Metabolomics Society Forum

library(igraph)
dfr <- read.csv('GnpsEdge_0_20202251139_COMBINED.csv') # combined edge list from GNPS
edgelist <- dfr[,c("ID1", "ID2")] 
# I use "as.character" below because if the vector is integers instead of characters, 
# the resulting components output includes ALL integers 
# from 1 to the max value in edges. Thus, new nodes can be created artificially.  
edges <- as.character(as.vector(t(edgelist))) 
g1 <- graph(edges, directed=FALSE)
comp1 <- components(g1)
comp.membership <- data.frame(cbind(node=as.numeric(names(comp1$membership))
                                  , annotation.group=as.numeric(comp1$membership)))

j <- grep("NEG_DDA", fileNames(xdata)) # index of DDA files
MS2.file.paths <- fileNames(xdata)[j] # file paths of DDA files
MS2.file.names <- gsub(".*/", "", MS2.file.paths) # names of DDA files
fd <- featureDefinitions(xdata) # extract feature defs as a new object
fv <- featureValues(xdata) # extract feature values as a new object
fv.filtered <- fv[, colnames(fv) %in% MS2.file.names] # filter feature values to include only DDA samples
cp <- chromPeaks(xdata) # extract chromatographic peaks as a new object
cp <- cbind(rowid=seq(nrow(cp)), cp) # add a temporary column to cp for matching with 'peakidx' in feature definitions (suggested by CoreyG)
cp.filtered <- cp[which(cp[,which(colnames(cp) == "sample")] %in% j),] # filter cp to include only DDA samples

peakidx.filtered <- list() # create object to store results of loop below
for(i in 1:length(fd@listData$peakidx)){
     temp <- match(fd@listData$peakidx[[i]], cp.filtered[,"rowid"])
     peakidx.filtered[[i]] <- temp[which(!is.na(temp))]
} # end loop; this filters to include only peakidx in fd that correspond to peaks in DDA samples
fd.filtered <- fd # duplicate original feature definitions
fd.filtered@listData$peakidx <- peakidx.filtered # overwrite peakidx in duplicated fd with filtered peakidx generated by loop above

xdata.filtered <- filterFile(xdata, MS2.file.names, keepAdjustedRtime=TRUE) # create a new xdata object with only DDA samples; correspondence results are removed and will be added back in below

file_factor <- factor(cp.filtered[, "sample"]) # outputs a vector to correspond peak number with sample number
cp.filtered.split <- split.data.frame(cp.filtered, f=file_factor) # splits cp.filtered (a dataframe) into a list of dataframes (one for each remaining DDA sample)

cp.filtered.v2 <- c() # new object for storing results
for(i in 1:length(cp.filtered.split)){
    cp.filtered.split[[i]][,"sample"] <- i
    cp.filtered.v2 <- rbind(cp.filtered.v2, cp.filtered.split[[i]])
} #end loop; renumbers samples in filtered cp list, starting at 1

chromPeaks(xdata.filtered) <- cp.filtered.v2 # GOAL !!!
featureDefinitions(xdata.filtered) <- fd.filtered # GOAL !!!

# export MS1 and MS2 features
filteredMs2Spectra <- featureSpectra(xdata.filtered, return.type = "Spectra")
filteredMs2Spectra <- clean(filteredMs2Spectra, all = TRUE)
filteredMs2Spectra <- formatSpectraForGNPS(filteredMs2Spectra)
writeMgfData(filteredMs2Spectra, paste(run.name, "_ms2spectra_all.mgf", sep=""))

# generate data table (i.e. peak table) in format needed for GNPS/FBMN
setwd(save.dir)
featuresDef <- featureDefinitions(xdata.filtered)
featuresIntensities <- featureValues(xdata.filtered, value = "into")
dataTable <- merge(featuresDef, featuresIntensities, by = 0, all = TRUE)
dataTable <- dataTable[, !(colnames(dataTable) %in% c("peakidx"))]
write.table(dataTable, paste(run.name, "_xcms_all.txt", sep=""), sep = "\t", quote = FALSE, row.names = FALSE) # UPLOAD TO GNPS for FBMN

#export MS2 features only
setwd(save.dir)
filteredMs2Spectra_maxTic <- combineSpectra(filteredMs2Spectra,
                                            fcol = "feature_id",
                                            method = maxTic)
writeMgfData(filteredMs2Spectra_maxTic, paste(run.name, "_ms2spectra_maxTic.mgf", sep="")) # UPLOAD TO GNPS for FBMN
filteredDataTable <- dataTable[which(dataTable$Row.names %in% filteredMs2Spectra@elementMetadata$feature_id),]
write.table(filteredDataTable, paste(run.name, "_xcms_onlyMS2.txt", sep=""), sep = "\t", quote = FALSE, row.names = FALSE) #NOTE: apparently this is the table that can be used with GNPS/FBMN

> featureValues(xdata.filtered) <- fv.filtered
Error in featureValues(xdata.filtered) <- fv.filtered : 
  could not find function "featureValues<-"

featureDefinitions(xdata.filtered)<-fd.filtered 

> all(fv.filtered == featureValues(xdata.filtered))
[1] FALSE

pks <- chromPeaks(xdata) # extract chromatographic peaks as a separate object
xdata_filtered <- filterMsLevel(as(xdata, "OnDiskMSnExp"), 2L) # extract MS2 data as a separate object
## Split data per file
file_factor <- factor(pks[, "sample"]) #define each sample as a separate factor
pks <- split.data.frame(pks, f = file_factor) # make a separate dataframe for each sample
xdata_filtered <- lapply(as.integer(levels(file_factor)), filterFile, object = xdata_filtered)

## You then need to loop through xdata_filtered and pks for the samples you need. Each entry in xdata_filtered becomes 'x'; and 'pks' is the corresponding entry in the pks list.
n <- 1 # index in xdata_filtered and pks
method = 'closest_mz'

sps <- spectra(xdata_filtered[[n]])
pmz <- precursorMz(xdata_filtered[[n]])
rtm <- rtime(xdata_filtered[[n]])

#https://github.com/sneumann/xcms/blob/557b936967271690140e19224be707d87ea63168/R/functions-XCMSnExp.R#L1877
## Make sure you define all the required parameters i.e. method = 'closest_mz'
res <- vector(mode = "list", nrow(pks[[n]]))
for (i in 1:nrow(pks[[n]])) {
   if (is.na(pks[[n]][i, "mz"]))
     next
   idx <- which(pmz >= pks[[n]][i, "mzmin"] & pmz <= pks[[n]][i, "mzmax"] &
                  rtm >= pks[[n]][i, "rtmin"] & rtm <= pks[[n]][i, "rtmax"])
   if (length(idx)) {
     if (length(idx) > 1 & method != "all") {
       if (method == "closest_rt")
         idx <- idx[order(abs(rtm[idx] - pks[[n]][i, "rt"]))][1]
       if (method == "closest_mz")
         idx <- idx[order(abs(pmz[idx] - pks[[n]][i, "mz"]))][1]
       if (method == "signal") {
         sps_sub <- sps[idx]
         ints <- vapply(sps_sub, function(z) sum(intensity(z)),
                        numeric(1))
         idx <- idx[order(abs(ints - pks[[n]][i, "maxo"]))][1]
       }
     }
     res[[i]] <- lapply(sps[idx], function(z) {
       z@fromFile = fromFile
       z
     })
   }
 }

Error in (function (cl, name, valueClass)  : 
  assignment of an object of class “standardGeneric” is not valid for @‘fromFile’ in an object of class “Spectrum2”; is(value, "integer") is not TRUE

j <- grep("NEG_DDA", fileNames(xdata))
MS2.file.paths <- fileNames(xdata)[j]
MS2.file.names <- gsub(".*/", "", MS2.file.paths)

# extracting parts of xdata & filtering them ------------------
fd <- featureDefinitions(xdata)
fv <- featureValues(xdata)
fv.filtered <- fv[, colnames(fv) %in% MS2.file.names]
cp <- chromPeaks(xdata)
cp.filtered <- cp[which(cp[,which(colnames(cp) == "sample")] %in% j),]

# filtering files in xdata -----------------------------------------
xdata.filtered <- filterFile(xdata, MS2.file.names, keepAdjustedRtime=TRUE)

# changing sample numbers in filtered, extracted chromatographic peak object to match those in filtered xdata object
# if the sample numbers do not match, an error is generated
file_factor <- factor(cp.filtered[, "sample"])
cp.filtered.split <- split.data.frame(cp.filtered, f=file_factor)
cp.filtered.v2 <- c()
for(i in 1:length(cp.filtered.split)){
    cp.filtered.split[[i]][,"sample"] <- i
    cp.filtered.v2 <- rbind(cp.filtered.v2, cp.filtered.split[[i]])
} #end for loop

# incorporate filtered peaks back into filtered xdata
chromPeaks(xdata.filtered) <- cp.filtered.v2

featureDefinitions(xdata.filtered) <- fd

Error in validObject(object) : 
  invalid class “XCMSnExp” object: Some of the indices in column 'peakidx' of element 'featureDefinitions' do not match rows of the 'chromPeaks' matrix!

peak.list <- as.numeric(gsub("CP", "", row.names(chromPeaks(xdata.filtered))))
for(i in 1:length(fd@listData$peakidx)){ # this loop removes chromatographic peaks in "fd" that are not present in cp.filtered.v2 (i.e. chromPeaks(xdata.filtered) )
    k <- which(fd@listData$peakidx[[i]] %in% peak.list)
    fd@listData$peakidx[[i]] <- fd@listData$peakidx[[i]][k]
} # end for loop

featureDefinitions(xdata.filtered) <- fd