view pecan.xml @ 0:69274b068a48 draft default tip

planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tools/pecan commit 21855dec2949420d3bccfd6c9912578e9aacecf0

<tool id="pecan" name="PECAN" version="0.1.0">
    <description>detect peptides directly from DIA Mass Spec</description>
    <requirements>
        <requirement type="package" version="0.9.9.3">pecan</requirement>
    </requirements>
    <stdio>
        <exit_code range="1:" />
    </stdio>
    <version_command>pecan --VERSION</version_command>
    <command><![CDATA[
        ## ln mzML file
        #if $isolation.isolationsrc == 'entered':
            #set $isolationfile = $isolationSheme
        #else
            #set $isolationfile = $isolation.isolation_windows
        #end if
        #set $outputRoot = 'test'
        awk 'NF>1 && $1 ~ /^[0-9]+(\.[0-9]*)?$/{print \$1; print \$2; print "$outputRoot"}' $isolationfile | 
        xargs -n3 -P\${GALAXY_SLOTS} 
        pecan 
        #if $backgroundProteome:
            --backgroundProteome='$backgroundProteome'
        #end if
            --bgDecoyNumber=$bgDecoyNumber
        #if $decoyTag and str($decoyTag):
            --decoyTag=$decoyTag
        #end if
        --ionTypes=$--ionTypes
        --minElution=$minElution
        --maxElution=$maxElution
        --overlap=$advanced_options.overlap 
        #if $advanced_options.ms1.ms1Unit == 'ppm':
            --ms1Unit=ppm
            --ms1Tolerance=$advanced_options.ms1.ms1Tolerance
        #elif $advanced_options.ms1.ms1Unit == 'mz':
            --ms1Unit=mz
            --ms1Tolerance=$advanced_options.ms1.ms1Tolerance
        #end if
        #if $advanced_options.ms2.ms2Unit == 'ppm':
            --ms2Unit=ppm
            --ms2Tolerance=$advanced_options.ms2.ms2Tolerance
        #elif $advanced_options.ms2.ms2Unit == 'mz':
            --ms2Unit=mz
            --ms2Tolerance=$advanced_options.ms2.ms2Tolerance
        #end if
        --minCharge=$advanced_options.minCharge
        --maxCharge=$advanced_options.maxCharge
        #if $advanced_options.ms2Boundaries.specify == 'specify':
            --ms2Boundaries="$advanced_options.ms2Boundaries.ms2BoundariesMin,$advanced_options.ms2Boundaries.ms2BoundariesMax"
        #end if
        --alpha=$advanced_options.alpha
        --beta=$advanced_options.beta
        --idotp=$advanced_options.idotp
        --topX=$advanced_options.topX
        #if $advanced_options.fixedMod and str($advanced_options.fixedMod):
            --fixedMod='$advanced_options.fixedMod'
        #end if
        $input $peptides  2>1 > run.log
        && cat "$outputRoot"*.log > '$log_file'   
        && cat "$outputRoot"*.td.feature | awk 'NR==1{print $0};!/peakCalibratedScore/{print $0}' > '$feature_file'
    ]]></command>
    
    <configfiles>
        <configfile name="isolationSheme">#slurp
#import re
#if $isolation.isolationsrc == 'entered':
#echo $re.sub(r'(\d+[.]\d+)\s+(\d+[.]\d)\s*',r'\1\t\2\n',$isolation.isolation_windows.strip())
#end if
  #slurp</configfile>
    </configfiles>
    <inputs>
        <param name="input" type="data" format="mzml" label="centroided mzML file"/>
        <param name="sciex" type="boolean" truevalue="--sciex" falsevalue="" checked="false" 
            label="mzML files were generated from AB SCIEX MS Data Converter"/>
        <param name="peptides" type="data" format="tabular" multiple="true" label="target peptides"/>

        <conditional name="isolation">
            <param name="isolationsrc" type="select" label="isolation windows">
                <option value="entered">Enter isolation windows</option>  
                <option value="history">Read isolation windows from a history dataset</option>  
            </param>
            <when value="entered">
                <param name="isolation_windows" type="text" area="true" size="20x20" value="" label="isolation windows">
                    <validator type="regex" message="enter pairs of window boudaries, e.g.:400.0 420.0">^(?ms)(\d+[.]\d+\s\d+[.]\d)(\s\d+[.]\d+\s\d+[.]\d)*$</validator>
                </param>
            </when>
            <when value="history">
                <param name="isolation_windows" type="data" format="tabular" label="isolation windows"/>
            </when>
        </conditional>

        <!-- species - default proteomes for (ecoli/yeast/mouse/human) -->
        <param name="backgroundProteome" type="data" format="tabular" optional="true" label="BackgroundProteome" 
            help="Peptide list file name of background proteome. Recommended if querying for a small number of peptides."/>
        <param name="bgDecoyNumber" type="integer" value="2000" min="0" label="Number of decoys used for background estimation" 
            optional="True" help="(-n) Default= 2000" />
        <param name="decoyTag" type="text" value="decoy" optional="true" label="Tag for decoy sequences in the database"/>

        <param name="ionTypes" type="select" label="Fragment ion types" help="suggested HCD: y, reCID: by">
            <option value="by">by</option>
            <option value="y">y</option>
            <option value="b">b</option>
        </param>
        
        <param name="minElution" type="integer" value="12" min="1" label="Minimum time in seconds a peptide is expected to elute" 
            optional="True" help="(-m) Default= 12 seconds" />
        <param name="maxElution" type="integer" value="30" min="1" label="Maximum time in seconds a peptide is expected to elute" 
            optional="True" help="(-x) Default= 30 seconds" />

        <section name="advanced_options" expanded="false" title="Advanced Options">
            <param name="overlap" type="integer" value="0" min="0" max="100" label="Overlap percentage of isolation window"/>
            <conditional name="ms1">
                <param name="ms1Unit" type="select" label="Unit for precursor ion extraction tolerance">
                    <option value="ppm">ppm</option>
                    <option value="mz">mz</option>
                </param>
                <when value="ppm">
                    <param name="ms1Tolerance" type="integer" value="10" min="0" label="ms1Tolerance" 
                        help="Mass error for precursor ion extracting"/>
                </when>
                <when value="mz">
                    <param name="ms1Tolerance" type="float" value="10" min="0" label="ms1Tolerance" 
                        help="Mass error for precursor ion extracting"/>
                </when>
            </conditional>
            <conditional name="ms2">
                <param name="ms2Unit" type="select" label="Unit for fragment ion extraction tolerance">
                    <option value="ppm">ppm</option>
                    <option value="mz">mz</option>
                </param>
                <when value="ppm">
                    <param name="ms2Tolerance" type="integer" value="10" min="0" label="ms2Tolerance" 
                        help="Mass error for precursor ion extracting"/>
                </when>
                <when value="mz">
                    <param name="ms2Tolerance" type="float" value="10" min="0" label="ms2Tolerance" 
                        help="Mass error for precursor ion extracting"/>
                </when>
            </conditional>
            <param name="minCharge" type="integer" value="2" min="1" label="Minimum charge state searched"/>
            <param name="maxCharge" type="integer" value="3" min="1" label="Maximum charge state searched"/>
            <conditional name="ms2Boundaries">
                <param name="specify" type="select" label="Unit for fragment ion extraction tolerance">
                    <option value="calculate">Let PECAN calculate MS2 Boundaries</option>
                    <option value="specify">Specify MS2 Boundaries</option>
                </param>
                <when value="calculate"/>
                <when value="specify">
                    <param name="ms2BoundariesMin"  type="float" value="" label="Specify the MS2 scan lower m/z limit for the target window"/>
                    <param name="ms2BoundariesMax"  type="float" value="" label="Specify the MS2 scan upper m/z limit for the target window"/>
                </when>
            </conditional>
            <param name="alpha" type="float" value="1.8" min="0.0" label="Alpha hyperparameter to set score threshold for cIons" />
            <param name="beta" type="float" value="0.4" min="0.0" max="1.0" label="Beta hyperparameter to set score threshold for cIons" />
            <param name="idotp" type="float" value="0.0" min="0.0" label="MS1 idotp threshold used for peak picking" />
            <param name="topX" type="integer" value="1" min="1" label="Reporting top x peaks per charged peptide"/>
            
            <param name="fixedMod" type="text" value="" optional="true" label="Fixed modifications in the sample">
                <help><![CDATA[
                    Fixed modifications in the sample. For example, default C[+57.021] will treat every Cys
                    in the query sequences and background proteome as C[+57.021] while overwriting any
                    specific modifications on individual Cys
                    If multiple fixed modifications are desired use "," to separate each fixedMod tag
                    (e.g. C[+57.021],M[+15.995])
                ]]></help>
                <validator type="regex" message="C[+57.021],M[+15.995]">^([A-Z]\[[+-]\d+[.]\d+\](,[A-Z]\[[+-]\d+[.]\d+\])*)*$</validator>
                <sanitizer sanitize="False"/>
            </param>
        </section>
        <!--
        verbosity#Verbosity level for file logging (WARNING/INFO/DEBUG) Default=INFO
        -->
    </inputs>
    <outputs>
        <data name="log_file" format="txt" label="" />
        <data name="feature_file" format="percin" label="" />
    </outputs>
    <help><![CDATA[
**PECAN**

PECAN (PEptide-Centric ANalysis) is a tool for peptide detection directly from DIA data without the need of a spectral library. PECAN takes a list of peptide sequences and query each peptide against the DIA data (in centroid mzML format) and reports the best evidence of detection for each peptide. The PECAN report is designed for Percolator to separate correct from incorrect matches with false discovery rate (FDR) control.

usage: pecan [-h] [-d DECOYTAG] [--ms1Unit MS1UNIT] [--ms2Unit MS2UNIT]
             [-p MS1TOLERANCE] [-q MS2TOLERANCE] [-m MINELUTION]
             [-x MAXELUTION] [-b BACKGROUNDPROTEOME] [-n BGDECOYNUMBER]
             [-i IDOTP] [-v VERBOSITY] [-t TOPX] [--minCharge MINCHARGE]
             [--maxCharge MAXCHARGE] [-s IONTYPES] [--overlap OVERLAP]
             [--SCIEX] [--VERSION] [--ms2Boundaries MS2BOUNDARIES]
             [--alpha ALPHA] [--beta BETA] [--fixedMod FIXEDMOD]
             mzMLFileName peptideListFileName isolationStart isolationEnd
             outputRoot

        

positional arguments:
  mzMLFileName          path to the centroid .mzML file
  peptideListFileName   path to the query peptide list
  isolationStart        precursor m/z of isolation start
  isolationEnd          precursor m/z of isolation end
  outputRoot            name base of the output files

optional arguments:
  -h, --help            show this help message and exit
  -d DECOYTAG, --decoyTag DECOYTAG
                        Tag for decoy sequences in the database
                        Default= decoy
                        
  --ms1Unit MS1UNIT     Unit (ppm/mz) for precursor ion extraction tolerance
                        Used with -p
                        Default=ppm 
                        
  --ms2Unit MS2UNIT     Unit (ppm/mz) for fragment ion extraction tolerance
                        Used with -q
                        Default=ppm 
                        
  -p MS1TOLERANCE, --ms1Tolerance MS1TOLERANCE
                        Mass error for precursor ion extracting
                        Default=10 
                        
  -q MS2TOLERANCE, --ms2Tolerance MS2TOLERANCE
                        Mass error for fragment ion extracting
                        Default=10 
                        
  -m MINELUTION, --minElution MINELUTION
                        Minimum time in seconds a peptide is expected to elute
                        Default= 12 seconds
                        
  -x MAXELUTION, --maxElution MAXELUTION
                        Maximum time in seconds a peptide is expected to elute
                        Default= 30 seconds
                        
  -b BACKGROUNDPROTEOME, --backgroundProteome BACKGROUNDPROTEOME
                        Peptide list file name of background proteome
                        Recommended if querying for a small number of peptides
                        Default=None
                        
  -n BGDECOYNUMBER, --bgDecoyNumber BGDECOYNUMBER
                        Number of decoys used for background estimation
                        Default=2000
                        
  -i IDOTP, --idotp IDOTP
                        MS1 idotp threshold used for peak picking
                        Default=0.000000
                        
  -v VERBOSITY, --verbosity VERBOSITY
                        Verbosity level for file logging (WARNING/INFO/DEBUG)
                        Default=INFO
                        
  -t TOPX, --topX TOPX  Reporting top x peaks per charged peptide
                        Default=1
                        
  --minCharge MINCHARGE
                        Minimum charge state searched
                        Default=2
                        
  --maxCharge MAXCHARGE
                        Maximum charge state searched
                        Default=3
                        
  -s IONTYPES, --ionTypes IONTYPES
                        == To be added == Fragment ion types (b, y, or by)
                        Default=by
                        
  --overlap OVERLAP     Overlap percentage of isolation window
                        Default=0 
                        
  --SCIEX               Flag if mzML is generated from AB Sciex converter
  --VERSION             Show current PECAN version
                        
  --ms2Boundaries MS2BOUNDARIES
                        Specify the MS2 scan lower and upper m/z limit such as 200,2000 for the target window
                        If not specify pecan will determine it with a function
                        
  --alpha ALPHA         Alpha hyperparameter to set score threshold for cIons
                        
  --beta BETA           Beta hyperparameter to set cutoff for number of cIons
                        
  --fixedMod FIXEDMOD   Fixed modifications in the sample. For example, default C[+57.021] will treat every Cys
                        in the query sequences and background proteome as C[+57.021] while overwriting any
                        specific modifications on individual Cys
                        If multiple fixed modifications are desired use "," to separate each fixedMod tag
                        (e.g. C[+57.021],M[+15.995])
                        

    ]]></help>
</tool>