rliterman@12: <tool id="csp2-screen" name="CSP2 (Screening Mode)" version="0.9.7_Dev21">
rliterman@12: 	<description>Screen query assemblies against reference assemblies</description>
rliterman@12: 	<requirements>
rliterman@12: 		<requirement type="package" version="24.10.1">nextflow</requirement>
rliterman@12: 		<requirement type="package" version="1.5.8">micromamba</requirement>
rliterman@12: 	</requirements>
rliterman@12: 	<version_command>nextflow -version</version_command>
rliterman@12: 	<command detect_errors="aggressive"><![CDATA[
rliterman@12: export CSP2_DIR=\$PWD;
rliterman@12: mkdir -p \$CSP2_DIR/queries \$CSP2_DIR/references;
rliterman@12: 
rliterman@12: if [ -n "$query_fasta" ] && [ "$query_fasta" != "None" ]; then
rliterman@12: 		#for query in $query_fasta:
rliterman@12: 				ln -sf ${query} \$CSP2_DIR/queries/${query.element_identifier};
rliterman@12: 		#end for
rliterman@12: 		export QUERY_FASTA_ARG="--fasta \$CSP2_DIR/queries";
rliterman@12: else
rliterman@12: 		export QUERY_FASTA_ARG="";
rliterman@12: fi;
rliterman@12: 
rliterman@12: if [ -n "$query_reads" ] && [ "$query_reads" != "None" ]; then
rliterman@12: 		#for query in $query_reads:
rliterman@12: 				ln -sf ${query} \$CSP2_DIR/queries/${query.element_identifier};
rliterman@12: 		#end for
rliterman@12: 		export QUERY_READS_ARG="--reads \$CSP2_DIR/queries";
rliterman@12: else
rliterman@12: 		export QUERY_READS_ARG="";
rliterman@12: fi;
rliterman@12: 
rliterman@12: if [ -n "$ref_fasta" ] && [ "$ref_fasta" != "None" ]; then
rliterman@12: 		#for ref in $ref_fasta:
rliterman@12: 				ln -sf ${ref} \$CSP2_DIR/references/${ref.element_identifier};
rliterman@12: 		#end for
rliterman@12: 		export REF_FASTA_ARG="--ref_fasta \$CSP2_DIR/references";
rliterman@12: else
rliterman@12: 		export REF_FASTA_ARG="";
rliterman@12: fi;
rliterman@12: 
rliterman@12: if [ -n "$ref_reads" ] && [ "$ref_reads" != "None" ]; then
rliterman@12: 		#for ref in $ref_reads:
rliterman@12: 				ln -sf ${ref} \$CSP2_DIR/references/${ref.element_identifier};
rliterman@12: 		#end for
rliterman@12: 		export REF_READS_ARG="--ref_reads \$CSP2_DIR/references";
rliterman@12: else
rliterman@12: 		export REF_READS_ARG="";
rliterman@12: fi;
rliterman@12: 
rliterman@12: if [ -n "$trim_name" ] && [ "$trim_name" != "None" ]; then
rliterman@12: 		export TRIM_ARG="--trim_name $trim_name";
rliterman@12: else
rliterman@12: 		export TRIM_ARG="";
rliterman@12: fi;
rliterman@12: 
rliterman@12: if [ -n "$ref_id" ] && [ "$ref_id" != "None" ]; then
rliterman@12: 		export REF_ID_ARG="--ref_id $ref_id";
rliterman@12: else
rliterman@12: 		export REF_ID_ARG="";
rliterman@12: fi;
rliterman@12: 
rliterman@26: 		nextflow run ${__tool_directory__}/CSP2/CSP2.nf -profile csp2_galaxy --cores $cores --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 --cores 8 --out \$CSP2_DIR/CSP2_Screen_Output > Nextflow_Log.txt 2>&1;
rliterman@21: 
rliterman@26: 		zip -r work.zip work;
rliterman@26: 		zip -r csp2.zip CSP2_Screen_Output;
rliterman@12: 	]]>
rliterman@12: 	</command>
rliterman@12: 	<inputs>
rliterman@12: 		<param name="query_fasta" type="data" format="fasta" value="" label="Query assemblies" multiple="true" optional="true" />
rliterman@12: 		<param name="ref_fasta" type="data" format="fasta" value="" label="Reference assemblies" multiple="true" optional="true" />
rliterman@12: 		<param name="query_reads" type="data" format="fastq,fastq.gz" value="" label="Query reads" multiple="true" optional="true" />
rliterman@12: 		<param name="ref_reads" type="data" format="fastq,fastq.gz" value="" label="Reference reads" multiple="true" optional="true" />
rliterman@12: 		<param name="min_cov" type="float" value="85" label="Minimum reference genome coverage to proceed with distance estimation" optional="true" />
rliterman@12: 		<param name="min_iden" type="float" value="99" label="Minimum alignment percent identity to detect SNPs" optional="true" />
rliterman@12: 		<param name="min_len" type="integer" value="500" label="Minimum alignment length to detect SNPs" optional="true" />
rliterman@12: 		<param name="ref_edge" type="integer" value="150" label="Prune SNPs within this many bases of reference contig edge" optional="true" />
rliterman@12: 		<param name="query_edge" type="integer" value="150" label="Prune SNPs within this many bases of query contig edge" optional="true" />
rliterman@26:                 <param name="cores" type="integer" value="2" label="Available cores per node" optional="true" />
rliterman@12: 		<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@12: 		<param name="wsnps" type="text" value="3,2,1" label="Comma-separated list of maximum SNP counts per density window" optional="true" />
rliterman@12: 		<param name="readext" type="text" value="fastq.gz" label="Read extension format (e.g., fastq.gz)" optional="true" />
rliterman@12: 		<param name="forward" type="text" value="_1.fastq.gz" label="Forward read suffix (e.g. _1.fastq.gz)" optional="true" />
rliterman@12: 		<param name="reverse" type="text" value="_2.fastq.gz" label="Forward read suffix (e.g. _2.fastq.gz)" optional="true" />
rliterman@12: 		<param name="trim_name" type="text" value="" label="Text to remove from all file names (e.g., _contigs_skesa)" optional="true" />
rliterman@12: 		<param name="ref_id" type="text" value="" label="Comma-separated list of desired Reference IDs (e.g., Sample_A,Sample_B)" optional="true" />
rliterman@12: 	</inputs>
rliterman@12: 	<outputs>
rliterman@12: 		<data name="screening_results" format="tabular" label="Screening Results" from_work_dir="CSP2_Screen_Output/Screening_Results.tsv" />
rliterman@19: 		<data name="raw_mummer" format="tabular" label="Raw MUMmer Output" from_work_dir="CSP2_Screen_Output/Raw_MUMmer_Summary.tsv" />
rliterman@12: 		<data name="isolate_data" format="tabular" label="Isolate Data" from_work_dir="CSP2_Screen_Output/Isolate_Data.tsv" />
rliterman@12: 		<data name="nextflow_log" format="txt" label="Nextflow Log" from_work_dir="Nextflow_Log.txt" />
rliterman@25:                 <data name="out_dir" format="zip" label="CSP2 Directory" from_work_dir="csp2.zip" />
rliterman@21:                 <data name="work_dir" format="zip" label="Nextflow Work Directory" from_work_dir="work.zip" />
rliterman@12: 	</outputs>
rliterman@12: 	<tests>
rliterman@12: 		<test>
rliterman@12: 			<param name="query_fasta">
rliterman@12: 					<collection type="list">
rliterman@12: 							<element name="Sample_A" value="assemblies/Sample_A.fasta" />
rliterman@12: 							<element name="Sample_B" value="assemblies/Sample_B.fasta" />
rliterman@12: 							<element name="Sample_C" value="assemblies/Sample_C.fasta" />
rliterman@12: 							<element name="Sample_D" value="assemblies/Sample_D.fasta" />
rliterman@12: 							<element name="Sample_E" value="assemblies/Sample_E.fasta" />
rliterman@12: 							<element name="Sample_F" value="assemblies/Sample_F.fasta" />
rliterman@12: 							<element name="Sample_G" value="assemblies/Sample_G.fasta" />
rliterman@12: 							<element name="Sample_H" value="assemblies/Sample_H.fasta" />
rliterman@12: 							<element name="Sample_I" value="assemblies/Sample_I.fasta" />
rliterman@12: 							<element name="Sample_J" value="assemblies/Sample_J.fasta" />
rliterman@12: 							<element name="Sample_K" value="assemblies/Sample_K.fasta" />
rliterman@12: 							<element name="Sample_L" value="assemblies/Sample_L.fasta" />
rliterman@12: 							<element name="Sample_M" value="assemblies/Sample_M.fasta" />
rliterman@12: 							<element name="Sample_N" value="assemblies/Sample_N.fasta" />
rliterman@12: 							<element name="Sample_O" value="assemblies/Sample_O.fasta" />
rliterman@12: 					</collection>
rliterman@12: 			</param>
rliterman@12: 			<param name="query_reads">
rliterman@12: 					<collection type="list">
rliterman@12: 							<element name="Forward" value="reads/Week_42_Reads_1.fq.gz" />
rliterman@12: 							<element name="Reverse" value="reads/Week_42_Reads_2.fq.gz" />
rliterman@12: 					</collection>
rliterman@12: 			</param>
rliterman@12: 
rliterman@12: 			<param name="ref_id" value="Sample_A,Sample_B" />
rliterman@12: 			<param name="readext" value="fq.gz" />
rliterman@12: 			<param name="forward" value="_1.fq.gz" />
rliterman@12: 			<param name="reverse" value="_2.fq.gz" />
rliterman@12: 
rliterman@12: 			<output name="screening_results" value="Screening_Results.tsv" />
rliterman@12: 			<output name="isolate_data" value="Isolate_Data.tsv" />
rliterman@12: 		</test>
rliterman@12: 	</tests>
rliterman@12: 	<help>
rliterman@12: 	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@12: 	</help>
rliterman@12: 	<citations>
rliterman@12: 		<citation type="doi">10.XXXX/placeholder.doi</citation>
rliterman@12: 		<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@12: 		</citation>
rliterman@12: 	</citations>
rliterman@12: </tool>
rliterman@12: