annotate smap_design.xml @ 0:fe9f46c3bd28 draft default tip

Uploaded

<tool id="smap_design" name="SMAP-design" version="0.1.0" python_template_version="3.5">
    <requirements>
        <!--<requirement type="package">smap-design</requirement>-->
        <requirement type="package">primer3-py</requirement>
        <requirement type="package">pandas</requirement>
        <requirement type="package">biopython</requirement>
        <requirement type="package">matplotlib</requirement>
        <requirement type="package">gffutils</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
        #if $input_source_conditional.input_source == "pregenerated_inputs":
            design_input_fasta=$input_source_conditional.input_fasta;
            design_input_gff3=$input_source_conditional.input_gff3;
        #else:
            ln -s $input_source_conditional.fasta_file species_fasta.fasta;
            ln -s $input_source_conditional.gff_file species.gff3;
            ln -s $input_source_conditional.gene_families_file families_file.txt;
            #if $input_source_conditional.interest_input_selector.interest_input == "genes":
                ln -s $input_source_conditional.interest_input_selector.genes_file genes_file.txt;
            #else
                ln -s $input_source_conditional.interest_input_selector.homology_groups_file hom_groups_file.txt;
            #end if
        python3 $__tool_directory__/Get_fasta_and_gff_for_selected_hom_groups_extended_flanking_region.py ./species.gff3 ./species_fasta.fasta ./families_file.txt $input_source_conditional.species --region $input_source_conditional.region
            #if $input_source_conditional.interest_input_selector.interest_input == "genes":
                --genes genes_file.txt
            #else
                --hom_groups hom_groups_file.txt
            #end if
            &&
            ######  need to set the design_input_fasta and design_input_gff3 vars
            design_input_fasta=`ls *_bp.fasta` &&
            design_input_gff3=`ls *_bp.gff`;
        #end if
        python3 $__tool_directory__/smap_design/SMAPdesign.py
            --output 'OUTPUT' 
            --minimumAmpliconLength $min_amplicon_length 
            --maximumAmpliconLength $max_amplicon_length 
            --numbergRNAs $number_grnas 
            --numberAmplicons $number_amplicons 
            #if $print_summary:
                --summary
            #end if
            $bordersOnly
            --distance $distance
            $print_summary
            $allAmplicons
            --gRNAfile $gRNAfile
            --gRNAsource $gRNAsource.value
            #if $selectGenes:
                --selectGenes $selectGenes
            #end if
            --gRNAoverlap $gRNAoverlap
            --generateAmplicons $generateAmplicons
            --threshold $threshold
            --borderLength $borderLength
            $ampliconLabel
            --targetRegion5 $targetRegion5
            --targetRegion3 $targetRegion3
            #if $targetSpecificRegion:
                --targetSpecificRegion $targetSpecificRegion
            #end if
            $misPrimingAllowed
            $restrictPrimerDesign 
            --promoter '$promoter'
            --scaffold '$scaffold'
            \$design_input_fasta \$design_input_gff3;
        mv OUTPUT.gff3 $gff_output;
        mv OUTPUT_primers.tsv $primers_output;
        mv OUTPUT_gRNAs.tsv $gRNA_per_gene;
        #if $print_summary:
            mv OUTPUT_SMAPdesign_summary.tsv $summary;
        #end if
        #if $bordersOnly:
            mv OUTPUT_SMAPdesign_borders.gff3 $borders_output;
        #end if
        ;
    ]]></command>
    <inputs>
        <conditional name="input_source_conditional">
        <param name="input_source" type="select">
            <option value="smap_selector">Generate the target gene using SMAP selector</option>
            <option value="pregenerated_inputs">Use pregenerated inputs</option>
        </param>
        <when value="smap_selector">
            <param name="species" type="text" label="Species, corresponding with species indicated in the gene family info file"/>
            <param name="gene_families_file" type="data" format="tsv,tabular" label="gene family information file (tab-delimited) for the (coding) genes, separated per gene family type"/>
            <param name="fasta_file" type="data" format="fasta" label="FASTA file containing the genomic sequence of the species"/>
            <param name="gff_file" type="data" format="gff,gff3" label="gff3 file (tab-delimited) of the species containing gene, CDS, and exon features with positions relative to the fasta file"/>
            <conditional name="interest_input_selector">
                <param name="interest_input" type="select">
                    <option value="genes">list with genes of interest</option>
                    <option value="homology_groups">list with homology groups of interest</option>
                </param>
                <when value="genes">
                    <param name="genes_file" type="data" format="txt" label="Genes file"/>
                </when>
                <when value="homology_groups">
                    <param name="homology_groups_file" type="data" format="txt" label="Homology groups file"/>
                </when>
            </conditional>
            <param argument="--region" type="integer" value="0" label="Region to extend the FASTA sequence of the genes of interest on both sides with the given number of basepairs or with the maximum possible" help="default: 0"/>

        </when>
        <when value="pregenerated_inputs">
            <param name="input_fasta" type="data" optional="false" label="FASTA file containing all genes to screen" format="fasta"/>
            <param name="input_gff3" type="data" optional="false" label="GFF3 File" help="GFF3 file with at least the CDS feature with positions relative to the FASTA file " format="gff3,gff"/>
        </when>
        </conditional>

        <param name="number_amplicons" type="integer" value="2" label="The maximum number of non-overlapping amplicons in the output (default = 2)" help="sets the seed for the random number generator"/>
        <param name="number_grnas" type="integer" value="2" label="Maximum number of gRNAs to retain per amplicon (default = 2)" help=""/>
        <param name="min_amplicon_length" type="integer" value="120" label="The minimum length of the amplicons in base pairs (default = 120)" help=""/>
        <param name="max_amplicon_length" type="integer" value="150" label="The maximum length of the amplicons in base pairs (default = 120)" help=""/>
        <param name="distance" type="integer" optional="true" value="15" label="Minimum number of bases between primer and gRNA (default = 15)" help=""/>
        <param name="print_summary" type="boolean" truevalue="--summary" falsevalue="" label="Write summary file and plot of the output"/>
        <param argument='--allAmplicons' type="boolean" truevalue="--allAmplicons" falsevalue="" label='Return all amplicons with their respective gRNAs per gene (extra file)'/>
        <param name='bordersOnly' type="boolean" truevalue="-bo" falsevalue="" checked="false" label='Write additional GFF file with only borders (for SMAP)'/>
        <param argument='--gRNAfile' type="data" format="tsv,tabular" label='CRISPOR, FlashFry or other gRNA design program output file' help="The CRISPOR and Flashfry file must contain a header and 12 columns. Check the manual for specifics"/>
        <param name='gRNAsource' type="select" label="What is the source of the gRNA file?">
            <option value="crispor">CRISPOR</option>
            <option value="flashfry">FlashFry</option>
            <option value="other">Other</option>
        </param>
        <param argument='--selectGenes' type="data" optional="true" format="txt" label="OPTIONAL List of genes (one per line) to which amplicons and guides must be designed" help='The other genes in the fasta file will be used to check for specificity only. default: if no list is used then, for all genes in the fasta the design is done'/>
        <param argument='--gRNAoverlap' label='The minimum number of bases between the start of two gRNAs' type="integer" value="5"/>
        <param argument='--generateAmplicons' label="Number of amplicons to generate per gene by Primer3." help="To generate 50 amplicons per 1000 bases per gene enter -1" value="150" type="integer"/>
        <param argument='--threshold' label='Minimum gRNA MIT score allowed (default = 80)' value="80" type="integer"/>
        <param argument='--borderLength' value="10" type="integer" label='The length of the borders (for SMAP)'/>
        <param argument='--ampliconLabel' type="boolean" truevalue="--ampliconLabel" falsevalue="" checked="false" label='Number the amplicons from left to right instead of from best to worst'/>
        <param argument='--gRNAlabel' type="boolean" truevalue="--gRNAlabel" falsevalue="" checked="false" label='Number the gRNAs from left to right instead of from best to worst  (based on specificity scores)'/>
        <param argument='--targetRegion5' label="The fraction of the coding sequencing that cannot be targeted at the 5' end as indicated by a float between 0 and 1 (default = 0.2)" value="0.2" type="float"/>
        <param argument='--targetRegion3' label="The fraction of the coding sequencing that cannot be targeted at the 3' end as indicated by a float between 0 and 1 (default = 0.2)" value="0.2" type="float"/>
        <param argument='--targetSpecificRegion' label='Only target a specific region in the gene indicated by the feature name in the GFF file' value="" type="text"/>
        <param argument='--misPrimingAllowed' type="boolean" truevalue="--misPrimingAllowed" falsevalue="" label='Do not check for mispriming in the gene set when designing primers' help='By default Primer3 will not allow primers that can prime at other genes in the gene set' checked="false"/>
        <param argument='--restrictPrimerDesign' type="boolean" truevalue="--restrictPrimerDesign" falsevalue="" label='Restrict primer design in large introns' help="Increases the speed of amplicon design. This should have no impact on the output" checked="false"/>
        <param argument="--promoter" type="text" label='Give the last 6 bases of the promoter that will be used to express the gRNA. This will be taken into account when checking for BsaI or BbsI sites in the gRNA' help="default: U6 promoter = GTAGTG)" value="GTAGTG" />

        <param argument="--scaffold" type="text" label="Give the first 6 bases of the scaffold that will be used. This will be taken into account when checking for BsaI or BbsI sites in the gRNA" value="GTTTTA"/>

    </inputs>
    <outputs>
        <data format="gff3" name="gff_output" label="GFF3 design" />
        <data format="tsv" name="primers_output" label="Primers sequences per gene" />
        <data format="tsv" name="gRNA_per_gene" label="gRNA sequences per gene" />
        <data format="tsv" name="summary" label="Summary">
            <filter>print_summary == "--summary"</filter>
        </data>
        <data format="gff3" name="borders_output" label="Borders">
            <filter>bordersOnly == "-bo"</filter>
        </data>
    </outputs>
    <help><![CDATA[
        TODO: Fill in help.
    ]]></help>
</tool>