<tool id="csp2-snp" name="CSP2 (SNP Pipeline Mode)" version="0.9.7_Dev21">
        <description>Run SNP Pipeline analysis on isolates using one or more references.</description>
        <requirements>
                <requirement type="package" version="24.10.1">nextflow</requirement>
                <requirement type="package" version="1.5.8">micromamba</requirement>
        </requirements>
        <version_command>nextflow -version</version_command>
        <command detect_errors="aggressive"><![CDATA[
mkdir ./queries ./references;

if [[ "$query_fasta.query_fasta_select" == "none" ]]; then
        export QUERY_FASTA_ARG="";
elif [[ "$query_fasta.query_fasta_select" == "history" ]]; then
        #for query in $query_fasta.list:
                ln -sf ${query} ./queries/${query.element_identifier};
        #end for
        export QUERY_FASTA_ARG="--fasta ./queries";
else
        #for query in $query_fasta.coll.elements:
                ln -sf ${query} ./queries/${query.element_identifier};
        #end for
        export QUERY_FASTA_ARG="--fasta ./queries";
fi;

if [[ "$ref_fasta.ref_fasta_select" == "none" ]]; then
        export REF_FASTA_ARG="";
elif [[ "$ref_fasta.ref_fasta_select" == "history" ]]; then
        #for ref in $ref_fasta.list:
                ln -sf ${ref} ./references/${ref.element_identifier};
        #end for
        export REF_FASTA_ARG="--ref_fasta ./references";
else
        #for ref in $ref_fasta.coll.elements:
                ln -sf ${ref} ./references/${ref.element_identifier};
        #end for
        export REF_FASTA_ARG="--ref_fasta ./references";
fi;

if [[ "$query_reads.query_reads_select" == "none" ]]; then
        export QUERY_READS_ARG="";
elif [[ "$query_reads.query_reads_select" == "history" ]]; then
        #for query in $query_reads.list:
                ln -sf ${query} ./queries/${query.element_identifier};
        #end for
        export QUERY_READS_ARG="--reads ./queries";
else
        #for query in $query_reads.coll.forward:
                ln -sf ${query} ./queries/${query.element_identifier};
        #end for
        #for query in $query_reads.coll.reverse:
                ln -sf ${query} ./queries/${query.element_identifier};
        #end for
        export QUERY_READS_ARG="--reads ./queries";
fi;

if [[ "$ref_reads.ref_reads_select" == "none" ]]; then
        export REF_READS_ARG="";
elif [[ "$ref_reads.ref_reads_select" == "history" ]]; then
        #for ref in $ref_reads.list:
                ln -sf ${ref} ./references/${ref.element_identifier};
        #end for
        export REF_READS_ARG="--ref_reads ./references";
else
        #for ref in $ref_reads.coll.forward:
                ln -sf ${ref} ./references/${ref.element_identifier};
        #end for
        #for ref in $ref_reads.coll.reverse:
                ln -sf ${ref} ./references/${ref.element_identifier};
        #end for
        export REF_READS_ARG="--ref_reads ./references";
fi;

if [[ "$trim_name" != "DUMMYCSP2TRIM" ]]; then
        export TRIM_ARG="--trim_name $trim_name";
else
        export TRIM_ARG="";
fi;

if [[ "$rescue" == "true" ]]; then
        export RESCUE_ARG="--rescue";
else
        export RESCUE_ARG="";
fi;

nextflow run ${__tool_directory__}/CSP2/CSP2.nf -profile csp2_galaxy --runmode snp \$QUERY_FASTA_ARG \$REF_FASTA_ARG \$QUERY_READS_ARG \$REF_READS_ARG \$TRIM_ARG \$RESCUE_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 --max_missing $max_missing --n_ref $n_ref --out ./CSP2_SNP_Output > Nextflow_Log.txt 2>&1 &&
mkdir CSP2_Output &&
cat CSP2_SNP_Output/logs/Reference_IDs.txt | while read line; do cp 'CSP2_SNP_Output/SNP_Analysis/'\$line'/CSP2_SNP_Pipeline.log' 'CSP2_Output/'\$line'_CSP2_SNP_Pipeline.log'; cp 'CSP2_SNP_Output/SNP_Analysis/'\$line'/Reference_Screening.tsv' 'CSP2_Output/'\$line'_Reference_Screening.tsv';  cp 'CSP2_SNP_Output/SNP_Analysis/'\$line'/snp_distance_matrix_preserved.tsv' 'CSP2_Output/'\$line'_snp_distance_matrix_preserved.tsv'; cp 'CSP2_SNP_Output/SNP_Analysis/'\$line'/snp_distance_pairwise_preserved.tsv' 'CSP2_Output/'\$line'_snp_distance_pairwise_preserved.tsv'; cp 'CSP2_SNP_Output/SNP_Analysis/'\$line'/snpma_preserved.fasta' 'CSP2_Output/'\$line'_snpma_preserved.fasta';done &&
ls -la CSP2_Output;

]]>
        </command>
        <inputs>
                <conditional name="query_fasta">
                <param name="query_fasta_select" type="select" label="Get query assemblies from a collection or your history">
                        <option value="none" selected="true">Do not provide query assemblies</option>
                        <option value="collection">Query assemblies from a collection_type</option>
                        <option value="history">Query assemblies from your history</option>
                </param>
                <when value="none">
                </when>
                <when value="collection">
                        <param label="Query Assembly Collection" name="coll" type="data_collection" format="fasta" collection_type="list" />
                </when>
                <when value="history">
                        <param label="Query Assembly List" type="data" name="list" format="fasta"  multiple="true" />
                </when>
                </conditional>

                <conditional name="ref_fasta">
                <param name="ref_fasta_select" type="select" label="Get reference assemblies from a collection or your history">
                        <option value="none" selected="true">Do not provide reference assemblies</option>
                        <option value="collection">Reference assemblies from a collection_type</option>
                        <option value="history">Reference assemblies from your history</option>
                </param>
                <when value="none">
                </when>
                <when value="collection">
                        <param label="Reference Assembly Collection" name="coll" type="data_collection" format="fasta" collection_type="list" />
                </when>
                <when value="history">
                        <param label="Reference Assembly List" type="data" name="list" format="fasta"  multiple="true" />
                </when>
                </conditional>   
                <conditional name="query_reads">
                <param name="query_reads_select" type="select" label="Get query reads from a paired-end collection or your history">
                        <option value="none" selected="true">Do not provide query reads</option>
                        <option value="collection">Query reads from a collection_type</option>
                        <option value="history">Query reads from your history</option>
                </param>
                <when value="none">
                </when>
                <when value="collection">
                        <param label="Query Reads Collection" name="coll" type="data_collection" format="fastq,fastqsanger,fastq.gz,fastqsanger.gz,fastq.bz2,fastqsanger.bz2" collection_type="paired" />
                </when>
                <when value="history">
                        <param label="Query Reads List" type="data" name="list" format="fastq,fastqsanger,fastq.gz,fastqsanger.gz,fastq.bz2,fastqsanger.bz2"  multiple="true" />
                </when>
                </conditional>

                <conditional name="ref_reads">
                <param name="ref_reads_select" type="select" label="Get reference reads from a paired-end collection or your history">
                        <option value="none" selected="true">Do not provide reference reads</option>
                        <option value="collection">Refrence reads from a collection_type</option>
                        <option value="history">Refrence reads from your history</option>
                </param>
                <when value="none">
                </when>
                <when value="collection">
                        <param label="Refrence Reads Collection" name="coll" type="data_collection" format="fastq,fastqsanger,fastq.gz,fastqsanger.gz,fastq.bz2,fastqsanger.bz2" collection_type="paired" />
                </when>
                <when value="history">
                        <param label="Refrence Reads List" type="data" name="list" format="fastq,fastqsanger,fastq.gz,fastqsanger.gz,fastq.bz2,fastqsanger.bz2"  multiple="true" />
                </when>
                </conditional>             
                
                <param name="min_cov" type="float" value="85" label="Minimum reference genome coverage to proceed with distance estimation" optional="true" />
                <param name="min_iden" type="float" value="99" label="Minimum alignment percent identity to detect SNPs" optional="true" />
                <param name="min_len" type="integer" value="500" label="Minimum alignment length to detect SNPs" optional="true" />
                <param name="ref_edge" type="integer" value="150" label="Prune SNPs within this many bases of reference contig edge" optional="true" />
                <param name="query_edge" type="integer" value="150" label="Prune SNPs within this many bases of query contig edge" optional="true" />
                <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" />
                <param name="wsnps" type="text" value="3,2,1" label="Comma-separated list of maximum SNP counts per density window" optional="true" />
                <param name="readext" type="text" value="fastq.gz" label="Read extension format (e.g., fastq.gz)" optional="true" />
                <param name="forward" type="text" value="_1.fastq.gz" label="Forward read suffix (e.g. _1.fastq.gz)" optional="true" />
                <param name="reverse" type="text" value="_2.fastq.gz" label="Forward read suffix (e.g. _2.fastq.gz)" optional="true" />
                <param name="trim_name" type="text" value="DUMMYCSP2TRIM" label="Text to remove from all file names (e.g., _contigs_skesa)" optional="true" />
                <param name="rescue" type="boolean" value="false" label="Enable SNP edge rescue mode" optional="true" />
                <param name="max_missing" type="float" value="50" label="Maximum percent of isolates allowed to have missing data to retain a SNP" optional="true" />
                <param name="n_ref" type="integer" value="1" label="Number of reference genomes to select" optional="true" />

        </inputs>
        <outputs>
                <data name="isolate_data" format="tabular" label="Isolate Data" from_work_dir="CSP2_SNP_Output/Isolate_Data.tsv" />
                <data name="raw_mummer" format="tabular" label="Raw MUMmer Output" from_work_dir="CSP2_SNP_Output/Raw_MUMmer_Summary.tsv" />
                
                <collection name="csp2_logs" type='list' label="CSP2 Log">
                        <discover_datasets pattern="(?P&lt;designation&gt;.+)_CSP2_SNP_Pipeline\.log" directory="CSP2_Output" format='txt'/>
                </collection>
                <collection name="csp2_alignments" type='list' label="CSP2 Alignment">
                        <discover_datasets pattern="(?P&lt;designation&gt;.+)_snpma_preserved\.fasta" directory="CSP2_Output" format='fasta'/>
                </collection>
                <collection name="csp2_matrices" type='list' label="CSP2 Matrices">
                        <discover_datasets pattern="(?P&lt;designation&gt;.+)_snp_distance_matrix_preserved\.tsv" directory="CSP2_Output" format='tabular'/>
                </collection>
                <collection name="csp2_pairwise" type='list' label="CSP2 Pairwise Distances">
                        <discover_datasets pattern="(?P&lt;designation&gt;.+)_snp_distance_pairwise_preserved\.tsv" directory="CSP2_Output" format='tabular'/>
                </collection>
                <collection name="csp2_ref_screening" type='list' label="CSP2 Reference Screening">
                        <discover_datasets pattern="(?P&lt;designation&gt;.+)_Reference_Screening\.tsv" directory="CSP2_Output" format='tabular'/>
                </collection>
        </outputs>
        <tests>
                <test>
                        <param name="query_fasta">
                                <collection type="list">
                                        <element name="Sample_A" value="assemblies/Sample_A.fasta" />
                                        <element name="Sample_B" value="assemblies/Sample_B.fasta" />
                                        <element name="Sample_C" value="assemblies/Sample_C.fasta" />
                                        <element name="Sample_D" value="assemblies/Sample_D.fasta" />
                                        <element name="Sample_E" value="assemblies/Sample_E.fasta" />
                                        <element name="Sample_F" value="assemblies/Sample_F.fasta" />
                                        <element name="Sample_G" value="assemblies/Sample_G.fasta" />
                                        <element name="Sample_H" value="assemblies/Sample_H.fasta" />
                                        <element name="Sample_I" value="assemblies/Sample_I.fasta" />
                                        <element name="Sample_J" value="assemblies/Sample_J.fasta" />
                                        <element name="Sample_K" value="assemblies/Sample_K.fasta" />
                                        <element name="Sample_L" value="assemblies/Sample_L.fasta" />
                                        <element name="Sample_M" value="assemblies/Sample_M.fasta" />
                                        <element name="Sample_N" value="assemblies/Sample_N.fasta" />
                                        <element name="Sample_O" value="assemblies/Sample_O.fasta" />
                                </collection>
                        </param>
                        <param name="query_reads">
                                <collection type="list">
                                        <element name="Forward" value="reads/Week_42_Reads_1.fq.gz" />
                                        <element name="Reverse" value="reads/Week_42_Reads_2.fq.gz" />
                                </collection>
                        </param>

                        <param name="ref_id" value="Sample_A,Sample_B" />
                        <param name="readext" value="fq.gz" />
                        <param name="forward" value="_1.fq.gz" />
                        <param name="reverse" value="_2.fq.gz" />

                        <output name="isolate_data" value="Isolate_Data.tsv" />
                </test>
        </tests>
        <help>
        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.
        </help>
        <citations>
                <citation type="doi">10.XXXX/placeholder.doi</citation>
                <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}}
                </citation>
        </citations>
</tool>