Mercurial > repos > galaxyp > cardinal_classification

comparison classification.xml @ 0:2fdbbb1be2b0 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/cardinal commit 0825a4ccd3ebf4ca8a298326d14f3e7b25ae8415
author galaxyp
date Mon, 01 Oct 2018 01:06:08 -0400
parents
children 6a03b201bc12
comparison
equal deleted inserted replaced
-1:000000000000 0:2fdbbb1be2b0
1 <tool id="cardinal_classification" name="MSI classification" version="@VERSION@.0">
2 <description>spatial classification of mass spectrometry imaging data</description>
3 <macros>
4 <import>macros.xml</import>
5 </macros>
6 <expand macro="requirements">
7 <requirement type="package" version="2.2.1">r-gridextra</requirement>
8 <requirement type="package" version="0.20-35">r-lattice</requirement>
9 <requirement type="package" version="2.2.1">r-ggplot2</requirement>
10 </expand>
11 <command detect_errors="exit_code">
12 <![CDATA[
13
14 @INPUT_LINKING@
15 cat '${MSI_segmentation}' &&
16 Rscript '${MSI_segmentation}'
17
18 ]]>
19 </command>
20 <configfiles>
21 <configfile name="MSI_segmentation"><![CDATA[
22
23
24 ################################# load libraries and read file #########################
25
26 library(Cardinal)
27 library(gridExtra)
28 library(lattice)
29 library(ggplot2)
30
31 @READING_MSIDATA@
32
33 ## create full matrix to make processed imzML files compatible with classification
34 iData(msidata) <- iData(msidata)[]
35
36 @DATA_PROPERTIES@
37
38 ######################################## PDF ###################################
39 ################################################################################
40 ################################################################################
41
42 Title = "Prediction"
43
44 #if str( $type_cond.type_method) == "training":
45
46 Title = "$type_cond.method_cond.class_method"
47 #end if
48
49 pdf("classificationpdf.pdf", fonts = "Times", pointsize = 12)
50 plot(0,type='n',axes=FALSE,ann=FALSE)
51
52
53 title(main=paste0(Title," for file: \n\n", "$infile.display_name"))
54
55
56 ##################### I) numbers and control plots #############################
57 ###############################################################################
58
59 ## table with values
60 grid.table(property_df, rows= NULL)
61
62
63 if (npeaks > 0 && sum(is.na(spectra(msidata)))==0){
64
65 opar <- par()
66
67 ######################## II) Training #############################
68 #############################################################################
69 #if str( $type_cond.type_method) == "training":
70 print("training")
71
72
73 ## load y response (will be needed in every training scenario)
74
75 y_tabular = read.delim("$type_cond.annotation_file", header = $type_cond.tabular_header, stringsAsFactors = FALSE)
76
77 #if str($type_cond.column_fold) == "None":
78 y_input = y_tabular[,c($type_cond.column_x, $type_cond.column_y, $type_cond.column_response)]
79 #else
80 y_input = y_tabular[,c($type_cond.column_x, $type_cond.column_y, $type_cond.column_response, $type_cond.column_fold)]
81 #end if
82
83 colnames(y_input)[1:2] = c("x", "y")
84 ## merge with coordinate information of msidata
85 msidata_coordinates = cbind(coord(msidata)[,1:2], c(1:ncol(msidata)))
86 colnames(msidata_coordinates)[3] = "pixel_index"
87 merged_response = merge(msidata_coordinates, y_input, by=c("x", "y"), all.x=TRUE)
88 merged_response[is.na(merged_response)] = "NA"
89 merged_response = merged_response[order(merged_response\$pixel_index),]
90 y_vector = as.factor(merged_response[,4])
91
92 ## plot of y vector
93
94 position_df = cbind(coord(msidata)[,1:2], y_vector)
95 y_plot = ggplot(position_df, aes(x=x, y=y, fill=y_vector))+
96 geom_tile() +
97 coord_fixed()+
98 ggtitle("Distribution of the response variable y")+
99 theme_bw()+
100 theme(text=element_text(family="ArialMT", face="bold", size=15))+
101 theme(legend.position="bottom",legend.direction="vertical")+
102 guides(fill=guide_legend(ncol=4,byrow=TRUE))
103 coord_labels = aggregate(cbind(x,y)~y_vector, data=position_df, mean, na.rm=TRUE, na.action="na.pass")
104 coord_labels\$file_number = gsub( "_.*$", "", coord_labels\$y_vector)
105 print(y_plot)
106
107
108 ## plot of folds
109
110 #if str($type_cond.column_fold) != "None":
111 fold_vector = as.factor(merged_response[,5])
112
113
114 position_df = cbind(coord(msidata)[,1:2], fold_vector)
115 fold_plot = ggplot(position_df, aes(x=x, y=y, fill=fold_vector))+
116 geom_tile() +
117 coord_fixed()+
118 ggtitle("Distribution of the fold variable")+
119 theme_bw()+
120 theme(text=element_text(family="ArialMT", face="bold", size=15))+
121 theme(legend.position="bottom",legend.direction="vertical")+
122 guides(fill=guide_legend(ncol=4,byrow=TRUE))
123 coord_labels = aggregate(cbind(x,y)~fold_vector, data=position_df, mean, na.rm=TRUE, na.action="na.pass")
124 coord_labels\$file_number = gsub( "_.*$", "", coord_labels\$fold_vector)
125 print(fold_plot)
126
127 #end if
128
129 ######################## PLS #############################
130 #if str( $type_cond.method_cond.class_method) == "PLS":
131 print("PLS")
132
133 ######################## PLS - CV #############################
134 #if str( $type_cond.method_cond.analysis_cond.PLS_method) == "cvapply":
135 print("PLS cv")
136
137 ## set variables for components and number of response groups
138 components = c($type_cond.method_cond.analysis_cond.plscv_comp)
139 number_groups = length(levels(y_vector))
140
141 ## PLS-cvApply:
142 msidata.cv.pls <- cvApply(msidata, .y = y_vector, .fold = fold_vector, .fun = "PLS", ncomp = components)
143
144 ## create table with summary
145 count = 1
146 summary_plscv = list()
147 accuracy_vector = numeric()
148 for (iteration in components){
149 summary_iteration = summary(msidata.cv.pls)\$accuracy[[paste0("ncomp = ", iteration)]]
150 ## change class of numbers into numeric to round and calculate mean
151 summary_iteration2 = round(as.numeric(summary_iteration), digits=2)
152 summary_matrix = matrix(summary_iteration2, nrow=4, ncol=number_groups)
153 accuracy_vector[count] = mean(summary_matrix[1,]) ## vector with accuracies to find later maximum for plot
154 summary_iteration3 = cbind(rownames(summary_iteration), summary_matrix) ## include rownames in table
155 summary_iteration4 = t(summary_iteration3)
156 summary_iteration5 = cbind(c(paste0("ncomp = ", iteration), colnames(summary_iteration)), summary_iteration4)
157 summary_plscv[[count]] = summary_iteration5
158 count = count+1} ## create list with summary table for each component
159 summary_plscv = do.call(rbind, summary_plscv)
160 summary_df = as.data.frame(summary_plscv)
161 colnames(summary_df) = NULL
162
163 ## plots
164 ## plot to find ncomp with highest accuracy
165 plot(components, accuracy_vector, ylab = "mean accuracy",type="o", main="Mean accuracy of PLS classification")
166 ncomp_max = components[which.max(accuracy_vector)] ## find ncomp with max. accuracy
167 ## one image for each sample/fold, 4 images per page
168 minimumy = min(coord(msidata.cv.pls)[,2])
169 maximumy = max(coord(msidata.cv.pls)[,2])
170 image(msidata.cv.pls, model = list(ncomp = ncomp_max),ylim= c(maximumy+0.2*maximumy,minimumy-0.2*minimumy),layout = c(2, 2))
171
172 ## print table with summary in pdf
173 par(opar)
174 plot(0,type='n',axes=FALSE,ann=FALSE)
175 title(main="Summary for the different components\n", adj=0.5)
176 ## 20 rows fits in one page:
177 if (nrow(summary_df)<=20){
178 grid.table(summary_df, rows= NULL)
179 }else{
180 grid.table(summary_df[1:20,], rows= NULL)
181 mincount = 21
182 maxcount = 40
183 for (count20 in 1:(ceiling(nrow(summary_df)/20)-1)){
184 plot(0,type='n',axes=FALSE,ann=FALSE)
185 if (maxcount <= nrow(summary_df)){
186 grid.table(summary_df[mincount:maxcount,], rows= NULL)
187 mincount = mincount+20
188 maxcount = maxcount+20
189 }else{### stop last page with last sample otherwise NA in table
190 grid.table(summary_df[mincount:nrow(summary_df),], rows= NULL)}
191 }
192 }
193
194 ## optional output as .RData
195 #if $output_rdata:
196 save(msidata.cv.pls, file="$classification_rdata")
197 #end if
198
199
200 ######################## PLS - analysis ###########################
201 #elif str( $type_cond.method_cond.analysis_cond.PLS_method) == "PLS_analysis":
202 print("PLS analysis")
203
204 ## set variables for components and number of response groups
205 component = c($type_cond.method_cond.analysis_cond.pls_comp)
206 number_groups = length(levels(y_vector))
207
208 ### pls analysis and coefficients plot
209 msidata.pls <- PLS(msidata, y = y_vector, ncomp = component, scale=$type_cond.method_cond.analysis_cond.pls_scale)
210 plot(msidata.pls, main="PLS coefficients per m/z")
211
212 ### summary table of PLS
213 summary_table = summary(msidata.pls)\$accuracy[[paste0("ncomp = ",component)]]
214 summary_table2 = round(as.numeric(summary_table), digits=2)
215 summary_matrix = matrix(summary_table2, nrow=4, ncol=number_groups)
216 summary_table3 = cbind(rownames(summary_table), summary_matrix) ## include rownames in table
217 summary_table4 = t(summary_table3)
218 summary_table5 = cbind(c(paste0("ncomp = ", component), colnames(summary_table)), summary_table4)
219 plot(0,type='n',axes=FALSE,ann=FALSE)
220 grid.table(summary_table5, rows= NULL)
221
222 ### image of the best m/z
223 minimumy = min(coord(msidata)[,2])
224 maximumy = max(coord(msidata)[,2])
225 print(image(msidata, mz = topLabels(msidata.pls)[1,1], normalize.image = "linear", contrast.enhance = "histogram",ylim= c(maximumy+0.2*maximumy,minimumy-0.2*minimumy), smooth.image="gaussian", main="best m/z heatmap"))
226
227 ### m/z and pixel information output
228 pls_classes = data.frame(msidata.pls\$classes[[1]])
229 pixel_names = gsub(", y = ", "_", names(pixels(msidata)))
230 pixel_names = gsub(" = ", "y_", pixel_names)
231 x_coordinates = matrix(unlist(strsplit(pixel_names, "_")), ncol=3, byrow=TRUE)[,2]
232 y_coordinates = matrix(unlist(strsplit(pixel_names, "_")), ncol=3, byrow=TRUE)[,3]
233
234 pls_classes2 = data.frame(pixel_names, x_coordinates, y_coordinates, pls_classes)
235 colnames(pls_classes2) = c("pixel names", "x", "y","predicted condition")
236 pls_toplabels = topLabels(msidata.pls, n=$type_cond.method_cond.analysis_cond.pls_toplabels)
237 write.table(pls_toplabels, file="$mzfeatures", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
238 write.table(pls_classes2, file="$pixeloutput", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
239
240 ## image with predicted classes
241 prediction_df = cbind(coord(msidata.pls)[,1:2], pls_classes)
242 colnames(prediction_df) = c("x", "y", "predicted_classes")
243
244 prediction_plot = ggplot(prediction_df, aes(x=x, y=y, fill=predicted_classes))+
245 geom_tile() +
246 coord_fixed()+
247 ggtitle("Predicted condition for each pixel")+
248 theme_bw()+
249 theme(text=element_text(family="ArialMT", face="bold", size=15))+
250 theme(legend.position="bottom",legend.direction="vertical")+
251 guides(fill=guide_legend(ncol=4,byrow=TRUE))
252 coord_labels = aggregate(cbind(x,y)~predicted_classes, data=prediction_df, mean, na.rm=TRUE, na.action="na.pass")
253 coord_labels\$file_number = gsub( "_.*$", "", coord_labels\$predicted_classes)
254 print(prediction_plot)
255
256 ### optional output as .RData
257 #if $output_rdata:
258 save(msidata.pls, file="$classification_rdata")
259 #end if
260
261 #end if
262
263 ######################## OPLS #############################
264 #elif str( $type_cond.method_cond.class_method) == "OPLS":
265 print("OPLS")
266
267 ######################## OPLS -CV #############################
268 #if str( $type_cond.method_cond.opls_analysis_cond.opls_method) == "opls_cvapply":
269 print("OPLS cv")
270
271 ## set variables for components and number of response groups
272 components = c($type_cond.method_cond.opls_analysis_cond.opls_cvcomp)
273 number_groups = length(levels(y_vector))
274
275 ## OPLS-cvApply:
276 msidata.cv.opls <- cvApply(msidata, .y = y_vector, .fold = fold_vector, .fun = "OPLS", ncomp = components, keep.Xnew = $type_cond.method_cond.opls_analysis_cond.xnew_cv)
277
278 ## create table with summary
279 count = 1
280 summary_oplscv = list()
281 accuracy_vector = numeric()
282 for (iteration in components){
283
284 summary_iteration = summary(msidata.cv.opls)\$accuracy[[paste0("ncomp = ", iteration)]]
285 ## change class of numbers into numeric to round and calculate mean
286 summary_iteration2 = round(as.numeric(summary_iteration), digits=2)
287 summary_matrix = matrix(summary_iteration2, nrow=4, ncol=number_groups)
288 accuracy_vector[count] = mean(summary_matrix[1,]) ## vector with accuracies to find later maximum for plot
289 summary_iteration3 = cbind(rownames(summary_iteration), summary_matrix) ## include rownames in table
290 summary_iteration4 = t(summary_iteration3)
291 summary_iteration5 = cbind(c(paste0("ncomp = ", iteration), colnames(summary_iteration)), summary_iteration4)
292 summary_oplscv[[count]] = summary_iteration5
293 count = count+1} ## create list with summary table for each component
294 summary_oplscv = do.call(rbind, summary_oplscv)
295 summary_df = as.data.frame(summary_oplscv)
296 colnames(summary_df) = NULL
297
298 ## plots
299 ## plot to find ncomp with highest accuracy
300 plot(components, accuracy_vector, ylab = "mean accuracy", type="o", main="Mean accuracy of OPLS classification")
301 ncomp_max = components[which.max(accuracy_vector)] ## find ncomp with max. accuracy
302 ## one image for each sample/fold, 4 images per page
303 minimumy = min(coord(msidata.cv.opls)[,2])
304 maximumy = max(coord(msidata.cv.opls)[,2])
305 image(msidata.cv.opls, model = list(ncomp = ncomp_max),ylim= c(maximumy+0.2*maximumy,minimumy-0.2*minimumy),layout = c(2, 2))
306
307 ## print table with summary in pdf
308 par(opar)
309 plot(0,type='n',axes=FALSE,ann=FALSE)
310 title(main="Summary for the different components\n", adj=0.5)
311 ## 20 rows fits in one page:
312 if (nrow(summary_df)<=20){
313 grid.table(summary_df, rows= NULL)
314 }else{
315 grid.table(summary_df[1:20,], rows= NULL)
316 mincount = 21
317 maxcount = 40
318 for (count20 in 1:(ceiling(nrow(summary_df)/20)-1)){
319 plot(0,type='n',axes=FALSE,ann=FALSE)
320 if (maxcount <= nrow(summary_df)){
321 grid.table(summary_df[mincount:maxcount,], rows= NULL)
322 mincount = mincount+20
323 maxcount = maxcount+20
324 }else{### stop last page with last sample otherwise NA in table
325 grid.table(summary_df[mincount:nrow(summary_df),], rows= NULL)}
326 }
327 }
328
329 ## optional output as .RData
330 #if $output_rdata:
331 save(msidata.cv.opls, file="$classification_rdata")
332 #end if
333
334
335 ######################## OPLS -analysis ###########################
336 #elif str( $type_cond.method_cond.opls_analysis_cond.opls_method) == "opls_analysis":
337 print("OPLS analysis")
338
339 ## set variables for components and number of response groups
340 component = c($type_cond.method_cond.opls_analysis_cond.opls_comp)
341 number_groups = length(levels(y_vector))
342
343
344 ### opls analysis and coefficients plot
345 msidata.opls <- PLS(msidata, y = y_vector, ncomp = component, scale=$type_cond.method_cond.opls_analysis_cond.opls_scale, keep.Xnew = $type_cond.method_cond.opls_analysis_cond.xnew)
346 plot(msidata.opls, main="OPLS coefficients per m/z")
347
348 ### summary table of OPLS
349 summary_table = summary(msidata.opls)\$accuracy[[paste0("ncomp = ",component)]]
350 summary_table2 = round(as.numeric(summary_table), digits=2)
351 summary_matrix = matrix(summary_table2, nrow=4, ncol=number_groups)
352 summary_table3 = cbind(rownames(summary_table), summary_matrix) ## include rownames in table
353 summary_table4 = t(summary_table3)
354 summary_table5 = cbind(c(paste0("ncomp = ", component), colnames(summary_table)), summary_table4)
355 plot(0,type='n',axes=FALSE,ann=FALSE)
356 grid.table(summary_table5, rows= NULL)
357
358 ### image of the best m/z
359 minimumy = min(coord(msidata)[,2])
360 maximumy = max(coord(msidata)[,2])
361 print(image(msidata, mz = topLabels(msidata.opls)[1,1], normalize.image = "linear", contrast.enhance = "histogram",smooth.image="gaussian", ylim= c(maximumy+0.2*maximumy,minimumy-0.2*minimumy), main="best m/z heatmap"))
362
363 ## m/z and pixel information output
364 opls_classes = data.frame(msidata.opls\$classes[[1]])
365 pixel_names = gsub(", y = ", "_", names(pixels(msidata)))
366 pixel_names = gsub(" = ", "y_", pixel_names)
367 x_coordinates = matrix(unlist(strsplit(pixel_names, "_")), ncol=3, byrow=TRUE)[,2]
368 y_coordinates = matrix(unlist(strsplit(pixel_names, "_")), ncol=3, byrow=TRUE)[,3]
369 opls_classes2 = data.frame(pixel_names, x_coordinates, y_coordinates, opls_classes)
370 colnames(opls_classes2) = c("pixel names", "x", "y","predicted condition")
371
372 opls_toplabels = topLabels(msidata.opls, n=$type_cond.method_cond.opls_analysis_cond.opls_toplabels)
373 write.table(opls_toplabels, file="$mzfeatures", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
374 write.table(opls_classes2, file="$pixeloutput", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
375
376 ## image with predicted classes
377 prediction_df = cbind(coord(msidata.opls)[,1:2], opls_classes)
378 colnames(prediction_df) = c("x", "y", "predicted_classes")
379
380 prediction_plot = ggplot(prediction_df, aes(x=x, y=y, fill=predicted_classes))+
381 geom_tile() +
382 coord_fixed()+
383 ggtitle("Predicted condition for each pixel")+
384 theme_bw()+
385 theme(text=element_text(family="ArialMT", face="bold", size=15))+
386 theme(legend.position="bottom",legend.direction="vertical")+
387 guides(fill=guide_legend(ncol=4,byrow=TRUE))
388 coord_labels = aggregate(cbind(x,y)~predicted_classes, data=prediction_df, mean, na.rm=TRUE, na.action="na.pass")
389 coord_labels\$file_number = gsub( "_.*$", "", coord_labels\$predicted_classes)
390 print(prediction_plot)
391
392 ## optional output as .RData
393 #if $output_rdata:
394 save(msidata.opls, file="$classification_rdata")
395 #end if
396 #end if
397
398
399 ######################## SSC #############################
400 #elif str( $type_cond.method_cond.class_method) == "spatialShrunkenCentroids":
401 print("SSC")
402
403 ######################## SSC - CV #############################
404 #if str( $type_cond.method_cond.ssc_analysis_cond.ssc_method) == "ssc_cvapply":
405 print("SSC cv")
406
407 ## set variables for components and number of response groups
408 number_groups = length(levels(y_vector))
409
410 ## SSC-cvApply:
411 msidata.cv.ssc <- cvApply(msidata, .y = y_vector,.fold = fold_vector,.fun = "spatialShrunkenCentroids", r = c($type_cond.method_cond.ssc_r), s = c($type_cond.method_cond.ssc_s), method = "$type_cond.method_cond.ssc_kernel_method")
412
413 ## create table with summary
414 count = 1
415 summary_ssccv = list()
416 accuracy_vector = numeric()
417 iteration_vector = character()
418 for (iteration in names(msidata.cv.ssc@resultData[[1]][,1])){
419
420 summary_iteration = summary(msidata.cv.ssc)\$accuracy[[iteration]]
421 ## change class of numbers into numeric to round and calculate mean
422 summary_iteration2 = round(as.numeric(summary_iteration), digits=2)
423 summary_matrix = matrix(summary_iteration2, nrow=4, ncol=number_groups)
424 accuracy_vector[count] = mean(summary_matrix[1,]) ## vector with accuracies to find later maximum for plot
425 summary_iteration3 = cbind(rownames(summary_iteration), summary_matrix) ## include rownames in table
426 summary_iteration4 = t(summary_iteration3)
427 summary_iteration5 = cbind(c(iteration, colnames(summary_iteration)), summary_iteration4)
428 summary_ssccv[[count]] = summary_iteration5
429 iteration_vector[count] = unlist(strsplit(iteration, "[,]"))[3]
430 count = count+1} ## create list with summary table for each component
431 summary_ssccv = do.call(rbind, summary_ssccv)
432 summary_df = as.data.frame(summary_ssccv)
433 colnames(summary_df) = NULL
434
435 ## plot to find parameters with highest accuracy
436 plot(c($type_cond.method_cond.ssc_s),accuracy_vector[!duplicated(iteration_vector)], type="o",ylab="Mean accuracy", xlab = "Shrinkage parameter (s)", main="Mean accuracy of SSC classification")
437 best_params = names(msidata.cv.ssc@resultData[[1]][,1])[which.max(accuracy_vector)] ## find parameters with max. accuracy
438 r_value = as.numeric(substring(unlist(strsplit(best_params, ","))[1], 4))
439 s_value = as.numeric(substring(unlist(strsplit(best_params, ","))[3], 5)) ## remove space
440 minimumy = min(coord(msidata.cv.ssc)[,2])
441 maximumy = max(coord(msidata.cv.ssc)[,2])
442 image(msidata.cv.ssc, model = list( r = r_value, s = s_value ), ylim= c(maximumy+0.2*maximumy,minimumy-0.2*minimumy),layout=c(2,2))
443
444 ## print table with summary in pdf
445 par(opar)
446 plot(0,type='n',axes=FALSE,ann=FALSE)
447 title(main="Summary for the different parameters\n", adj=0.5)
448 ## 20 rows fits in one page:
449 if (nrow(summary_df)<=20){
450 grid.table(summary_df, rows= NULL)
451 }else{
452 grid.table(summary_df[1:20,], rows= NULL)
453 mincount = 21
454 maxcount = 40
455 for (count20 in 1:(ceiling(nrow(summary_df)/20)-1)){
456 plot(0,type='n',axes=FALSE,ann=FALSE)
457 if (maxcount <= nrow(summary_df)){
458 grid.table(summary_df[mincount:maxcount,], rows= NULL)
459 mincount = mincount+20
460 maxcount = maxcount+20
461 }else{### stop last page with last sample otherwise NA in table
462 grid.table(summary_df[mincount:nrow(summary_df),], rows= NULL)}
463 }
464 }
465
466 ## optional output as .RData
467 #if $output_rdata:
468 save(msidata.cv.ssc, file="$classification_rdata")
469 #end if
470
471 ######################## SSC -analysis ###########################
472 #elif str( $type_cond.method_cond.ssc_analysis_cond.ssc_method) == "ssc_analysis":
473 print("SSC analysis")
474
475 ## set variables for components and number of response groups
476 number_groups = length(levels(y_vector))
477
478 ## SSC analysis and plot
479 msidata.ssc <- spatialShrunkenCentroids(msidata, y = y_vector, .fold = fold_vector,
480 r = c($type_cond.method_cond.ssc_r), s = c($type_cond.method_cond.ssc_s), method = "$type_cond.method_cond.ssc_kernel_method")
481 plot(msidata.ssc, mode = "tstatistics", model = list("r" = c($type_cond.method_cond.ssc_r), "s" = c($type_cond.method_cond.ssc_s)))
482
483 ### summary table SSC
484 ##############summary_table = summary(msidata.ssc)
485 summary_table = summary(msidata.ssc)\$accuracy[[names(msidata.ssc@resultData)]]
486 summary_table2 = round(as.numeric(summary_table), digits=2)
487 summary_matrix = matrix(summary_table2, nrow=4, ncol=number_groups)
488 summary_table3 = cbind(rownames(summary_table), summary_matrix) ## include rownames in table
489 summary_table4 = t(summary_table3)
490 summary_table5 = cbind(c(names(msidata.ssc@resultData),colnames(summary_table)), summary_table4)
491 plot(0,type='n',axes=FALSE,ann=FALSE)
492 grid.table(summary_table5, rows= NULL)
493
494 ### image of the best m/z
495 minimumy = min(coord(msidata)[,2])
496 maximumy = max(coord(msidata)[,2])
497 print(image(msidata, mz = topLabels(msidata.ssc)[1,1], normalize.image = "linear", contrast.enhance = "histogram",smooth.image="gaussian", ylim= c(maximumy+0.2*maximumy,minimumy-0.2*minimumy), main="best m/z heatmap"))
498
499 ## m/z and pixel information output
500 ssc_classes = data.frame(msidata.ssc\$classes[[1]])
501 pixel_names = gsub(", y = ", "_", names(pixels(msidata)))
502 pixel_names = gsub(" = ", "y_", pixel_names)
503 x_coordinates = matrix(unlist(strsplit(pixel_names, "_")), ncol=3, byrow=TRUE)[,2]
504 y_coordinates = matrix(unlist(strsplit(pixel_names, "_")), ncol=3, byrow=TRUE)[,3]
505 ssc_classes2 = data.frame(pixel_names, x_coordinates, y_coordinates, ssc_classes)
506 colnames(ssc_classes2) = c("pixel names", "x", "y","predicted condition")
507 ssc_toplabels = topLabels(msidata.ssc, n=$type_cond.method_cond.ssc_analysis_cond.ssc_toplabels)
508 write.table(ssc_toplabels, file="$mzfeatures", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
509 write.table(ssc_classes2, file="$pixeloutput", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
510
511 ## image with predicted classes
512 prediction_df = cbind(coord(msidata.ssc)[,1:2], ssc_classes)
513 colnames(prediction_df) = c("x", "y", "predicted_classes")
514
515 prediction_plot = ggplot(prediction_df, aes(x=x, y=y, fill=predicted_classes))+
516 geom_tile() +
517 coord_fixed()+
518 ggtitle("Predicted condition for each pixel")+
519 theme_bw()+
520 theme(text=element_text(family="ArialMT", face="bold", size=15))+
521 theme(legend.position="bottom",legend.direction="vertical")+
522 guides(fill=guide_legend(ncol=4,byrow=TRUE))
523 coord_labels = aggregate(cbind(x,y)~predicted_classes, data=prediction_df, mean, na.rm=TRUE, na.action="na.pass")
524 coord_labels\$file_number = gsub( "_.*$", "", coord_labels\$predicted_classes)
525 print(prediction_plot)
526
527
528 ## optional output as .RData
529 #if $output_rdata:
530 save(msidata.ssc, file="$classification_rdata")
531 #end if
532
533 #end if
534 #end if
535
536
537
538 ######################## II) Prediction #############################
539 #############################################################################
540
541 #elif str($type_cond.type_method) == "prediction":
542 print("prediction")
543
544 training_data = loadRData("$type_cond.training_result")
545
546 #if str($type_cond.new_y_values_cond.new_y_values) == "new_response":
547 print("new response")
548
549 new_y_tabular = read.delim("$type_cond.new_y_values_cond.new_response_file", header = $type_cond.new_y_values_cond.new_tabular_header, stringsAsFactors = FALSE)
550 new_y_input = new_y_tabular[,c($type_cond.new_y_values_cond.column_new_x, $type_cond.new_y_values_cond.column_new_y, $type_cond.new_y_values_cond.column_new_response)]
551 colnames(new_y_input)[1:2] = c("x", "y")
552 ## merge with coordinate information of msidata
553 msidata_coordinates = cbind(coord(msidata)[,1:2], c(1:ncol(msidata)))
554 colnames(msidata_coordinates)[3] = "pixel_index"
555 merged_response = merge(msidata_coordinates, new_y_input, by=c("x", "y"), all.x=TRUE)
556 merged_response[is.na(merged_response)] = "NA"
557 merged_response = merged_response[order(merged_response\$pixel_index),]
558 new_y_vector = as.factor(merged_response[,4])
559 prediction = predict(training_data,msidata, newy = new_y_vector)
560
561 #else
562 prediction = predict(training_data,msidata)
563 #end if
564
565 ## m/z and pixel information output
566 predicted_classes = data.frame(prediction\$classes[[1]])
567 pixel_names = gsub(", y = ", "_", names(pixels(msidata)))
568 pixel_names = gsub(" = ", "y_", pixel_names)
569 x_coordinates = matrix(unlist(strsplit(pixel_names, "_")), ncol=3, byrow=TRUE)[,2]
570 y_coordinates = matrix(unlist(strsplit(pixel_names, "_")), ncol=3, byrow=TRUE)[,3]
571 predicted_classes2 = data.frame(pixel_names, x_coordinates, y_coordinates, predicted_classes)
572 colnames(predicted_classes2) = c("pixel names", "x", "y","predicted condition")
573 predicted_toplabels = topLabels(prediction, n=$type_cond.predicted_toplabels)
574 write.table(predicted_toplabels, file="$mzfeatures", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
575 write.table(predicted_classes2, file="$pixeloutput", quote = FALSE, row.names = FALSE, col.names=TRUE, sep = "\t")
576
577 ## image with predicted classes
578
579 prediction_df = cbind(coord(prediction)[,1:2], predicted_classes)
580 colnames(prediction_df) = c("x", "y", "predicted_classes")
581
582 prediction_plot = ggplot(prediction_df, aes(x=x, y=y, fill=predicted_classes))+
583 geom_tile() +
584 coord_fixed()+
585 ggtitle("Predicted condition for each pixel")+
586 theme_bw()+
587 theme(text=element_text(family="ArialMT", face="bold", size=15))+
588 theme(legend.position="bottom",legend.direction="vertical")+
589 guides(fill=guide_legend(ncol=4,byrow=TRUE))
590 coord_labels = aggregate(cbind(x,y)~predicted_classes, data=prediction_df, mean, na.rm=TRUE, na.action="na.pass")
591 coord_labels\$file_number = gsub( "_.*$", "", coord_labels\$predicted_classes)
592 print(prediction_plot)
593
594 ## Summary table prediction
595 summary_table = summary(prediction)\$accuracy[[names(prediction@resultData)]]
596 summary_table2 = round(as.numeric(summary_table), digits=2)
597 summary_matrix = matrix(summary_table2, nrow=4, ncol=ncol(summary_table))
598 summary_table3 = cbind(rownames(summary_table), summary_matrix) ## include rownames in table
599 summary_table4 = t(summary_table3)
600 summary_table5 = cbind(c(names(prediction@resultData),colnames(summary_table)), summary_table4)
601 plot(0,type='n',axes=FALSE,ann=FALSE)
602 grid.table(summary_table5, rows= NULL)
603
604 ## optional output as .RData
605 #if $output_rdata:
606 msidata = prediction
607 save(msidata, file="$classification_rdata")
608 #end if
609
610 #end if
611
612 dev.off()
613
614 }else{
615 print("Inputfile has no intensities > 0 or contains NA values")
616 dev.off()
617 }
618
619
620 ]]></configfile>
621 </configfiles>
622 <inputs>
623 <expand macro="reading_msidata"/>
624 <conditional name="type_cond">
625 <param name="type_method" type="select" label="Analysis step to perform">
626 <option value="training" selected="True">training</option>
627 <option value="prediction">prediction</option>
628 </param>
629 <when value="training">
630
631 <param name="annotation_file" type="data" format="tabular" label="Load tabular file with pixel coordinates and their classes"
632 help="Three or four columns: x values, y values, response values, optionally fold values"/>
633 <param name="column_x" data_ref="annotation_file" label="Column with x values" type="data_column"/>
634 <param name="column_y" data_ref="annotation_file" label="Column with y values" type="data_column"/>
635 <param name="column_response" data_ref="annotation_file" label="Column with response (condition) values" type="data_column" help="This is the condition (pixel group) which will be classified"/>
636 <param name="column_fold" data_ref="annotation_file" optional="True" label="Column with fold values - only neccessary for cvapply" type="data_column" help="Each fold must contain pixels of all response groups and is used for cross validation"/>
637 <param name="tabular_header" type="boolean" label="Tabular files contain a header line" truevalue="TRUE" falsevalue="FALSE"/>
638
639 <conditional name="method_cond">
640 <param name="class_method" type="select" label="Select the method for classification">
641 <option value="PLS" selected="True">PLS-DA</option>
642 <option value="OPLS">OPLS-DA</option>
643 <option value="spatialShrunkenCentroids">spatial shrunken centroids</option>
644 </param>
645 <when value="PLS">
646
647 <conditional name="analysis_cond">
648 <param name="PLS_method" type="select" label="Crossvalidation or analysis">
649 <option value="cvapply" selected="True">cvApply</option>
650 <option value="PLS_analysis">PLS-DA analysis</option>
651 </param>
652 <when value="cvapply">
653 <param name="plscv_comp" type="text" value="1:2"
654 label="The number of PLS-DA components" help="For cvapply multiple values are allowed (e.g. 1,2,3 or 2:5)">
655 <expand macro="sanitizer_multiple_digits"/>
656 </param>
657 </when>
658 <when value="PLS_analysis">
659 <param name="pls_comp" type="integer" value="5"
660 label="The optimal number of PLS-DA components as indicated by cross-validations" help="Run cvApply first to optain optiaml number of PLS-DA components"/>
661 <param name="pls_scale" type="boolean" label="Data scaling" truevalue="TRUE" falsevalue="FALSE"/>
662 <param name="pls_toplabels" type="integer" value="100"
663 label="Number of toplabels (m/z features) which should be written in tabular output"/>
664 </when>
665 </conditional>
666 </when>
667
668 <when value="OPLS">
669
670 <conditional name="opls_analysis_cond">
671 <param name="opls_method" type="select" label="Analysis step to perform">
672 <option value="opls_cvapply" selected="True">cvApply</option>
673 <option value="opls_analysis">OPLS-DA analysis</option>
674 </param>
675
676 <when value="opls_cvapply">
677 <param name="opls_cvcomp" type="text" value="1:2"
678 label="The number of OPLS-DA components" help="For cvapply multiple values are allowed (e.g. 1,2,3 or 2:5)">
679 <expand macro="sanitizer_multiple_digits"/>
680 </param>
681 <param name="xnew_cv" type="boolean" truevalue="TRUE" falsevalue="FALSE" label="Keep new matrix"/>
682 </when>
683
684 <when value="opls_analysis">
685 <param name="opls_comp" type="integer" value="5"
686 label="The optimal number of OPLS-DA components as indicated by cross-validations" help="Run cvApply first to optain optiaml number of OPLS-DA components"/>
687 <param name="xnew" type="boolean" truevalue="TRUE" falsevalue="FALSE" label="Keep new matrix"/>
688 <param name="opls_scale" type="boolean" truevalue="TRUE" falsevalue="FALSE" label="Data scaling"/>
689 <param name="opls_toplabels" type="integer" value="100"
690 label="Number of toplabels (m/z features) which should be written in tabular output"/>
691 </when>
692 </conditional>
693 </when>
694
695 <when value="spatialShrunkenCentroids">
696 <conditional name="ssc_analysis_cond">
697 <param name="ssc_method" type="select" label="Analysis step to perform">
698 <option value="ssc_cvapply" selected="True">cvApply</option>
699 <option value="ssc_analysis">spatial shrunken centroids analysis</option>
700 </param>
701 <when value="ssc_cvapply"/>
702
703 <when value="ssc_analysis">
704 <param name="ssc_toplabels" type="integer" value="100"
705 label="Number of toplabels (m/z features) which should be written in tabular output"/>
706 </when>
707 </conditional>
708 <param name="ssc_r" type="text" value="2"
709 label="The spatial neighborhood radius of nearby pixels to consider (r)" help="For cvapply multiple values are allowed (e.g. 1,2,3 or 2:5)">
710 <expand macro="sanitizer_multiple_digits"/>
711 </param>
712 <param name="ssc_s" type="text" value="2"
713 label="The sparsity thresholding parameter by which to shrink the t-statistics (s)" help="For cvapply multiple values are allowed (e.g. 1,2,3 or 2:5)">
714 <expand macro="sanitizer_multiple_digits"/>
715 </param>
716 <param name="ssc_kernel_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">
717 <option value="gaussian">gaussian</option>
718 <option value="adaptive" selected="True">adaptive</option>
719 </param>
720
721 </when>
722 </conditional>
723 </when>
724
725 <when value="prediction">
726 <param name="training_result" type="data" format="rdata" label="Result from previous classification training"/>
727 <param name="predicted_toplabels" type="integer" value="100"
728 label="Number of toplabels (m/z features) which should be written in tabular output"/>
729 <conditional name="new_y_values_cond">
730 <param name="new_y_values" type="select" label="Should new response values be used">
731 <option value="no_new_response" selected="True">old response should be used</option>
732 <option value="new_response">load new response from tabular file</option>
733 </param>
734 <when value="no_new_response"/>
735 <when value="new_response">
736 <param name="new_response_file" type="data" format="tabular" label="Load tabular file with pixel coordinates and the new response"/>
737 <param name="column_new_x" data_ref="new_response_file" label="Column with x values" type="data_column"/>
738 <param name="column_new_y" data_ref="new_response_file" label="Column with y values" type="data_column"/>
739 <param name="column_new_response" data_ref="new_response_file" label="Column with new response values" type="data_column"/>
740 <param name="new_tabular_header" type="boolean" label="Tabular files contain a header line" truevalue="TRUE" falsevalue="FALSE"/>
741 </when>
742 </conditional>
743 </when>
744 </conditional>
745 <param name="output_rdata" type="boolean" label="Results as .RData output" help="Can be used to generate a classification prediction on new data"/>
746 </inputs>
747 <outputs>
748 <data format="pdf" name="classification_images" from_work_dir="classificationpdf.pdf" label = "${tool.name} on ${on_string}"/>
749 <data format="tabular" name="mzfeatures" label="${tool.name} on ${on_string}: features"/>
750 <data format="tabular" name="pixeloutput" label="${tool.name} on ${on_string}: pixels"/>
751 <data format="rdata" name="classification_rdata" label="${tool.name} on ${on_string}: results.RData">
752 <filter>output_rdata</filter>
753 </data>
754 </outputs>
755 <tests>
756 <test expect_num_outputs="3">
757 <param name="infile" value="testfile_squares.rdata" ftype="rdata"/>
758 <conditional name="type_cond">
759 <param name="type_method" value="training"/>
760 <param name="annotation_file" value= "pixel_annotation_file1.tabular" ftype="tabular"/>
761 <param name="column_x" value="1"/>
762 <param name="column_y" value="2"/>
763 <param name="column_response" value="4"/>
764 <param name="column_fold" value="3"/>
765 <param name="tabular_header" value="False"/>
766 <conditional name="method_cond">
767 <param name="class_method" value="PLS"/>
768 <conditional name="analysis_cond">
769 <param name="PLS_method" value="cvapply"/>
770 <param name="plscv_comp" value="2:4"/>
771 </conditional>
772 </conditional>
773 </conditional>
774 <output name="mzfeatures" file="features_test1.tabular"/>
775 <output name="pixeloutput" file="pixels_test1.tabular"/>
776 <output name="classification_images" file="test1.pdf" compare="sim_size" delta="2000"/>
777 </test>
778
779 <test expect_num_outputs="4">
780 <param name="infile" value="testfile_squares.rdata" ftype="rdata"/>
781 <conditional name="type_cond">
782 <param name="type_method" value="training"/>
783 <param name="annotation_file" value= "pixel_annotation_file1.tabular" ftype="tabular"/>
784 <param name="column_x" value="1"/>
785 <param name="column_y" value="2"/>
786 <param name="column_response" value="4"/>
787 <param name="tabular_header" value="False"/>
788 <conditional name="method_cond">
789 <param name="class_method" value="PLS"/>
790 <conditional name="analysis_cond">
791 <param name="PLS_method" value="PLS_analysis"/>
792 <param name="pls_comp" value="2"/>
793 <param name="pls_scale" value="TRUE"/>
794 <param name="pls_toplabels" value="100"/>
795 </conditional>
796 </conditional>
797 </conditional>
798 <param name="output_rdata" value="True"/>
799 <output name="mzfeatures" file="features_test2.tabular"/>
800 <output name="pixeloutput" file="pixels_test2.tabular"/>
801 <output name="classification_images" file="test2.pdf" compare="sim_size"/>
802 <output name="classification_rdata" file="test2.rdata" compare="sim_size"/>
803 </test>
804
805 <test expect_num_outputs="3">
806 <param name="infile" value="testfile_squares.rdata" ftype="rdata"/>
807 <conditional name="type_cond">
808 <param name="type_method" value="training"/>
809 <param name="annotation_file" value= "random_factors.tabular" ftype="tabular"/>
810 <param name="column_x" value="1"/>
811 <param name="column_y" value="2"/>
812 <param name="column_response" value="4"/>
813 <param name="column_fold" value="3"/>
814 <param name="tabular_header" value="False"/>
815 <conditional name="method_cond">
816 <param name="class_method" value="OPLS"/>
817 <conditional name="opls_analysis_cond">
818 <param name="opls_method" value="opls_cvapply"/>
819 <param name="opls_cvcomp" value="1:2"/>
820 <param name="xnew_cv" value="FALSE"/>
821 </conditional>
822 </conditional>
823 </conditional>
824 <output name="mzfeatures" file="features_test3.tabular"/>
825 <output name="pixeloutput" file="pixels_test3.tabular"/>
826 <output name="classification_images" file="test3.pdf" compare="sim_size"/>
827 </test>
828
829 <test expect_num_outputs="4">
830 <param name="infile" value="testfile_squares.rdata" ftype="rdata"/>
831 <conditional name="type_cond">
832 <param name="type_method" value="training"/>
833 <param name="annotation_file" value= "random_factors.tabular" ftype="tabular"/>
834 <param name="column_x" value="1"/>
835 <param name="column_y" value="2"/>
836 <param name="column_response" value="4"/>
837 <param name="tabular_header" value="False"/>
838 <conditional name="method_cond">
839 <param name="class_method" value="OPLS"/>
840 <conditional name="opls_analysis_cond">
841 <param name="opls_method" value="opls_analysis"/>
842 <param name="opls_comp" value="3"/>
843 <param name="xnew" value="FALSE"/>
844 <param name="opls_scale" value="FALSE"/>
845 <param name="opls_toplabels" value="100"/>
846 </conditional>
847 </conditional>
848 </conditional>
849 <param name="output_rdata" value="True"/>
850 <output name="mzfeatures" file="features_test4.tabular"/>
851 <output name="pixeloutput" file="pixels_test4.tabular"/>
852 <output name="classification_images" file="test4.pdf" compare="sim_size"/>
853 <output name="classification_rdata" file="test4.rdata" compare="sim_size"/>
854 </test>
855
856 <test expect_num_outputs="3">
857 <param name="infile" value="testfile_squares.rdata" ftype="rdata"/>
858 <conditional name="type_cond">
859 <param name="type_method" value="training"/>
860 <param name="annotation_file" value= "pixel_annotation_file1.tabular" ftype="tabular"/>
861 <param name="column_x" value="1"/>
862 <param name="column_y" value="2"/>
863 <param name="column_response" value="3"/>
864 <param name="column_fold" value="4"/>
865 <param name="tabular_header" value="False"/>
866 <conditional name="method_cond">
867 <param name="class_method" value="spatialShrunkenCentroids"/>
868 <conditional name="ssc_analysis_cond">
869 <param name="ssc_method" value="ssc_cvapply"/>
870 <param name="ssc_r" value="1:2"/>
871 <param name="ssc_s" value="2:3"/>
872 <param name="ssc_kernel_method" value="adaptive"/>
873 </conditional>
874 </conditional>
875 </conditional>
876 <output name="mzfeatures" file="features_test5.tabular"/>
877 <output name="pixeloutput" file="pixels_test5.tabular"/>
878 <output name="classification_images" file="test5.pdf" compare="sim_size"/>
879 </test>
880
881 <test expect_num_outputs="4">
882 <param name="infile" value="testfile_squares.rdata" ftype="rdata"/>
883 <conditional name="type_cond">
884 <param name="type_method" value="training"/>
885 <param name="annotation_file" value= "random_factors.tabular" ftype="tabular"/>
886 <param name="column_x" value="1"/>
887 <param name="column_y" value="2"/>
888 <param name="column_response" value="4"/>
889 <conditional name="method_cond">
890 <param name="class_method" value="spatialShrunkenCentroids"/>
891 <conditional name="ssc_analysis_cond">
892 <param name="ssc_method" value="ssc_analysis"/>
893 <param name="ssc_toplabels" value="20"/>
894 </conditional>
895 <param name="ssc_r" value="2"/>
896 <param name="ssc_s" value="2"/>
897 <param name="ssc_kernel_method" value="adaptive"/>
898 </conditional>
899 </conditional>
900 <param name="output_rdata" value="True"/>
901 <output name="mzfeatures" file="features_test6.tabular"/>
902 <output name="pixeloutput" file="pixels_test6.tabular"/>
903 <output name="classification_images" file="test6.pdf" compare="sim_size"/>
904 <output name="classification_rdata" file="test6.rdata" compare="sim_size" />
905 </test>
906
907 <test expect_num_outputs="4">
908 <param name="infile" value="testfile_squares.rdata" ftype="rdata"/>
909 <conditional name="type_cond">
910 <param name="type_method" value="prediction"/>
911 <param name="training_result" value="test2.rdata" ftype="rdata"/>
912 <conditional name="new_y_values_cond">
913 <param name="new_y_values" value="new_response"/>
914 <param name="new_response_file" value="pixel_annotation_file1.tabular" ftype="tabular"/>
915 <param name="column_new_x" value="1"/>
916 <param name="column_new_y" value="2"/>
917 <param name="column_new_response" value="4"/>
918 <param name="new_tabular_header" value="False"/>
919 </conditional>
920 </conditional>
921 <param name="output_rdata" value="True"/>
922 <output name="mzfeatures" file="features_test7.tabular"/>
923 <output name="pixeloutput" file="pixels_test7.tabular"/>
924 <output name="classification_images" file="test7.pdf" compare="sim_size"/>
925 <output name="classification_rdata" file="test7.rdata" compare="sim_size" />
926 </test>
927 </tests>
928 <help>
929 <![CDATA[
930
931
932 @CARDINAL_DESCRIPTION@
933
934 -----
935
936 This tool provides three different Cardinal functions for supervised classification of mass-spectrometry imaging data.
937
938 @MSIDATA_INPUT_DESCRIPTION@
939 - For training: tabular file with condition and fold for each pixel: Two columns for pixel coordinates (x and y values); one column with the condition for the pixel, which will be used for classification; for the cross validation (cvapply) another column with a fold is necessary, each fold must contain pixels of all response groups and is used for cross validation. Condition and fold columns are treated as factor to perform discriminant analysis (also when numeric values are provided).
940
941 ::
942
943 x_coord y_coord condition fold
944 1 1 A f1
945 2 1 A f2
946 3 1 A f3
947 1 2 B f1
948 2 2 B f2
949 3 2 B f3
950 ...
951 ...
952
953
954 - For prediction: RData output from previous classification run is needed as input, optionally new response values can be loaded with a tabular file containing x values, y values and the response
955
956
957 **Options**
958
959 - PLS-DA: partial least square discriminant analysis
960 - O-PLS-DA: Orthogonal partial least squares discriminant analysis
961 - Spatial shrunken centroids
962
963 **Tips**
964
965 - The classification function will only run on files with valid intensity values (NA are not allowed)
966 - Only a single input file is accepted, several files have to be combined previously, for example with the msi_combine tool.
967
968
969 **Output**
970
971 - Pdf with the heatmaps and plots for the classification
972 - Tabular file with information on m/z features and pixels: toplabels/classes
973 - Optional: RData output that can be used to predict new data or to explore the results more deeply with the Cardinal package in R
974
975 ]]>
976 </help>
977 <expand macro="citations"/>
978 </tool>