comparison maldi_quant_peakdetection.xml @ 4:c6b47c89a2f5 draft

planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/MALDIquant commit ecdc3a64aa245d80dbc5487b2bf10a85a43adc6d

3:c42549f04fdd

<tool id="maldi_quant_peak_detection" name="MALDIquant peak detection" version="@VERSION@.3">
    <description>
        Peak detection, binning and filtering for mass-spectrometry imaging data
    </description>
    <macros>
        <import>maldi_macros.xml</import>

title(main=paste("$filename"))

## plot input file spectrum: 
#if $centroids:
    ## Choose random spectra for QC plots
    random_spectra = sample(1:length(peaks), 4, replace=FALSE)
    par(mfrow = c(2, 2), oma=c(0,0,2,0))
    for (random_sample in random_spectra){
        plot(peaks[[random_sample]],sub="", main=paste0("spectrum ", random_sample))}
    title("Input spectra", outer=TRUE, line=0)

#else
    ## Choose random spectra for QC plots
    random_spectra = sample(1:length(maldi_data), 4, replace=FALSE)
    par(mfrow = c(2, 2), oma=c(0,0,2,0))
    for (random_sample in random_spectra){
        plot(maldi_data[[random_sample]],sub="", main=paste0("spectrum ", random_sample))}
    title("Input spectra", outer=TRUE, line=0)
#end if


## QC numbers for input file

        ## QC plot and numbers
        ## plot old spectra with baseline in blue and picked peaks in green
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            noise = estimateNoise(maldi_data[[random_sample]], method= "$method.methods_conditional.peak_method")
            plot(maldi_data[[random_sample]], sub="", main=paste0("spectrum ", random_sample))
            lines(noise[,1], noise[,2]*$method.methods_conditional.snr, col="blue")
            points(peaks[[random_sample]], col="green", pch=20)}
            title("S/N in blue and picked peaks in green", outer=TRUE, line=0)

        ## plot new spectrum
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", random_sample))}
        title("Picked peaks", outer=TRUE, line=0)

        pixel_number = length(peaks)
        minmz = round(min(unlist(lapply(peaks,mass))), digits=4)
        maxmz = round(max(unlist(lapply(peaks,mass))), digits=4)

        ## keep peaks to plot them with monoisotopic peaks
        picked_peaks = peaks

        peaks = monoisotopicPeaks(peaks, minCor=$method.methods_conditional.minCor, 
                tolerance=$method.methods_conditional.tolerance,
                distance=$method.methods_conditional.distance, 
                size=$method.methods_conditional.size)

        ## plot old spectrum with picked isotopes as green dots
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(picked_peaks[[random_sample]], sub="", main=paste0("spectrum ", random_sample))
            points(peaks[[random_sample]], col="green", pch=20)}
            title(paste0("Monoisotopic peaks in green"), outer=TRUE, line=0)


        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", random_sample))}
            title("Monoisotopic peaks", outer=TRUE, line=0)

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

        #end if

        ## QC plot and numbers
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", random_sample))}
        title("Aligned spectra", outer=TRUE, line=0)
        minmz = round(min(unlist(lapply(peaks,mass))), digits=4)
        maxmz = round(max(unlist(lapply(peaks,mass))), digits=4)
        mean_features = round(length(unlist(lapply(peaks,mass)))/length(peaks), digits=2)
        medint = round(median(unlist(lapply(peaks,intensity))), digits=2)

        peaks <- binPeaks(peaks, tolerance=$method.methods_conditional.bin_tolerance, method="$method.methods_conditional.bin_method")

        ## QC plot and numbers
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", random_sample))}
        title("Binned spectra", outer=TRUE, line=0)
        minmz = round(min(unlist(lapply(peaks,mass))), digits=4)
        maxmz = round(max(unlist(lapply(peaks,mass))), digits=4)
        mean_features = round(length(unlist(lapply(peaks,mass)))/length(peaks), digits=2)
        medint =round( median(unlist(lapply(peaks,intensity))), digits=2)

        #end if

        ##QC plot and numbers
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", random_sample))}
        title("Filtered spectra", outer=TRUE, line=0)
        minmz = round(min(unlist(lapply(peaks,mass))), digits=4)
        maxmz = round(max(unlist(lapply(peaks,mass))), digits=4)
        mean_features = round(length(unlist(lapply(peaks,mass)))/length(peaks), digits=2)
        medint = round(median(unlist(lapply(peaks,intensity))), digits=2)

        </conditional>
        <repeat name="methods" title="Method" min="1">
            <conditional name="methods_conditional">
                <param name="method" type="select" label="Select a method">
                    <option value="Peak_detection">Peak detection</option>
                    <option value="monoisotopic_peaks">Keep only monoisotopic peaks</option>
                    <option value="Align">Align Spectra (warping/phase correction)</option>
                    <option value="Binning">Binning</option>
                    <option value="Filtering">Filtering</option>
                </param>
                <when value="Peak_detection">

                    <param name="snr" type="integer" value="2" label="Signal-to-noise-ratio" help=""/>
                    <param name="use_annotations" type="boolean" label="Detect peaks on average mass spectra" help="Spectra with same annotation are averaged before peak detection, no imzML output" truevalue="TRUE" falsevalue="FALSE"/>
                </when>
                <when value="monoisotopic_peaks">
                    <param name="minCor" type="float" value="0.95" label="Minimal correlation"
                        help="Minimal correlation between the peak pattern generated by the model and the experimental peaks in the MassPeaks object to be recognized as isotopic pattern"/>
                    <param name="tolerance" type="float" label="Tolerance" value="0.00005"
                        help="Maximal relative deviation of a peak position (m/z) to be considered as identical: abs(((mass[i]+distance)-mass[i+1])/mass[i]) smaller than 'tolerance'. For 50ppm use 0.00005 or 50e-6" />
                    <param name="distance" type="float" label="Distance" value="1.00235" help="Distance between two consecutive peaks in an isotopic pattern. 1.00235 is average distance for polypeptides."/>
                    <param name="size" type="integer" label="Size" value="3" help="Size (length) of isotopic pattern, longer patterns are prefered over shorter ones, min size is 2."/>
                </when>

                <when value="Align">
                    <param name="warping_method" type="select" label="Warping methods">
                        <option value="lowess" selected="True">Lowess</option>

                </when>

                <when value="Binning">
                    <param name="bin_tolerance" type="float" value="0.002" label="Tolerance"
                        help="After the alignment the peak positions (mass) are very similar but not identical. The binning is needed to make similar peak mass values identical."/>
                    <param name="bin_method" display="radio" type="select" label="Bin creation rule" help="strict - creates bins never containing two or more peaks of the sampe sample. relaxed - allows multiple peaks of the same sample in one bin.">
                    <option value="strict" selected="True" >strict</option>
                    <option value="relaxed" >relaxed</option>
                </param>
                </when>
                <when value="Filtering">

            <param name="centroids" value="TRUE"/>
             <repeat name="methods">
                <conditional name="methods_conditional">
                <param name="method" value="monoisotopic_peaks"/>
                <param name="tolerance" value="0.0004"/>
            </conditional>
            </repeat>
            <output name="plots" file="peakdetection2_QC.pdf" compare="sim_size"/>
            <output name="masspeaks" file="masspeaks2.tabular"/>
            <output name="intensity_matrix" file="int2.tabular"/>

    - Detect peaks on average mass spectra: Spectra with the same annotation (taken from the annotation tabular input) are averaged and peak picking is performed on the average spectrum of each annotation group. The exported imzML is empty and cannot be used for further analysis steps. The peaklist and intensity matrix outputs contain the annotation group names with their averaged intensity values. Filtering steps have to be done in the same run as the peak picking.


- Monoisotopic peaks: Keeps only the monoisotopic peaks


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

    - without external reference m/z: internal reference is obtained by filtering (default 90%) and binning the peaks to find landmark peaks and their average m/z
    - with external reference m/z: the m/z provided in a tabular file are used as a reference, at least 10 reference values are recommended

4:c6b47c89a2f5

<tool id="maldi_quant_peak_detection" name="MALDIquant peak detection" version="@VERSION@.4">
    <description>
        Peak detection, binning and filtering for mass-spectrometry imaging data
    </description>
    <macros>
        <import>maldi_macros.xml</import>

title(main=paste("$filename"))

## plot input file spectrum: 
#if $centroids:
    ## Choose random spectra for QC plots
    print(length(peaks))
    random_spectra = sample(1:length(peaks), 4, replace=FALSE)
    random_spectra_name = pixelnames[random_spectra]
    par(mfrow = c(2, 2), oma=c(0,0,2,0))
    for (random_sample in random_spectra){
        plot(peaks[[random_sample]],sub="", main=paste0("spectrum ", pixelnames[random_sample]))}
    title("Input spectra", outer=TRUE, line=0)

#else
    ## Choose random spectra for QC plots
    random_spectra = sample(1:length(maldi_data), 4, replace=FALSE)
    par(mfrow = c(2, 2), oma=c(0,0,2,0))
    for (random_sample in random_spectra){
        plot(maldi_data[[random_sample]],sub="", main=paste0("spectrum ", pixelnames[random_sample]))
    }
    title("Input spectra", outer=TRUE, line=0)
#end if


## QC numbers for input file

        ## QC plot and numbers
        ## plot old spectra with baseline in blue and picked peaks in green
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            noise = estimateNoise(maldi_data[[random_sample]], method= "$method.methods_conditional.peak_method")
            plot(maldi_data[[random_sample]], sub="", main=paste0("spectrum ", pixelnames[random_sample]))
            lines(noise[,1], noise[,2]*$method.methods_conditional.snr, col="blue")
            points(peaks[[random_sample]], col="green", pch=20)}
            title("S/N in blue and picked peaks in green", outer=TRUE, line=0)

        ## plot new spectrum
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", pixelnames[random_sample]))}
        title("Picked peaks", outer=TRUE, line=0)

        pixel_number = length(peaks)
        minmz = round(min(unlist(lapply(peaks,mass))), digits=4)
        maxmz = round(max(unlist(lapply(peaks,mass))), digits=4)

        ## keep peaks to plot them with monoisotopic peaks
        picked_peaks = peaks

        peaks = monoisotopicPeaks(peaks, minCor=$method.methods_conditional.minCor, 
                tolerance=$method.methods_conditional.tolerance,
                distance=c($method.methods_conditional.distance),
                size=$method.methods_conditional.size)

        ## plot old spectrum with picked isotopes as green dots
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(picked_peaks[[random_sample]], sub="", main=paste0("spectrum ", pixelnames[random_sample]))
            points(peaks[[random_sample]], col="green", pch=20)}
            title(paste0("Monoisotopic peaks in green"), outer=TRUE, line=0)


        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", pixelnames[random_sample]))}
            title("Monoisotopic peaks", outer=TRUE, line=0)

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

        #end if

        ## QC plot and numbers
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", pixelnames[random_sample]))}
        title("Aligned spectra", outer=TRUE, line=0)
        minmz = round(min(unlist(lapply(peaks,mass))), digits=4)
        maxmz = round(max(unlist(lapply(peaks,mass))), digits=4)
        mean_features = round(length(unlist(lapply(peaks,mass)))/length(peaks), digits=2)
        medint = round(median(unlist(lapply(peaks,intensity))), digits=2)

        peaks <- binPeaks(peaks, tolerance=$method.methods_conditional.bin_tolerance, method="$method.methods_conditional.bin_method")

        ## QC plot and numbers
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", pixelnames[random_sample]))}
        title("Binned spectra", outer=TRUE, line=0)
        minmz = round(min(unlist(lapply(peaks,mass))), digits=4)
        maxmz = round(max(unlist(lapply(peaks,mass))), digits=4)
        mean_features = round(length(unlist(lapply(peaks,mass)))/length(peaks), digits=2)
        medint =round( median(unlist(lapply(peaks,intensity))), digits=2)

        #end if

        ##QC plot and numbers
        par(mfrow = c(2, 2), oma=c(0,0,2,0))
        for (random_sample in random_spectra){
            plot(peaks[[random_sample]], sub="", main=paste0("spectrum ", pixelnames[random_sample]))}
        title("Filtered spectra", outer=TRUE, line=0)
        minmz = round(min(unlist(lapply(peaks,mass))), digits=4)
        maxmz = round(max(unlist(lapply(peaks,mass))), digits=4)
        mean_features = round(length(unlist(lapply(peaks,mass)))/length(peaks), digits=2)
        medint = round(median(unlist(lapply(peaks,intensity))), digits=2)

        </conditional>
        <repeat name="methods" title="Method" min="1">
            <conditional name="methods_conditional">
                <param name="method" type="select" label="Select a method">
                    <option value="Peak_detection">Peak detection</option>
                    <option value="monoisotopic_peaks">Monoisotopic peaks</option>
                    <option value="Align">Align Spectra (warping/phase correction)</option>
                    <option value="Binning">Binning</option>
                    <option value="Filtering">Filtering</option>
                </param>
                <when value="Peak_detection">

                    <param name="snr" type="integer" value="2" label="Signal-to-noise-ratio" help=""/>
                    <param name="use_annotations" type="boolean" label="Detect peaks on average mass spectra" help="Spectra with same annotation are averaged before peak detection, no imzML output" truevalue="TRUE" falsevalue="FALSE"/>
                </when>
                <when value="monoisotopic_peaks">
                    <param name="minCor" type="float" value="0.95" label="Minimal correlation"
                        help="Minimal correlation between the peak pattern generated by the model and the experimental peaks in the MassPeaks object to be recognized as isotopic pattern."/>
                    <param name="tolerance" type="float" label="Tolerance" value="0.0001"
                        help="Maximal relative deviation of a peak position (m/z) to be considered as identical: abs(((mass[i]+distance)-mass[i+1])/mass[i]) smaller than 'tolerance'. For 100ppm use 0.0001" />
                    <param name="distance" type="text" label="Distance" value="1.00235" help="Distance between two consecutive peaks in an isotopic pattern. 1.00235 is average distance for polypeptides. Multiple values can be used to find multiple charged pattern e.g. 1, 0.5 ,0.33">
                        <sanitizer invalid_char="">
                            <valid initial="string.digits">
                                <add value="," />
                                <add value=":" />
                                <add value="." />
                            </valid>
                        </sanitizer>
                    </param>
                    <param name="size" type="text" label="Size" value="3:10" help="Size (length) of isotopic pattern, longer patterns are prefered over shorter ones, min size is 2, a range can be used.">
                        <sanitizer invalid_char="">
                            <valid initial="string.digits">
                                <add value=":" />
                            </valid>
                        </sanitizer>
                    </param>
                </when>

                <when value="Align">
                    <param name="warping_method" type="select" label="Warping methods">
                        <option value="lowess" selected="True">Lowess</option>

                </when>

                <when value="Binning">
                    <param name="bin_tolerance" type="float" value="0.002" label="Tolerance"
                        help="After the alignment the peak positions (mass) are very similar but not identical. The binning is needed to make similar peak mass values identical."/>
                    <param name="bin_method" display="radio" type="select" label="Bin creation rule" help="strict - creates bins never containing two or more peaks of the same sample. relaxed - allows multiple peaks of the same sample in one bin.">
                    <option value="strict" selected="True" >strict</option>
                    <option value="relaxed" >relaxed</option>
                </param>
                </when>
                <when value="Filtering">

            <param name="centroids" value="TRUE"/>
             <repeat name="methods">
                <conditional name="methods_conditional">
                <param name="method" value="monoisotopic_peaks"/>
                <param name="tolerance" value="0.0004"/>
                <param name="size" value="3"/>
            </conditional>
            </repeat>
            <output name="plots" file="peakdetection2_QC.pdf" compare="sim_size"/>
            <output name="masspeaks" file="masspeaks2.tabular"/>
            <output name="intensity_matrix" file="int2.tabular"/>

    - Detect peaks on average mass spectra: Spectra with the same annotation (taken from the annotation tabular input) are averaged and peak picking is performed on the average spectrum of each annotation group. The exported imzML is empty and cannot be used for further analysis steps. The peaklist and intensity matrix outputs contain the annotation group names with their averaged intensity values. Filtering steps have to be done in the same run as the peak picking.


- Monoisotopic peaks: Keeps only the monoisotopic peaks

    - Based on poisson model for isotopic patterns as decribed in (`Breen et al. <https://doi.org/10.1002/1522-2683(20000601)21:11%3C2243::AID-ELPS2243%3E3.0.CO;2-K>`_)
    - Isotopic pattern can be characterized and recognized by

        - the similarity of the experimental pattern with the modelled pattern
        - the distance between consecutive isotopic peaks. For polypeptides the average distance is 1.00235 (`Park et al. <https://pubs.acs.org/doi/abs/10.1021/ac800913b>`_). Multiply charged analytes have smaller distances between the peaks (e.g. z = 1 distance = ~1; z = 2: distance = ~0.5; z = 3: distance = ~0.3333) To search for differently charged isotopic pattern multiple distances can be applied - the order matters because the first distance that matches is reported (1, 0.5, 0.3333). 
        - the size (length) of the pattern, multiple values can be applied, longer patterns are prefered over shorter ones.


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

    - without external reference m/z: internal reference is obtained by filtering (default 90%) and binning the peaks to find landmark peaks and their average m/z
    - with external reference m/z: the m/z provided in a tabular file are used as a reference, at least 10 reference values are recommended