view segmentation.xml @ 7:4a2ac25d1063 draft

"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/cardinal commit f986c51abe33c7f622d429a3c4a79ee24b33c1f3"

<tool id="cardinal_segmentations" name="MSI segmentation" version="@VERSION@.0">
    <description>mass spectrometry imaging spatial clustering</description>
    <macros>
        <import>macros.xml</import>
    </macros>
    <expand macro="requirements">
        <requirement type="package" version="2.3">r-gridextra</requirement>
      </expand>
    <command detect_errors="exit_code">
    <![CDATA[

        @INPUT_LINKING@
        cat '${MSI_segmentation}' &&
        Rscript '${MSI_segmentation}'

    ]]>
    </command>
    <configfiles>
        <configfile name="MSI_segmentation"><![CDATA[

################################# load libraries and read file #################

library(Cardinal)
library(gridExtra)

@READING_MSIDATA@

       msidata = as(msidata, "MSImageSet") ##coercion to MSImageSet

## remove duplicated coordinates
msidata <- msidata[,!duplicated(coord(msidata))]


@DATA_PROPERTIES_INRAM@


######################################## PDF ###################################
################################################################################
################################################################################


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

title(main=paste0("Spatial segmentation for file: \n\n", "$infile.display_name"))


############################# I) numbers ####################################
#############################################################################
grid.table(property_df, rows= NULL)

if (npeaks > 0 && sum(is.na(spectra(msidata)))==0)
{

######################## II) segmentation tools #############################
#############################################################################
        #set $color_string = ','.join(['"%s"' % $color.feature_color for $color in $colours])
        colourvector = c($color_string)

        ## set seed to make analysis reproducible
        set.seed($setseed)

        #if str( $segm_cond.segmentationtool ) == 'pca':
            print('pca')
            ##pca

            component_vector = character()
            for (numberofcomponents in 1:$segm_cond.pca_ncomp)
                {component_vector[numberofcomponents]= paste0("PC", numberofcomponents)}

            pca_result = PCA(msidata, ncomp=$segm_cond.pca_ncomp, column = component_vector, superpose = FALSE, 
            method = "$segm_cond.pca_method", scale = $segm_cond.pca_scale, layout = c(ncomp, 1))

            ## remove msidata to clean up RAM space
            rm(msidata)
            gc()

            ### table in pdf file
            plot(0,type='n',axes=FALSE,ann=FALSE)
            sd_table = as.data.frame(round(pca_result@resultData\$ncomp\$sdev, digits=2))
            colnames(sd_table) = "Standard deviation"
            PC_vector = character()
            for (PCs in 1:$segm_cond.pca_ncomp){
                PC_vector[[PCs]] = c(paste0("PC",PCs))}
            sd_table = cbind(PC_vector, sd_table)
            colnames(sd_table)[1] = "Principal components"
            grid.table(sd_table, rows=NULL)
            ### images in pdf file
            print(image(pca_result, main="PCA image", strip = FALSE, col=colourvector, ylim=c(maximumy+2, minimumy-2)))
            for (PCs in 1:$segm_cond.pca_ncomp){
                print(image(pca_result, column = c(paste0("PC",PCs)),strip = FALSE, superpose = FALSE, main=paste0("PC", PCs), col.regions = risk.colors(100), ylim=c(maximumy+2, minimumy-2)))}
            ### plots in pdf file
            print(plot(pca_result, main="PCA plot", col= colourvector, strip = FALSE))
            for (PCs in 1:$segm_cond.pca_ncomp){
                print(plot(pca_result, column = c(paste0("PC",PCs)),main=paste0("PC", PCs),strip = FALSE,superpose = FALSE))}

            ### values in tabular files
            pcaloadings = formatC(pca_result@resultData\$ncomp\$loadings, format = "e", digits = 6)### loading for each m/z value
            pcaloadings2 = cbind(matrix(unlist(strsplit(rownames(pcaloadings), " = ")), ncol=2, byrow=TRUE)[,2], pcaloadings)
            colnames(pcaloadings2) = c("mz", colnames(pcaloadings))
            pcascores = round(pca_result@resultData\$ncomp\$scores, digits=6) ### scores for each pixel

            ## pixel names and coordinates
            ## to remove potential sample names and z dimension, split at comma and take only x and y 
            x_coords = unlist(lapply(strsplit(rownames(pcascores), ","), `[[`, 1))
            y_coords = unlist(lapply(strsplit(rownames(pcascores), ","), `[[`, 2))
            x_coordinates = gsub("x = ","",x_coords)
            y_coordinates = gsub(" y = ","",y_coords)

            pixel_names = paste0("xy_", x_coordinates, "_", y_coordinates)
            pcascores2 = data.frame(pixel_names, x_coordinates, y_coordinates, pcascores)
            colnames(pcascores2) = c("pixel names", "x", "y", colnames(pcascores))
            write.table(pcaloadings2, file="$mzfeatures", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
            write.table(pcascores2, file="$pixeloutput", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")

            ## optional output as .RData
            #if $output_rdata:
            ## save as (.RData)
            save(pca, file="$segmentation_rdata")

            #end if

        #elif str( $segm_cond.segmentationtool ) == 'kmeans':
            print('kmeans')
            ##k-means

            skm = spatialKMeans(msidata, r=c($segm_cond.kmeans_r), k=c($segm_cond.kmeans_k), method="$segm_cond.kmeans_method")

            ## remove msidata to clean up RAM space
            rm(msidata)
            gc()

            print(image(skm, key=TRUE, main="K-means clustering", strip=FALSE, col= colourvector, layout=c(1,1), ylim=c(maximumy+2, minimumy-2)))
            print(plot(skm, main="K-means plot", col= colourvector, strip=FALSE, layout=c(1,1)))

            skm_clusters = data.frame(matrix(NA, nrow = pixelcount, ncol = 0))
            for (iteration in 1:length(skm@resultData)){
                        skm_cluster = ((skm@resultData)[[iteration]]\$cluster)
            skm_clusters = cbind(skm_clusters, skm_cluster) }

            ## pixel names and coordinates
            ## to remove potential sample names and z dimension, split at comma and take only x and y 
            x_coords = unlist(lapply(strsplit(rownames(skm_clusters), ","), `[[`, 1))
            y_coords = unlist(lapply(strsplit(rownames(skm_clusters), ","), `[[`, 2))
            x_coordinates = gsub("x = ","",x_coords)
            y_coordinates = gsub(" y = ","",y_coords)
            pixel_names = paste0("xy_", x_coordinates, "_", y_coordinates)
            skm_clusters2 = data.frame(pixel_names, x_coordinates, y_coordinates, skm_clusters)
            colnames(skm_clusters2) = c("pixel names", "x", "y",names(skm@resultData))

            skm_toplabels = topFeatures(skm, n=$segm_cond.kmeans_toplabels)

            write.table(skm_toplabels, file="$mzfeatures", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
            write.table(skm_clusters2, file="$pixeloutput", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")

            ## optional output as .RData
            #if $output_rdata:

            ## save as (.RData)
            save(skm, file="$segmentation_rdata")

            #end if

        #elif str( $segm_cond.segmentationtool ) == 'centroids':
            print('centroids')
            ##centroids

            ssc = spatialShrunkenCentroids(msidata, r=c($segm_cond.centroids_r), k=c($segm_cond.centroids_k), s=c($segm_cond.centroids_s), method="$segm_cond.centroids_method")
            ## remove msidata to clean up RAM space
            rm(msidata)
            gc()
            print(image(ssc, key=TRUE, main="Spatial shrunken centroids", strip = TRUE, col= colourvector,layout=c(1,1), ylim=c(maximumy+2, minimumy-2)))
            print(plot(ssc, main="Spatial shrunken centroids plot", col= colourvector, strip = TRUE,layout=c(1,1)))
            print(plot(ssc, mode = "tstatistics",key = TRUE, layout = c(1,1), main="t-statistics", col=colourvector))

            plot(summary(ssc), main = "Number of segments")

            ssc_classes = data.frame(matrix(NA, nrow = pixelcount, ncol = 0))
            for (iteration in 1:length(ssc@resultData)){
            ssc_class = ((ssc@resultData)[[iteration]]\$classes)
            ssc_classes = cbind(ssc_classes, ssc_class) }

            ## pixel names and coordinates
            ## to remove potential sample names and z dimension, split at comma and take only x and y 
            x_coords = unlist(lapply(strsplit(rownames(ssc_classes), ","), `[[`, 1))
            y_coords = unlist(lapply(strsplit(rownames(ssc_classes), ","), `[[`, 2))
            x_coordinates = gsub("x = ","",x_coords)
            y_coordinates = gsub(" y = ","",y_coords)
            pixel_names = paste0("xy_", x_coordinates, "_", y_coordinates)
            ssc_classes2 = data.frame(pixel_names, x_coordinates, y_coordinates, ssc_classes)
            colnames(ssc_classes2) = c("pixel names", "x", "y", names(ssc@resultData))

            ssc_toplabels =  topFeatures(ssc, n=$segm_cond.centroids_toplabels)

            write.table(ssc_toplabels, file="$mzfeatures", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
            write.table(ssc_classes2, file="$pixeloutput", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")

            ## optional output as .RData
            #if $output_rdata:

            ## save as (.RData)
            save(ssc, file="$segmentation_rdata")

            #end if

        #end if

    dev.off()

        ## optional svg output with original coordinates
        #if $svg_pixelimage:
            print("svg image")
            ## reverse y axis for svg output = correct order and nice svg image


            svg(file="svg_pixel_output.svg", width=maximumx, height=maximumy)
            par(mar=c(0,0,0,0))
            #if str( $segm_cond.segmentationtool ) == 'pca':
                coord(pca_result)\$y <- max(coord(pca_result)\$y) - coord(pca_result)\$y + 1
                image(pca_result, strip = FALSE, colorkey=FALSE, axes=FALSE, xlab=NA, ylab=NA, col=colourvector)
            #elif str( $segm_cond.segmentationtool ) == 'kmeans':
                coord(skm)\$y <- max(coord(skm)\$y) - coord(skm)\$y + 1
                image(skm, key=FALSE, strip=FALSE, col= colourvector)
            #elif str( $segm_cond.segmentationtool ) == 'centroids':
                coord(ssc)\$y <- max(coord(ssc)\$y) - coord(ssc)\$y + 1
                image(ssc, key=FALSE, strip = FALSE, col= colourvector)
            #end if
            dev.off()
        #end if


}else{
    print("Inputfile has no intensities > 0")
    dev.off()
}

    ]]></configfile>
    </configfiles>
    <inputs>
        <expand macro="reading_msidata"/>
            <conditional name="segm_cond">
                <param name="segmentationtool" type="select" label="Select the tool for spatial clustering">
                    <option value="pca" selected="True">pca</option>
                    <option value="kmeans">k-means</option>
                    <option value="centroids">spatial shrunken centroids</option>
                </param>
                <when value="pca">
                    <param name="pca_ncomp" type="integer" value="2"
                           label="The number of principal components to calculate"/>
                    <param name="pca_method" type="select" 
                           label="The function used to calculate the singular value decomposition">
                        <option value="irlba" selected="True">irlba</option>
                        <option value="svd">svd</option>
                    </param>
                    <param name="pca_scale" type="boolean" truevalue="TRUE" falsevalue="FALSE" label="Scaling of data before analysis"/>
                </when>
                <when value="kmeans">
                    <param name="kmeans_r" type="text" value="2"
                           label="The spatial neighborhood radius of nearby pixels to consider (r)" help="Multiple values are allowed (e.g. 1,2,3 or 2:5)">
                        <expand macro="sanitizer_multiple_digits"/>
                    </param>
                    <param name="kmeans_k" type="text" value="3"
                           label="The number of clusters (k)" help="Multiple values are allowed (e.g. 1,2,3 or 2:5)">
                        <expand macro="sanitizer_multiple_digits"/>
                    </param>
                    <param name="kmeans_method" type="select" display="radio"
                           label="The method to use to calculate the spatial smoothing kernels for the embedding. The 'gaussian' method refers to spatially-aware (SA) clustering, and 'adaptive' refers to spatially-aware structurally-adaptive (SASA) clustering">
                        <option value="gaussian">gaussian</option>
                        <option value="adaptive" selected="True">adaptive</option>
                </param>
                <param name="kmeans_toplabels" type="integer" value="500"
                       label="Number of toplabels (m/z) which should be written in tabular output"/>
                 </when>

                <when value="centroids">
                    <param name="centroids_r" type="text" value="2"
                           label="The spatial neighborhood radius of nearby pixels to consider (r)" help="Multiple values are allowed (e.g. 1,2,3 or 2:5)">
                        <expand macro="sanitizer_multiple_digits"/>
                    </param>
                    <param name="centroids_k" type="text" value="5"
                           label="The initial number of clusters (k)" help="Multiple values are allowed (e.g. 1,2,3 or 2:5)">
                        <expand macro="sanitizer_multiple_digits"/>
                    </param>
                    <param name="centroids_s" type="text" value="2"
                           label="The sparsity thresholding parameter by which to shrink the t-statistics (s)"
                           help="As s increases, fewer m/z features (m/z values) will be used in the spatial segmentation, and only the informative m/z features will be retained. Multiple values are allowed (e.g. 1,2,3 or 2:5)">
                        <expand macro="sanitizer_multiple_digits"/>
                    </param>
                    <param name="centroids_method" type="select" display="radio" label="The method to use to calculate the spatial smoothing kernels for the embedding. The 'gaussian' method refers to spatially-aware (SA) weights, and 'adaptive' refers to spatially-aware structurally-adaptive (SASA) weights">
                        <option value="gaussian">gaussian</option>
                        <option value="adaptive" selected="True">adaptive</option>
                </param>
                <param name="centroids_toplabels" type="integer" value="500"
                       label="Number of toplabels (m/z) which should be written in tabular output"/>
                </when>
            </conditional>
            <param name="svg_pixelimage" type="boolean" label="Export first segmentation image as svg"/>
            <repeat name="colours" title="Colours for the plots" min="1" max="50">
                <param name="feature_color" type="color" label="Colours" value="#ff00ff" help="Numbers of colours should be the same as number of components">
                  <sanitizer>
                    <valid initial="string.letters,string.digits">
                      <add value="#" />
                    </valid>
                  </sanitizer>
                </param>
            </repeat>
        <param name="output_rdata" type="boolean" label="Results as .RData output"/>
        <param name="setseed" type="integer" value="1" label="set seed" help="Use same value to reproduce previous results"/>
    </inputs>
    <outputs>
        <data format="pdf" name="segmentationimages" from_work_dir="segmentationpdf.pdf" label = "${tool.name} on ${on_string}: results"/>
        <data format="tabular" name="mzfeatures" label="${tool.name} on ${on_string}: features"/>
        <data format="tabular" name="pixeloutput" label="${tool.name} on ${on_string}: pixels"/>
        <data format="rdata" name="segmentation_rdata" label="${tool.name} on ${on_string}: results.RData">
            <filter>output_rdata</filter>
        </data>
        <data format="svg" name="svg_output" from_work_dir="svg_pixel_output.svg" label="${tool.name} on ${on_string}: image.svg">
            <filter>svg_pixelimage</filter>
        </data>
    </outputs>
    <tests>
        <test>
            <expand macro="infile_imzml"/>
            <param name="segmentationtool" value="pca"/>
            <repeat name="colours">
                <param name="feature_color" value="#ff00ff"/>
            </repeat>
            <repeat name="colours">
                <param name="feature_color" value="#0000FF"/>
            </repeat>
            <output name="segmentationimages" file="pca_imzml.pdf" compare="sim_size"/>
            <output name="mzfeatures">
                <assert_contents>
                    <has_text text="300.1667" />
                    <has_text text="300.25" />
                    <has_text text="-4.234458e-04" />
                    <has_text text="3.878545e-10" />
                    <has_n_columns n="3" />
                </assert_contents>
            </output>
            <output name="pixeloutput" file="scores_pca.tabular"/>
        </test>
        <test>
            <expand macro="infile_imzml"/>
            <param name="segmentationtool" value="kmeans"/>
            <param name="kmeans_r" value="1:3"/>
            <param name="kmeans_k" value="2,3"/>
            <param name="kmeans_toplabels" value="20"/>
            <repeat name="colours">
                <param name="feature_color" value="#ff00ff"/>
            </repeat>
            <repeat name="colours">
                <param name="feature_color" value="#0000FF"/>
            </repeat>
            <repeat name="colours">
                <param name="feature_color" value="#00C957"/>
            </repeat>
            <param name="output_rdata" value="True"/>
            <output name="segmentationimages" file="kmeans_analyze.pdf" compare="sim_size"/>
            <output name="mzfeatures" file="toplabels_skm.tabular"/>
            <output name="pixeloutput" file="cluster_skm.tabular"/>
            <output name="segmentation_rdata" file="cluster_skm.RData" compare="sim_size"/>
        </test>
        <test>
            <param name="infile" value="preprocessed.RData" ftype="rdata"/>
            <param name="segmentationtool" value="centroids"/>
            <param name="centroids_r" value="1,2"/>
            <param name="centroids_k" value="3"/>
            <param name="centroids_toplabels" value="50"/>
            <repeat name="colours">
                <param name="feature_color" value="#0000FF"/>
            </repeat>
            <repeat name="colours">
                <param name="feature_color" value="#00C957"/>
            </repeat>
            <repeat name="colours">
                <param name="feature_color" value="#B0171F"/>
            </repeat>
            <output name="segmentationimages" file="centroids_rdata.pdf" compare="sim_size"/>
            <output name="mzfeatures" file="toplabels_ssc.tabular"/>
            <output name="pixeloutput" file="classes_ssc.tabular"/>
        </test>
        <test>
            <param name="infile" value="" ftype="imzml">
                <composite_data value="preprocessing_results3.imzml"/>
                <composite_data value="preprocessing_results3.ibd"/>
            </param>
            <conditional name="processed_cond">
                <param name="processed_file" value="processed"/>
                <param name="accuracy" value="1"/>
                <param name="units" value="mz"/>
            </conditional>
            <param name="segmentationtool" value="centroids"/>
            <param name="centroids_r" value="1"/>
            <param name="centroids_k" value="2,3"/>
            <param name="centroids_s" value="0,3"/>
            <param name="centroids_toplabels" value="100"/>
            <repeat name="colours">
                <param name="feature_color" value="#0000FF"/>
            </repeat>
            <repeat name="colours">
                <param name="feature_color" value="#00C957"/>
            </repeat>
            <repeat name="colours">
                <param name="feature_color" value="#B0171F"/>
            </repeat>
            <output name="segmentationimages" file="centroids_proc.pdf" compare="sim_size"/>
            <output name="mzfeatures" file="toplabels_proc.tabular"/>
            <output name="pixeloutput" file="classes_proc.tabular"/>
        </test>
    </tests>
    <help>
        <![CDATA[

@CARDINAL_DESCRIPTION@

-----

This tool provides three different Cardinal functions for unsupervised clustering/spatial segmentation of mass spectrometry imaging data.

@MSIDATA_INPUT_DESCRIPTION@
            - NA intensities are not allowed
            - duplicated coordinates will be removed


**Options**

- PCA: principal component analysis
- k-means: spatially-aware k-means clustering (adopted from `Alexandrov and Kobarg <https://doi.org/10.1093/bioinformatics/btr246>`_)
- spatial shrunken centroids: Allows the number of segments to decrease according to the data. This allows selection of the number of clusters (more details in `Bemis et al. <https://doi.org/10.1074/mcp.O115.053918>`_)

**Output**

- Pdf with the heatmaps and plots for the segmentation
- Tabular file with information on m/z and pixels: loadings/scores (PCA), toplabels/clusters (k-means), toplabels/classes (spatial shrunken centroids)
- Optional .RData file which contains the segmentation results and can be used for further exploration in R using the Cardinal package
- Optional: svg file with the first segmentation image

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