view searchgui.xml @ 41:86ab83258d40 draft

Fix datatypes for searchgui input and output

<tool id="search_gui" name="Search GUI" version="1.27.0">
    <description>
        Perform protein identification using various search engines and prepare results for input to Peptide Shaker
    </description>
    <requirements>
        <requirement type="package" version="1.27">searchgui</requirement>
    </requirements>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="stdio" />
    <command>
<![CDATA[
        #from datetime import datetime
        #set $exp_str = "Galaxy_Experiment_%s" % datetime.now().strftime("%Y%m%d%H%M%s")
        #set $samp_str = "Sample_%s" % datetime.now().strftime("%Y%m%d%H%M%s")
        #set $temp_stderr = "searchgui_stderr"

        mkdir output;
        mkdir output_reports;
        cwd=`pwd`;
        #for $mgf in $peak_lists:
            #set $input_name = $mgf.display_name.replace(".mgf", "") + ".mgf"
            ln -s -f '${mgf}' '${input_name}';
        #end for
        ##ln -s "${input_database}" input_database.fasta;
        cp "${input_database}" input_database.fasta;

        ###########################################
        ####       Creating decoy database     ####
        ###########################################
        #if $create_decoy:
            echo "Creating decoy database.";
            java -cp \$SEARCHGUI_JAR_PATH eu.isas.searchgui.cmd.FastaCLI -in input_database.fasta -decoy &&
            rm input_database.fasta &&
            cp input_database_concatenated_target_decoy.fasta input_database.fasta &&
            ##ln -sf input_database_concatenated_target_decoy.fasta input_database.fasta;
        #end if

        #####################################################
        ## generate IdentificationParameters for SearchGUI ##
        #####################################################

        (java -cp \$SEARCHGUI_JAR_PATH eu.isas.searchgui.cmd.IdentificationParametersCLI
            -out SEARCHGUI_IdentificationParameters.parameters

            @GENERAL_PARAMETERS@

            -db input_database.fasta

            #if $advanced.advanced_type_selector == "advanced":

                #if $advanced.xtandem.xtandem_selector == "yes"

                    -xtandem_npeaks ${advanced.xtandem.xtandem_npeaks}
                    -xtandem_min_peaks ${advanced.xtandem.xtandem_min_peaks}
                    -xtandem_min_frag_mz ${advanced.xtandem.xtandem_min_frag_mz}
                    -xtandem_min_prec_mass ${advanced.xtandem.xtandem_min_prec_mass}
                    -xtandem_noise_suppr ${advanced.xtandem.xtandem_noise_suppr}

                    #if $advanced.xtandem.xtandem_refine.xtandem_refine_selector == "yes"
                        -xtandem_refine 1
                        -xtandem_refine_unc ${advanced.xtandem.xtandem_refine.xtandem_refine_unc}
                        -xtandem_refine_semi ${advanced.xtandem.xtandem_refine.xtandem_refine_semi}
                        -xtandem_refine_p_mut ${advanced.xtandem.xtandem_refine.xtandem_refine_p_mut}
                        -xtandem_refine_snaps ${advanced.xtandem.xtandem_refine.xtandem_refine_snaps}
                        -xtandem_refine_spec_synt ${advanced.xtandem.xtandem_refine.xtandem_refine_spec_synt}
                    #end if
                #end if

                #if $advanced.omssa.omssa_selector == "yes"
                    -omssa_hitlist_length ${advanced.omssa.hitlist_length}
                    -omssa_remove_prec ${advanced.omssa.remove_precursor}
                    -omssa_scale_prec ${advanced.omssa.scale_precursor}
                    -omssa_estimate_charge ${advanced.omssa.estimate_charge}
                #end if

                #if $advanced.msgf.msgf_selector == "yes"
                    -msgf_min_pep_length ${advanced.msgf.msgf_min_pep_length}
                    -msgf_max_pep_length ${advanced.msgf.msgf_max_pep_length}
                    -msgf_termini ${advanced.msgf.msgf_termini}
                    -msgf_num_ptms ${advanced.msgf.msgf_num_ptms}
                #end if

                ##if $advanced.ms_amanda.ms_amanda_selector == "yes"
                ##end if

            #end if

        2> $temp_stderr)
        &&

        ################
        ## Search CLI ##
        ################
        (java -Djava.awt.headless=true -cp \$SEARCHGUI_JAR_PATH eu.isas.searchgui.cmd.SearchCLI 
            -temp_folder `pwd`
            -spectrum_files \$cwd
            -output_folder \$cwd/output
            -id_params SEARCHGUI_IdentificationParameters.parameters

            -threads "\${GALAXY_SLOTS:-12}"
            -correct_titles "${correct_titles}"
            $missing_titles
            -mgf_splitting "${mgf_splitting}"
            -mgf_spectrum_count "${mgf_spectrum_count}"

            ## Turn of the protein tree generation as it can produce errors if the search is finished before the tree is created
            ## the tree is generated afterwards in PeptideShaker
            -protein_index 0

            ##-makeblastdb_folder \$BLAST_ROOT_DIR

            #if $advanced.advanced_type_selector == "advanced":

                #if $advanced.xtandem.xtandem_selector == "yes"
                    -xtandem 1
                #else
                    -xtandem 0
                #end if

                #if $advanced.omssa.omssa_selector == "yes"
                    -omssa 1
                #else
                    -omssa 0
                #end if

                #if $advanced.msgf.msgf_selector == "yes"
                    -msgf 1
                #else
                    -msgf 0
                #end if

                #if $advanced.myrimatch.myrimatch_selector == "yes"
                    -myrimatch 1
                #else
                    -myrimatch 0
                #end if

                #if $advanced.comet.comet_selector == "yes"
                    -comet 1
                #else
                    -comet 0
                #end if

                #if $advanced.ms_amanda.ms_amanda_selector == "yes"
                    -ms_amanda 1
                #else
                    -ms_amanda 0
                #end if

                #if $advanced.tide.tide_selector == "yes"
                    -tide 1
                #else
                    -tide 0
                #end if

            #else
                -ms_amanda 0 -tide 0
            #end if
            
            ## single zip file
            -output_option 0

            ## mgf and database in output
            -output_data 1

        2>> $temp_stderr)

        &&

        (mv output/searchgui_out.zip searchgui_out.zip 2>> $temp_stderr)

        &&

        exit_code_for_galaxy=\$?;
        cat $temp_stderr 2>&1;
        (exit \$exit_code_for_galaxy)
]]>
    </command>
    <inputs>
        <param format="fasta" name="input_database" type="data" label="Protein Database"
            help="Select FASTA database from history"/>

        <param name="create_decoy" type="boolean" truevalue="True" falsevalue="False" checked="true"
            label="Create a concatenated target/decoy database before running PeptideShaker"
            help="Selecting this option will help PeptideShaker calculate FDR values" />

        <param name="peak_lists" format="mgf" type="data" multiple="true" label="Input Peak Lists (mgf)"
            help="Select appropriate MGF dataset(s) from history" />

        <expand macro="general_options"/>

        <param name="correct_titles" type="select" label="How should PeptideShaker deal with duplicate spectra?"
            help="Unless you suspect some input files to be genuine duplicates then rename spectra is the safest option">
            <option value="0">no correction</option>
            <option value="1" selected="True">rename spectra</option>
            <option value="2">delete spectra</option>
        </param>

        <param name="missing_titles" type="boolean" checked="false" truevalue="-missing_titles 1" falsevalue="-missing_titles 0"
            label="Add missing spectrum titles" help="(-missing_titles)"/>

        <param name="mgf_splitting" type="integer" value="1000" label="The maximum mgf file size in MB before splitting the mgf"
            help="Choose a smaller value if you are running on a machine with limited memory"/>

        <param name="mgf_spectrum_count" type="integer" value="25000" label="The maximum number of spectra per mgf file when splitting"
            help="Choose a smaller value if you are running on a machine with limited memory"/>

        <conditional name="advanced">
            <param name="advanced_type_selector" type="select" label="Basic or Advanced Search options">
                <option value="basic" selected="True">Basic</option>
                <option value="advanced">Advanced</option>
            </param>
            <when value="basic" />
            <when value="advanced">
                <conditional name="xtandem">
                    <param name="xtandem_selector" type="select" label="Run X!Tandem search">
                        <option value="yes" selected="True">Search with X!Tandem</option>
                        <option value="no">No X!Tandem search</option>
                    </param>
                    <when value="no" />
                    <when value="yes">
                        <param name="xtandem_npeaks" type="integer" value="50" 
                            label="X!Tandem: Total Peaks" help="Maximum number of peaks to be used from a spectrum"/>
                        <param name="xtandem_min_peaks" type="integer" value="15" 
                            label="X!Tandem: Min Peaks" help="Minimum number of peaks required for a spectrum to be considered"/>
                        <param name="xtandem_min_frag_mz" type="integer" value="200" 
                            label="X!Tandem: Min Frag m/z" help="Fragment mass peaks with m/z less than this value will be discarded"/>
                        <param name="xtandem_min_prec_mass" type="integer" value="200" 
                            label="X!Tandem: Min Precursor Mass" help="Minimum mass of 1+ mass of parent ion to be considered"/>
                        <param name="xtandem_noise_suppr" type="boolean" checked="true" truevalue="1" falsevalue="0"
                            label="X!Tandem: Noise Suppression" help="Use noise suppression"/>

                        <conditional name="xtandem_refine"><!-- -xtandem_refine -->
                            <param name="xtandem_refine_selector" type="select" label="X!Tandem peptide model refinement">
                                <option value="no" selected="True">Don't refine</option>
                                <option value="yes" >Use refinement</option>
                            </param>
                            <when value="no"/>
                            <when value="yes">
                                <param name="xtandem_refine_unc" type="boolean" truevalue="1" falsevalue="0"
                                    label="X!Tandem: Unanticipated cleavage, refinement" help="Allow for unanticipated cleavage during refinement"/>
                                <param name="xtandem_refine_semi" type="boolean" truevalue="1" falsevalue="0" 
                                    label="X!Tandem: Cleavage semi, refinement" help="Search for semi-tryptic peptides during refinement"/>
                                <param name="xtandem_refine_p_mut" type="boolean" truevalue="1" falsevalue="0" 
                                    label="X!Tandem: Point mutations, refinement" help="Allow for point mutations during refinement"/>
                                <param name="xtandem_refine_snaps" type="boolean" truevalue="1" falsevalue="0" 
                                    label="X!Tandem: snAPs, refinement" help="Search for known single amino acid polymorphisms during refinement"/>
                                <param name="xtandem_refine_spec_synt" type="boolean" truevalue="1" falsevalue="0" 
                                    label="X!Tandem: Spectrum synthesis, refinement" help="Use spectrum synthesis scoring"/>
                            </when>
                        </conditional>
                    </when>
                </conditional>

                <conditional name="omssa">
                    <param name="omssa_selector" type="select" label="Run OMSSA search">
                        <option value="yes" selected="True">Search with OMSSA</option>
                        <option value="no">No OMSSA search</option>
                    </param>
                    <when value="no" />
                    <when value="yes">
                        <param name="hitlist_length" label="OMSSA: Hit List Length" type="integer" value="25" />
                        <param name="remove_precursor" label="OMSSA: Remove Precurosr" type="boolean" truevalue="1" falsevalue="0" checked="true"/>
                        <param name="scale_precursor" label="OMSSA: Scale Precursor Mass" type="boolean" truevalue="1" falsevalue="0" checked="false"/>
                        <param name="estimate_charge" label="OMSSA: Estimate Charge" type="boolean" truevalue="1" falsevalue="0" checked="true" />
                    </when>
                </conditional>

                <conditional name="msgf">
                    <param name="msgf_selector" type="select" label="Run MSGF search">
                        <option value="yes" selected="True">Search with MSGF</option>
                        <option value="no">No MSGF search</option>
                    </param>
                    <when value="no" />
                    <when value="yes">
                        <param name="msgf_min_pep_length" type="integer" value="6"
                            label="Minimum Peptide Length" help="Minimum length for a peptide to be considered"/>
                        <param name="msgf_max_pep_length" type="integer" value="30" 
                            label="Maximum Peptide Length" help="Maximum length for a peptide to be considered"/>
                        <param name="msgf_termini" type="select" format="text" 
                            label="Number of tolerable termini" help="Searches will take much longer if selecting a value other than 2">
                            <option value="0">0 (ie non-specific cleavage)</option>
                            <option value="1">1 (ie semi-tryptic cleavage)</option>
                            <option value="2" selected="true">2 (ie fully-tryptic cleavage)</option>
                        </param>
                        <param name="msgf_num_ptms" label="Max PTMs per peptide" type="integer" value="2"/>
                    </when>
                </conditional>

                <conditional name="myrimatch">
                    <param name="myrimatch_selector" type="select" label="Run MyriMatch search">
                        <option value="yes" selected="True">Search with MyriMatch</option>
                        <option value="no">No MyriMatch search</option>
                    </param>
                    <when value="no" />
                    <when value="yes">
                    </when>
                </conditional>

                <conditional name="comet">
                    <param name="comet_selector" type="select" label="Run Comet search">
                        <option value="yes"  selected="True">Search with Comet</option>
                        <option value="no">No Comet search</option>
                    </param>
                    <when value="no" />
                    <when value="yes">
                    </when>
                </conditional>

                <conditional name="ms_amanda">
                    <param name="ms_amanda_selector" type="select" label="Run MS Amanda search">
                        <option value="yes">Search with MS Amanda</option>
                        <option value="no" selected="True">No MS Amanda search</option>
                    </param>
                    <when value="no" />
                    <when value="yes">
                    </when>
                </conditional>

                <conditional name="tide">
                    <param name="tide_selector" type="select" label="Run Tide search">
                        <option value="yes">Search with Tide</option>
                        <option value="no" selected="True">No Comet search</option>
                    </param>
                    <when value="no" />
                    <when value="yes">
                    </when>
                </conditional>

            </when>
        </conditional>
    </inputs>
    <outputs>
        <data name="searchgui_results" format="searchgui_zip" from_work_dir="searchgui_out.zip" label="${tool.name} on ${on_string}" />
    </outputs>
    <tests>
        <test>
            <param name="peak_lists" value="searchgui_tinyspectra1.mgf"/>
            <param name="input_database" value="searchgui_tinydb1.fasta" ftype="fasta"/>
            <param name="precursor_ion_tol" value="100"/>
            <param name="fixed_modifications" value="carbamidomethyl c"/>
            <param name="variable_modifications" value="oxidation of m"/>
            <param name="min_charge" value="1"/>
            <param name="max_charge" value="3"/>
            <param name="advanced_type_selector" value="advanced"/>
            <param name="xtandem_selector" value="yes"/>
            <param name="xtandem_selector.xtandem_refine_selector" value="yes"/>
            <param name="omssa_selector" value="no"/>
            <param name="msgf_selector" value="yes"/>
            <param name="ms_amanda_selector" value="no"/>
            <output name="output" file="tiny_searchgui_result1.zip" ftype="searchgui_archive" compare="sim_size" delta="600" /> 
        </test>
    </tests>
    <help>
**What it does**

Runs multiple search engines on any number of MGF peak lists using the SearchGUI.

By default X! Tandem, OMSSA, MS-GF+, Myrimatch and Comet are run. Optionally, Amanda and Tide engines can also be run.

    </help>
    <expand macro="citations" />
</tool>