -n "$trim_name" ] && [ "$trim_name" != "None" ]; then
rliterman@45:         export TRIM_ARG="--trim_name $trim_name";
rliterman@29: else
rliterman@45:         export TRIM_ARG="";
rliterman@43: fi;
rliterman@29: 
rliterman@29: if [ -n "$ref_id" ] && [ "$ref_id" != "None" ]; then
rliterman@45:         export REF_ID_ARG="--ref_id $ref_id";
rliterman@29: else
rliterman@45:         export REF_ID_ARG="";
rliterman@43: fi;
rliterman@29: 
rliterman@48: nextflow run ${__tool_directory__}/CSP2/CSP2.nf -profile csp2_galaxy --runmode screen \$QUERY_FASTA_ARG \$REF_FASTA_ARG \$QUERY_READS_ARG \$REF_READS_ARG \$REF_ID_ARG \$TRIM_ARG --readext $readext --forward $forward --reverse $reverse --ref_readext $readext --ref_forward $forward --ref_reverse $reverse --min_cov $min_cov --min_iden $min_iden --min_len $min_len --ref_edge $ref_edge --query_edge $query_edge --dwin $dwin --wsnps $wsnps --out ./CSP2_Screen_Output > Nextflow_Log.txt 2>&1;
rliterman@46: cp CSP2_Screen_Output/Screening_Results.tsv CSP2_Screen_Output/Raw_MUMmer_Summary.tsv CSP2_Screen_Output/Isolate_Data.tsv output/;
rliterman@30: ]]>
rliterman@29:         </command>
rliterman@29:         <inputs>
rliterman@29:                 <param name="query_fasta" type="data" format="fasta" value="" label="Query assemblies" multiple="true" optional="true" />
rliterman@29:                 <param name="ref_fasta" type="data" format="fasta" value="" label="Reference assemblies" multiple="true" optional="true" />
rliterman@29:                 <param name="query_reads" type="data" format="fastq,fastq.gz" value="" label="Query reads" multiple="true" optional="true" />
rliterman@29:                 <param name="ref_reads" type="data" format="fastq,fastq.gz" value="" label="Reference reads" multiple="true" optional="true" />
rliterman@29:                 <param name="min_cov" type="float" value="85" label="Minimum reference genome coverage to proceed with distance estimation" optional="true" />
rliterman@29:                 <param name="min_iden" type="float" value="99" label="Minimum alignment percent identity to detect SNPs" optional="true" />
rliterman@29:                 <param name="min_len" type="integer" value="500" label="Minimum alignment length to detect SNPs" optional="true" />
rliterman@29:                 <param name="ref_edge" type="integer" value="150" label="Prune SNPs within this many bases of reference contig edge" optional="true" />
rliterman@29:                 <param name="query_edge" type="integer" value="150" label="Prune SNPs within this many bases of query contig edge" optional="true" />
rliterman@29:                 <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" />
rliterman@29:                 <param name="wsnps" type="text" value="3,2,1" label="Comma-separated list of maximum SNP counts per density window" optional="true" />
rliterman@29:                 <param name="readext" type="text" value="fastq.gz" label="Read extension format (e.g., fastq.gz)" optional="true" />
rliterman@29:                 <param name="forward" type="text" value="_1.fastq.gz" label="Forward read suffix (e.g. _1.fastq.gz)" optional="true" />
rliterman@29:                 <param name="reverse" type="text" value="_2.fastq.gz" label="Forward read suffix (e.g. _2.fastq.gz)" optional="true" />
rliterman@29:                 <param name="trim_name" type="text" value="" label="Text to remove from all file names (e.g., _contigs_skesa)" optional="true" />
rliterman@29:                 <param name="ref_id" type="text" value="" label="Comma-separated list of desired Reference IDs (e.g., Sample_A,Sample_B)" optional="true" />
rliterman@29:         </inputs>
rliterman@29:         <outputs>
rliterman@46:                 <data name="raw_mummer" format="tabular" label="Raw MUMmer Output" from_work_dir="output/Raw_MUMmer_Summary.tsv" />
rliterman@46:                 <data name="isolate_data" format="tabular" label="Isolate Data" from_work_dir="output/Isolate_Data.tsv" />
rliterman@46:                 <data name="screening_results" format="tabular" label="Screening Results" from_work_dir="output/Screening_Results.tsv" />
rliterman@29:         </outputs>
rliterman@29:         <tests>
rliterman@29:                 <test>
rliterman@29:                         <param name="query_fasta">
rliterman@29:                                 <collection type="list">
rliterman@29:                                         <element name="Sample_A" value="assemblies/Sample_A.fasta" />
rliterman@29:                                         <element name="Sample_B" value="assemblies/Sample_B.fasta" />
rliterman@29:                                         <element name="Sample_C" value="assemblies/Sample_C.fasta" />
rliterman@29:                                         <element name="Sample_D" value="assemblies/Sample_D.fasta" />
rliterman@29:                                         <element name="Sample_E" value="assemblies/Sample_E.fasta" />
rliterman@29:                                         <element name="Sample_F" value="assemblies/Sample_F.fasta" />
rliterman@29:                                         <element name="Sample_G" value="assemblies/Sample_G.fasta" />
rliterman@29:                                         <element name="Sample_H" value="assemblies/Sample_H.fasta" />
rliterman@29:                                         <element name="Sample_I" value="assemblies/Sample_I.fasta" />
rliterman@29:                                         <element name="Sample_J" value="assemblies/Sample_J.fasta" />
rliterman@29:                                         <element name="Sample_K" value="assemblies/Sample_K.fasta" />
rliterman@29:                                         <element name="Sample_L" value="assemblies/Sample_L.fasta" />
rliterman@29:                                         <element name="Sample_M" value="assemblies/Sample_M.fasta" />
rliterman@29:                                         <element name="Sample_N" value="assemblies/Sample_N.fasta" />
rliterman@29:                                         <element name="Sample_O" value="assemblies/Sample_O.fasta" />
rliterman@29:                                 </collection>
rliterman@29:                         </param>
rliterman@29:                         <param name="query_reads">
rliterman@29:                                 <collection type="list">
rliterman@29:                                         <element name="Forward" value="reads/Week_42_Reads_1.fq.gz" />
rliterman@29:                                         <element name="Reverse" value="reads/Week_42_Reads_2.fq.gz" />
rliterman@29:                                 </collection>
rliterman@29:                         </param>
rliterman@29: 
rliterman@29:                         <param name="ref_id" value="Sample_A,Sample_B" />
rliterman@29:                         <param name="readext" value="fq.gz" />
rliterman@29:                         <param name="forward" value="_1.fq.gz" />
rliterman@29:                         <param name="reverse" value="_2.fq.gz" />
rliterman@29: 
rliterman@29:                         <output name="screening_results" value="Screening_Results.tsv" />
rliterman@29:                         <output name="isolate_data" value="Isolate_Data.tsv" />
rliterman@29:                 </test>
rliterman@29:         </tests>
rliterman@29:         <help>
rliterman@29:         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.
rliterman@29:         </help>
rliterman@29:         <citations>
rliterman@29:                 <citation type="doi">10.XXXX/placeholder.doi</citation>
rliterman@29:                 <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}}
rliterman@29:                 </citation>
rliterman@29:         </citations>
rliterman@29: </tool>
rliterman@29: