view maldi_quant_preprocessing.xml @ 4:82b1de5c142b draft default tip

planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/MALDIquant commit f1e1cd260ef2884d0ba12e2b614df3c72d0934dc

<tool id="maldi_quant_preprocessing" name="MALDIquant preprocessing" version="@VERSION@.0">
    <description>
        Preprocessing of mass-spectrometry imaging data
    </description>
    <macros>
        <import>maldi_macros.xml</import>
    </macros>
    <expand macro="requirements"/>
    <command detect_errors="exit_code">
    <![CDATA[
        cat '${maldi_quant_preprocessing}' &&
        #if $infile.ext == 'imzml'
            cp '${infile.extra_files_path}/imzml' infile.imzML &&
            cp '${infile.extra_files_path}/ibd' infile.ibd &&
        #elif $infile.ext == 'analyze75'
            cp '${infile.extra_files_path}/hdr' infile.hdr &&
            cp '${infile.extra_files_path}/img' infile.img &&
            cp '${infile.extra_files_path}/t2m' infile.t2m &&
            du infile.hdr &&
            du infile.img &&
            du -s -B1 infile.hdr &&
        #end if
        Rscript "${maldi_quant_preprocessing}" &&

        mkdir $outfile_imzml.files_path &&
        mv ./out.imzMl "${os.path.join($outfile_imzml.files_path, 'imzml')}" | true &&
        mv ./out.ibd "${os.path.join($outfile_imzml.files_path, 'ibd')}" | true &&
        echo "imzML file:" > $outfile_imzml &&
        ls -l "$outfile_imzml.files_path" >> $outfile_imzml
    ]]>
    </command>
    <configfiles>
        <configfile name="maldi_quant_preprocessing"><![CDATA[

@R_IMPORTS@

#if $restriction_conditional.restriction == 'restrict':

    print('Reading mask region')
    ## Import imzML file

    coordinate_matrix = as.matrix(read.delim("$restriction_conditional.coordinates_file", header = $restriction_conditional.coordinates_header, stringsAsFactors = FALSE))[,1:2]
    coordinate_matrix = coordinate_matrix[,c($restriction_conditional.column_x, $restriction_conditional.column_y)]

    maldi_data = importImzMl('infile.imzML',
                     coordinates = coordinate_matrix)
    pixelnames = paste0("x = ", coordinates(maldi_data)[,1],", y = ", coordinates(maldi_data)[,2])

#else:

    print('Reading entire file')
    #if $infile.ext == 'imzml'
        ## Import imzML file
        maldi_data = import( 'infile.imzML', type="imzML" )
        coordinates_info = cbind(coordinates(maldi_data)[,1:2], c(1:length(maldi_data)))
    #elif $infile.ext == 'analyze75'
        ## Import analyze7.5 file
        maldi_data = importAnalyze( 'infile.hdr' )
        coordinates_info = cbind(coordinates(maldi_data)[,1:2], c(1:length(maldi_data)))
    #end if

#end if

## Quality control plots during preprocessing

pdf("prepro_qc_plot.pdf", fonts = "Times", pointsize = 12)
plot(0,type='n',axes=FALSE,ann=FALSE)

## if no filename is given, name of file in Galaxy history is used
    #set $filename = $infile.display_name
title(main=paste("$filename"))


#################### Preprocessing methods #####################################

## QC plot on input file
avgSpectra = averageMassSpectra(maldi_data,method="mean")
plot(avgSpectra, main="Average spectrum for input file")

pixel_number = length(maldi_data)
minmz = round(min(unlist(lapply(maldi_data,mass))), digits=4)
maxmz = round(max(unlist(lapply(maldi_data,mass))), digits=4)
mean_features = round(length(unlist(lapply(maldi_data,mass)))/length(maldi_data), digits=2)
number_features = length(unique(unlist(lapply(maldi_data,mass))))
medint = round(median(unlist(lapply(maldi_data,intensity))), digits=2)
inputdata = c(minmz, maxmz,number_features,mean_features,medint, pixel_number)
QC_numbers= data.frame(inputdata = c(minmz, maxmz,number_features, mean_features, medint, pixel_number))
vectorofactions = "inputdata"


#for $method in $methods:

    #if str( $method.methods_conditional.method ) == 'Transformation':

        print('transforming')
        ##transformation
        maldi_data = transformIntensity(maldi_data, method="$method.methods_conditional.transform_method")
        ## QC plot and numbers
        avgSpectra = averageMassSpectra(maldi_data,method="mean")
        plot(avgSpectra, main="Average spectrum after transformation")
        pixel_number = length(maldi_data)
        minmz = round(min(unlist(lapply(maldi_data,mass))), digits=4)
        maxmz = round(max(unlist(lapply(maldi_data,mass))), digits=4)
        mean_features = round(length(unlist(lapply(maldi_data,mass)))/length(maldi_data), digits=2)
        medint = round(median(unlist(lapply(maldi_data,intensity))), digits=2)
        number_features = length(unique(unlist(lapply(maldi_data,mass))))
        transformed = c(minmz, maxmz,number_features,mean_features,medint,pixel_number)
        QC_numbers= cbind(QC_numbers, transformed)
        vectorofactions = append(vectorofactions, "transformed")


    #elif str( $method.methods_conditional.method ) == 'Smoothing':

        print('smoothing')
        ##smoothing

        #if str($method.methods_conditional.methods_for_smoothing.smooth_method ) == 'SavitzkyGolay':
        print('SavitzkyGolay')

            maldi_data = smoothIntensity(maldi_data,
                    method="SavitzkyGolay", polynomialOrder=$method.methods_conditional.methods_for_smoothing.polynomial,
                    halfWindowSize=$method.methods_conditional.halfWindowSize)

        #elif str($method.methods_conditional.methods_for_smoothing.smooth_method ) == 'MovingAverage':
            print('MovingAverage')

            maldi_data = smoothIntensity(maldi_data,
                    method="MovingAverage", weighted=$method.methods_conditional.methods_for_smoothing.weighted,
                    halfWindowSize=$method.methods_conditional.halfWindowSize)

        #end if

        ## QC plot and numbers
        avgSpectra = averageMassSpectra(maldi_data,method="mean")
        plot(avgSpectra, main="Average spectrum after smoothing", sub="")
        pixel_number = length(maldi_data)
        minmz = round(min(unlist(lapply(maldi_data,mass))), digits=4)
        maxmz = round(max(unlist(lapply(maldi_data,mass))), digits=4)
        mean_features = round(length(unlist(lapply(maldi_data,mass)))/length(maldi_data), digits=2)
        medint = round(median(unlist(lapply(maldi_data,intensity))), digits=2)
        number_features = length(unique(unlist(lapply(maldi_data,mass))))
        smoothed = c(minmz, maxmz,number_features,mean_features,medint,pixel_number)
        QC_numbers= cbind(QC_numbers, smoothed)
        vectorofactions = append(vectorofactions, "smoothed")


    #elif str( $method.methods_conditional.method ) == 'Baseline':

        print('baseline removing')
        ## Remove baseline

        ## Choose random spectra for QC plots
        random_spectra = sample(1:length(maldi_data), 4, replace=FALSE)

        #if str($method.methods_conditional.methods_for_baseline.baseline_method ) == 'SNIP':
        print('SNIP')

            par(mfrow = c(2,2))
            for (random_sample in random_spectra){
                maldi_data_baseline = estimateBaseline(maldi_data[[random_sample]],
                    method="SNIP", iterations=$method.methods_conditional.methods_for_baseline.iterations)
                plot(maldi_data[[random_sample]], sub="", main=paste0("Estimated baseline for spectrum ", random_sample))
                lines(maldi_data_baseline, col="blue", lwd=2)}

            maldi_data = removeBaseline(maldi_data,
                method="SNIP",
                iterations=$method.methods_conditional.methods_for_baseline.iterations)


        #elif str($method.methods_conditional.methods_for_baseline.baseline_method ) == 'TopHat':
        print('TopHat')

            par(mfrow = c(2,2))
            for (random_sample in random_spectra){
                maldi_data_baseline = estimateBaseline(maldi_data[[random_sample]],
                    method="TopHat", halfWindowSize=$method.methods_conditional.methods_for_baseline.tophat_halfWindowSize)
                plot(maldi_data[[random_sample]], sub="", main=paste0("Estimated baseline for spectrum ", random_sample))
                lines(maldi_data_baseline, col="blue", lwd=2)}

            maldi_data = removeBaseline(maldi_data,
                method="TopHat",
                halfWindowSize=$method.methods_conditional.methods_for_baseline.tophat_halfWindowSize)


        #elif str($method.methods_conditional.methods_for_baseline.baseline_method ) == 'ConvexHull':
        print('ConvexHull')

            par(mfrow = c(2,2))
            for (random_sample in random_spectra){
                maldi_data_baseline = estimateBaseline(maldi_data[[random_sample]],
                    method="ConvexHull")
                plot(maldi_data[[random_sample]], sub="", main=paste0("Estimated baseline for spectrum ", random_sample))
                lines(maldi_data_baseline, col="blue", lwd=2)}

            maldi_data = removeBaseline(maldi_data,
                method="ConvexHull")


        #elif str($method.methods_conditional.methods_for_baseline.baseline_method ) == 'median':
        print('median')

            par(mfrow = c(2,2))
            for (random_sample in random_spectra){
                maldi_data_baseline = estimateBaseline(maldi_data[[random_sample]],
                    method="median", halfWindowSize=$method.methods_conditional.methods_for_baseline.median_halfWindowSize)
                plot(maldi_data[[random_sample]], sub="", main=paste0("Estimated baseline for spectrum ", random_sample))
                lines(maldi_data_baseline, col="blue", lwd=2)}

            maldi_data = removeBaseline(maldi_data,
                method="median",
                halfWindowSize=$method.methods_conditional.methods_for_baseline.median_halfWindowSize)

        #end if

        ## QC plot and numbers
        par(mfrow = c(1,1))
        avgSpectra = averageMassSpectra(maldi_data,method="mean")
        plot(avgSpectra, main="Average spectrum after baseline removal")
        pixel_number = length(maldi_data)
        minmz = round(min(unlist(lapply(maldi_data,mass))), digits=4)
        maxmz = round(max(unlist(lapply(maldi_data,mass))), digits=4)
        mean_features = round(length(unlist(lapply(maldi_data,mass)))/length(maldi_data), digits=2)
        medint = round(median(unlist(lapply(maldi_data,intensity))), digits=2)
        number_features = length(unique(unlist(lapply(maldi_data,mass))))
        baseline_removed = c(minmz, maxmz,number_features,mean_features,medint,pixel_number)
        QC_numbers= cbind(QC_numbers, baseline_removed)
        vectorofactions = append(vectorofactions, "bl_removed")


    #elif str( $method.methods_conditional.method ) == 'Calibrate':

        print('calibrate')
        ##calibrate

        #if str($method.methods_conditional.cond_calibration_range) == "yes":
            ## calibrate only given m/z range
            maldi_data = calibrateIntensity(maldi_data,
                method="$method.methods_conditional.calibrate_method",
                range=c($method.methods_conditional.cond_calibration_range.mass_start, $method.methods_conditional.cond_calibration_range.mass_end))
        #else:
            maldi_data = calibrateIntensity(maldi_data,
            method="$method.methods_conditional.calibrate_method")
        #end if
        ## QC plot and numbers
        avgSpectra = averageMassSpectra(maldi_data,method="mean")
        plot(avgSpectra, main="Average spectrum after normalization")
        pixel_number = length(maldi_data)
        minmz = round(min(unlist(lapply(maldi_data,mass))), digits=4)
        maxmz = round(max(unlist(lapply(maldi_data,mass))), digits=4)
        mean_features = round(length(unlist(lapply(maldi_data,mass)))/length(maldi_data), digits=2)
        medint = round(median(unlist(lapply(maldi_data,intensity))), digits=2)
        number_features = length(unique(unlist(lapply(maldi_data,mass))))
        intensity_calibrated = c(minmz, maxmz,number_features,mean_features,medint,pixel_number)
        QC_numbers= cbind(QC_numbers, intensity_calibrated)
        vectorofactions = append(vectorofactions, "calibrated")


    #elif str( $method.methods_conditional.method ) == 'Align':

        print('align')
        ##align spectra with 3 separate functions

        ## create reference if needed

        ## 1) detect peaks:
        peaks <- detectPeaks(maldi_data, halfWindowSize=$method.methods_conditional.halfWindowSize,
                       method="$method.methods_conditional.peak_method", SNR=$method.methods_conditional.snr)


        #if str($method.methods_conditional.reference_for_alignment.align_ref) == 'no_reference':

        ## 2) calculate warping:
            warping_function <- determineWarpingFunctions(peaks,
                                  tolerance=$method.methods_conditional.tolerance, method="$method.methods_conditional.warping_method",
                                  allowNoMatches=$method.methods_conditional.allow_nomatch, minFrequency = $method.methods_conditional.reference_for_alignment.min_frequency)

        ## 3) warp spectra:
            maldi_data = warpMassSpectra(maldi_data, warping_function, emptyNoMatches=$method.methods_conditional.empty_nomatch)


        #elif str($method.methods_conditional.reference_for_alignment.align_ref) == 'yes_reference':

            ## create reference mass_vector from tabular file
            mass_vector = read.delim("$method.methods_conditional.reference_for_alignment.reference_file", header = $method.methods_conditional.reference_for_alignment.reference_header, stringsAsFactors = FALSE)[,$method.methods_conditional.reference_for_alignment.mz_column]
            int_vector = rep(1,length(mass_vector))
            mass_list = createMassPeaks(mass_vector, int_vector)

            #if str($method.methods_conditional.reference_for_alignment.separate_alignment) == "FALSE"
            print('default alignment')

                ## 2) calculate warping:
                    warping_function <- determineWarpingFunctions(peaks,
                                          tolerance=$method.methods_conditional.tolerance, method="$method.methods_conditional.warping_method",
                                          allowNoMatches=$method.methods_conditional.allow_nomatch, reference = mass_list)

                ## 3) warp spectra:
                    maldi_data = warpMassSpectra(maldi_data, warping_function, emptyNoMatches=$method.methods_conditional.empty_nomatch)

            #elif str($method.methods_conditional.reference_for_alignment.separate_alignment) == "TRUE"
            print('spectra wise alignment')

                maldi_data_new_list =list()

                for (pixelnb in 1:length(peaks))
                {
                    ## 2) calculate warping:
                        warping_function <- determineWarpingFunctions(peaks[[pixelnb]],
                                              tolerance=$method.methods_conditional.tolerance, method="$method.methods_conditional.warping_method",
                                              allowNoMatches=$method.methods_conditional.allow_nomatch, reference = mass_list)

                    ## 3) warp spectra:
                        maldi_data_new = warpMassSpectra(list(maldi_data[[pixelnb]]), warping_function, emptyNoMatches=$method.methods_conditional.empty_nomatch)
                        maldi_data_new_list = c(maldi_data_new_list, maldi_data_new)

                }
                maldi_data = maldi_data_new_list
            #end if
        #end if


        #if $method.methods_conditional.remove_empty:
            print(paste(length(findEmptyMassObjects(maldi_data)), " empty spectra were removed", sep=" "))

            ## only if there are empty spectra to remove

            if (length(findEmptyMassObjects(maldi_data))>0)

                {
                    maldi_data = removeEmptyMassObjects(maldi_data)
            }
        #end if

        ## QC plot

        if (length(maldi_data)>0){
            avgSpectra = averageMassSpectra(maldi_data,method="mean")
            plot(avgSpectra, main="Average spectrum after alignment")
            medint = round(median(unlist(lapply(maldi_data,intensity))), digits=2)
        }else{print("All spectra are empty")}

        pixel_number = length(maldi_data)
        minmz = round(min(unlist(lapply(maldi_data,mass))), digits=4)
        maxmz = round(max(unlist(lapply(maldi_data,mass))), digits=4)
        mean_features = round(length(unlist(lapply(maldi_data,mass)))/length(maldi_data), digits=2)

        number_features = length(unique(unlist(lapply(maldi_data,mass))))
        spectra_aligned = c(minmz, maxmz,number_features,mean_features, medint,pixel_number)
        QC_numbers= cbind(QC_numbers, spectra_aligned)
        vectorofactions = append(vectorofactions, "aligned")

    #elif str( $method.methods_conditional.method ) == 'skip_preprocessing':
        ##for now as option to filter large files

    #end if

#end for

rownames(QC_numbers) = c("min m/z", "max mz", "# features", "median \n# peaks (int.>0)", "median\nintensity", "pixel\nnumber")
colnames(QC_numbers) = vectorofactions
plot(0,type='n',axes=FALSE,ann=FALSE)
grid.table(t(QC_numbers))

dev.off()

## export imzML file
if (length(maldi_data)>0){
        MALDIquantForeign::exportImzMl(maldi_data, file="out.imzMl", processed=$export_processed)

}else{"All spectra are empty, outputfiles will be empty,too."}

    ]]>
        </configfile>
    </configfiles>
    <inputs>
        <param name="infile" type="data" format="imzml,analyze75" label="MSI data" help="Input file as imzML (composite upload). The file must be in profile mode, not centroided."/>
        <conditional name="restriction_conditional">
            <param name="restriction" type="select" label="Use only spectra of interest" help="This option only works for imzML files">
                <option value="no_restriction" selected="True">No, calculate on entire file</option>
                <option value="restrict">Yes, restrict to spectra of interest</option>
            </param>
            <when value="restrict">
                <param name="coordinates_file" type="data" format="tabular" label="Tabular file with coordinates"/>
                <param name="column_x" data_ref="coordinates_file" label="Column with x values" type="data_column"/>
                <param name="column_y" data_ref="coordinates_file" label="Column with y values" type="data_column"/>
                <param name="coordinates_header" type="boolean" label="File contains a header line" truevalue="TRUE" falsevalue="FALSE"/>
            </when>              
            <when value="no_restriction"/>
        </conditional>
        <repeat name="methods" title="Method" min="1">
            <conditional name="methods_conditional">
                <param name="method" type="select" label="Select a method">
                    <option value="Transformation" selected="True">Transformation</option>
                    <option value="Smoothing">Smoothing</option>
                    <option value="Baseline">Baseline removal</option>
                    <option value="Calibrate">Intensity calibration (normalization)</option>
                    <option value="Align">Align spectra (warping/phase correction)</option>
                    <option value="skip_preprocessing">Skip preprocessing</option>
                    <validator type="empty_field" />
                </param>
                <when value="Transformation">
                    <param name="transform_method" type="select" label="Transformation method">
                        <option value="sqrt" selected="True">sqrt</option>
                        <option value="log">log</option>
                        <option value="log2">log2</option>
                        <option value="log10">log10</option>
                        <validator type="empty_field" />
                    </param>
                </when>
                <when value="Smoothing">
                    <conditional name="methods_for_smoothing">
                        <param name="smooth_method" type="select" label="Smoothing method" help="This method smoothes the intensity values of a MassSpectrum object.">
                            <option value="SavitzkyGolay" selected="True">SavitzkyGolay</option>
                            <option value="MovingAverage">MovingAverage</option>
                        </param>
                        <when value="SavitzkyGolay">
                            <param name="polynomial" value="3" type="text" label="Polynomial order"
                                    help="Controls the order of the filter, should be smaller than the resulting window."/>
                        </when>
                        <when value="MovingAverage">
                            <param name="weighted" type="boolean" label="Weighted average" help = "Indicates if the average should be equal weight or if it should have weights depending on the distance from the center as calculated as 1/2^abs(-halfWindowSize:halfWindowSize) with the sum of all weigths normalized to 1" truevalue="TRUE" falsevalue="FALSE"/>
                        </when>
                    </conditional>
                    <param name="halfWindowSize" type="integer" value="10"
                        label="Half window size"
                        help="Number of data points, the resulting window reaches from 
                            mass[currentIndex-halfWindowSize] to mass[currentIndex+halfWindowSize]
                            (window size is 2*halfWindowSize+1)."/>
                </when>
                <when value="Baseline">
                    <conditional name="methods_for_baseline">
                        <param name="baseline_method" type="select" label="Baseline removal method">
                            <option value="SNIP" selected="True">SNIP</option>
                            <option value="TopHat">TopHat</option>
                            <option value="ConvexHull">ConvexHull</option>
                            <option value="median">median</option>
                            <validator type="empty_field" />
                        </param>
                        <when value="SNIP">
                            <param name="iterations" type="integer" value="100"
                        label="Number of iterations" help="Corresponds to half window size: The resulting window reaches from mass[cur_index-iterations] to mass[cur_index+iterations]"/>
                        </when>
                        <when value="TopHat">
                            <param name="tophat_halfWindowSize" type="integer" value="10"
                        label="Half window size"
                            help="Number of data points, the resulting window reaches from 
                            mass[currentIndex-halfWindowSize] to mass[currentIndex+halfWindowSize]"/>
                        </when>
                        <when value="ConvexHull"/>
                        <when value="median">
                            <param name="median_halfWindowSize" type="integer" value="10"
                        label="Half window size"
                            help="Number of data points, the resulting window reaches from 
                            mass[currentIndex-halfWindowSize] to mass[currentIndex+halfWindowSize]"/>
                        </when>
                    </conditional>
                </when>
                <when value="Calibrate">
                    <param name="calibrate_method" type="select" label="Intensity calibration method" help="Intensity normalization">
                        <option value="TIC" selected="True">TIC</option>
                        <option value="PQN">PQN</option>
                        <option value="median">median</option>
                        <validator type="empty_field" />
                    </param>
                    <conditional name="cond_calibration_range">
                        <param name="calibration_range" type="select" label="m/z range" help="Instead of the whole m/z range, a specified m/z range can be used to calculate the scaling factor.">
                            <option value="no" selected="True">complete m/z range</option>
                            <option value="yes">specify a m/z range</option>
                        </param>
                        <when value="no"/>
                        <when value="yes">
                            <param name="mass_start" type="integer" value="800"
                                label="Start of m/z range, has to be inside m/z range" 
                                help="Scaling factor is calculated on the mass range and applied to the whole spectrum."/>
                            <param name="mass_end" type="integer" value="3000"
                                label="End of m/z range, has to be inside m/z range"/>
                        </when>
                    </conditional>
                </when>
                <when value="Align">
                    <param name="warping_method" type="select" label="Alignment method">
                        <option value="lowess" selected="True">Lowess</option>
                        <option value="linear">Linear</option>
                        <option value="quadratic">Quadratic</option>
                        <option value="cubic">Cubic</option>
                    </param>

                    <param name="tolerance" type="float" value="0.00005"
                        label="Tolerance"
                        help="abs(mz1 - mz2)/mz2, maximal relative deviation of a peak position (m/z) to be considered as identical. For 50ppm use 0.00005 or 5e-5" />

                    <param name="halfWindowSize" type="integer" value="20"
                        label="Half window size"
                        help="Number of data points, the resulting window reaches from 
                            mass[currentIndex-halfWindowSize] to mass[currentIndex+halfWindowSize]
                            (window size is 2*halfWindowSize+1).
                            The best size differs depending on the selected smoothing method."/>

                    <param name="peak_method" type="select" label="Noise estimation function">
                        <option value="MAD" selected="True">MAD</option>
                        <option value="SuperSmoother">SuperSmoother</option>
                    </param>

                    <param name="snr" type="integer" value="2" label="Signal-to-noise-ratio"/>
                    <param name="allow_nomatch" type="boolean" label="Allow no matches" help="Don't throw an error when less than 2 reference m/z were found in a spectrum" truevalue="TRUE" falsevalue="FALSE"/>
                    <param name="empty_nomatch" type="boolean" label="Empty no matches" help="If TRUE the intensity values of MassSpectrum or MassPeaks objects with missing (NA) warping functions are set to zero" truevalue="TRUE" falsevalue="FALSE"/>
                    <param name="remove_empty" type="boolean" label="Remove empty spectra" truevalue="TRUE" falsevalue="FALSE"/>

                    <conditional name="reference_for_alignment">
                        <param name="align_ref" type="select" label="Reference" help="If given, samples will be aligned to reference, use internal calibrants to perform m/z calibration">
                            <option value="no_reference" selected="True">no reference</option>
                            <option value="yes_reference">reference from tabular file</option>
                        </param>
                        <when value="no_reference">
                    <param name="min_frequency" type="float" value="0.9" label = "minFrequency" help="Removal of all peaks which occur in less than minFrequency spectra to generate the reference m/z"/>
                        </when>
                        <when value="yes_reference">
                            <param name="reference_file" type="data" format="tabular"
                                label="Reference m/z values"
                                help="Tabular file"/>
                            <param name="mz_column" data_ref="reference_file" label="Column with m/z values" type="data_column"/>
                            <param name="reference_header" type="boolean" label="File contains a header line" truevalue="TRUE" falsevalue="FALSE"/>
                    <param name="separate_alignment" type="boolean" label="Spectrum wise alignment" help="Internal binning is omitted to avoid interaction between spectra" truevalue="TRUE" falsevalue="FALSE"/>
                        </when>
                    </conditional>
                </when>
                <when value="skip_preprocessing"/>
            </conditional>
        </repeat>
        <param name="export_processed" type="boolean" label="Export processed imzML" help="otherwise continuous imzML will be exported" truevalue="TRUE" falsevalue="FALSE"/>
    </inputs>
    <outputs>
        <data format="imzml" name="outfile_imzml" label="${tool.name} on ${on_string}" />
        <data format="pdf" name="plots" from_work_dir="prepro_qc_plot.pdf" label="${tool.name} on ${on_string}: QC"/>
    </outputs>
    <tests>
        <test>
            <param name="infile" value="" ftype="imzml">
                <composite_data value="Example_Continuous.imzML"/>
                <composite_data value="Example_Continuous.ibd"/>
            </param>
            <conditional name="restriction_conditional">
                <param name="restriction" value="restrict"/>
                <param name="coordinates_file" value="restricted_pixels.tabular"/>
                <param name="column_x" value="1"/>
                <param name="column_y" value="2"/>
            </conditional>
            <conditional name="methods_conditional">
                <param name="method" value="Transformation"/>
                <param name="transform_method" value="log2"/>
                <param name="method" value="Smoothing"/>
                <param name="smooth_method" value="SavitzkyGolay"/>
                <param name="method" value="Basline"/>
                <param name="baseline_method" value ="TopHat"/>
            </conditional>
            <output name="outfile_imzml" ftype="imzml" file="preprocessing1.imzml.txt" lines_diff="4">
                <extra_files type="file" file="outfile1.imzml" name="imzml" lines_diff="8"/>
                <extra_files type="file" file="outfile1.ibd" name="ibd" compare="sim_size"/>
            </output>
            <output name="plots" file="Preprocessing1_QC.pdf" compare="sim_size"/>
        </test>
        <test>
            <param name="infile" value="" ftype="imzml">
                <composite_data value="Example_Continuous.imzML"/>
                <composite_data value="Example_Continuous.ibd"/>
            </param>
            <conditional name="methods_conditional">
                <param name="method" value="Calibrate"/>
                <param name="calibrate_method" value="median"/>
            </conditional>
            <output name="outfile_imzml" ftype="imzml" file="preprocessing3.imzml.txt" lines_diff="4">
                <extra_files type="file" file="outfile3.imzml" name="imzml" lines_diff="8"/>
                <extra_files type="file" file="outfile3.ibd" name="ibd" compare="sim_size"/>
            </output>
            <output name="plots" file="Preprocessing3_QC.pdf" compare="sim_size"/>
        </test>
    </tests>
    <help><![CDATA[

@MADLI_QUANT_DESCRIPTION@

-----

**Input data**

- MSI data: 2 types of input data can be used:

    - imzml file (upload imzml and ibd file via the "composite" function) `Introduction to the imzml format <https://ms-imaging.org/imzml/>`_

- Optional: Tabular file with pixel coordinates to restrict reading of imzML files to coordinates of interest. Tabular files with any header name or no header at all are supported. 

    ::

       x_coord     y_coord 
         1            1    
         2            1    
         3            1    
        ...
        ...

- Optional: Tabular file with reference m/z for the spectra align function. At least 2 m/z values of the input list must be present in every spectrum to peform the alignment. First column must contain m/z values, without empty fields or letters. Tabular files with any header name or no header at all are supported. 

    ::

    
            m/z
          100.0
          100.01
          100.02
        ...
        ...    



**Options**

- Transformation: Variance stabilization through intensity transformation:'log', 'log2', 'log10' and 'squareroot' (sqrt) are available
- Smoothing: Smoothing of the peaks reduces noise and improves peak detection. Available smoothing methods are 'SavitzkyGolay' and 'Moving Average'

    - For all smoothing methods: The larger the 'Half window size', the stronger the smoothing. The resulting window should be smaller than the FWHM (full width at half maximum) of the typical peaks. Moving average needs smaller window size than SavitzkyGolay.
    - Moving average: Recommended for broader peaks/high m/z range spectra. Weighted moving average: Points in the center get larger weight factors than points away from the center.  
    - SavitzkyGolay: Recommended for sharp peaks/low m/z range, preserves the shape of the local maxima. The PolynomialOrder should be smaller than the resulting window. Negative values will be replaced with 0. 

- Baseline reduction: Baseline reduction removes background intensity generated by chemical noise (common in MALDI datasets). 

    - Available methods are SNIP, TopHat,ConvexHull and median:
    - SNIP is the default baseline reduction method in MALDIquant. 
    - ConvexHull is not appropriate for MALDI-TOF baseline removal. 
    - The moving median may generate negative intensities. 
    - Except for the ConvexHull all methods have a parameter for the 'Half window size' (in SNIP it is called 'iterations'). The smaller the window the more baseline will be removed but also parts of the peaks. Wider windows preserve the peak height better and produce a smoother baseline, but some local background variation will remain. 

- Intensity calibration (normalization): Normalization of intensities to Total Ion Current (TIC), median spectrum, Probabilistic Quotient Normalization (PQN)

    - TIC and median are local calibration methods: each spectrum is normalized on its own (each peak is divided by the TIC or median of the spectrum)
    - PQN is a global calibration method: In PQN all spectra are calibrated using the TIC calibration first. Subsequently, a median reference spectrum is created and the intensities in all spectra are standardized using the reference spectrum and a spectrum-specific median is calculated for each spectrum. Finally, each spectrum is rescaled by the median of the ratios of its intensity values and that of the reference spectrum

- Spectra alignment (warping): alignment for (re)calibration of m/z values. 

    - peak detection is performed, the reference peaks will be matched to those detected peaks
    - without external reference m/z: internal reference is obtained by filtering and binning the picked peaks to find landmark peaks and their average m/z
    - with external reference m/z: the given m/z are used as a reference, at least 10 reference values are recommended
    - non linear warping (parametric time warping plus binning) to match the reference peaks (internal or external) to the present (picked) peaks with the given tolerance. At least two m/z per spectrum are needed for the alignment. To prevent an error when this criterium is not fullfilled, "Don't throw an error when less than 2 reference m/z were found in a spectrum" should be set to yes. If the not aligned spectra should be set to zero select yes in "If TRUE the intensity values of MassSpectrum or MassPeaks objects with missing (NA) warping functions are set to zero". In order to remove such empty spectra set "Should empty spectra be removed" to yes. 


**Output**

- imzML file (imzML format can be continuous or processed)
- PDF with average mass spectra after each preprocessing step

.. _MALDIquant: http://strimmerlab.org/software/maldiquant/

        ]]>
    </help>
    <expand macro="citation"/>
</tool>