The R code given in the file Supplementary_AlignmentFunctions.txt is for the alignment of preprocessed LC/MS spectra. R is freely available statistical software, downloadable from http://cran.r-project.org/

To use the R code in Supplementary_AlignmentFunctions.txt install R on your computer and follow the instructions given below.



#######################################################################
####                                                               ####
#### workflow for the complete rentention-time alignment procedure ####
####                                                               ####
#######################################################################

1. load LC/MS data

The preprocessed (peak-picked) data from all LC/MS runs has to be combined to one 'map'. This map has to be loaded into your R workspace as a matrix called "data" with columnnames "RT", "MZ", "Intens", and "Sample". Further columnns are possible but not necessary.
- "RT" contains retention time of a peak
- "MZ" contains mass-to-charge value of a peak
- "Intens" contains intensity of a peak
- "Sample" indicates from which LC/MS run the measurement originates. Should be numbers from 1 to n, with n being the number of LC/MS runs.


2. set parameters for alignment functions

h = 1/2 * expected mass accuracy (in ppm) of mass spectrometer
diff.rt = expected maximal retention-time difference (in minutes) between LC/MS runs
nmiss = default should be about 1/3 * n; higher values will result in more 'well-behaved' clusters
ndup = default should be 0; higher values will result in more 'well-behaved' clusters
al.type = "global" to use linear regression, "local" to use loess regression
method = default should be "BIC" ; type of model selection criterion for span selection, i.e., "AIC", "BIC", ...
width = default should be 20; can be increased for less complex LC/MS samples
max.obs = default should be 400, but can be increased if not enough 'well-behaved' clusters are found
over = should be set to 1 (Da)
path = path to use for saving results in (within quotation marks); Note that R needs a double backslash "\\" for each backslash "\"

Further information on the parameters and their usage within the functions can be found directly in the R code. 

3. copy the functions from Supplementary_AlignmentFunctions.txt into the R workspace

4. alignment

The alignment can be obtained through the following code (to be typed into the R workspace):

  data.sort <- data[order(data$MZ),]
  list.good.data <- good.group.all2(data.sort,width=width,over=over,max.obs=max.obs,nmiss=nmiss,ndup=ndup,h=h,diff.rt=diff.rt)

  if (al.type=="local") {
    rtdev.data <- comp.RTcor(list.good.data,data=simu.data, method=method)
    pdf(paste(path,"RTcurves.pdf",sep=""))
    plot.rtdev(rtdev.data$rtdev, spans=rtdev.data$opt.span)
    dev.off()
  }

  if (al.type=="global") {
    rtdev.data <- comp.RTcor.global(list.good.data,data=data.sort)
    pdf(paste(path,"RTcurves.pdf",sep=""))
    plot.rtdev.global(rtdev.data$rtdev)
    dev.off()
  }

  data.sort <- cbind(data.sort,"RTcorrected"=rtdev.data$RTcor)
  data.cor <- data.sort[order(data.sort$Sample),]
  write.table(data.cor,file=paste(path,"correctedDataset.txt",sep=""),sep="\t",quote=FALSE)


5. results

The alignment results are saved as a .txt file where the corrected retention times are given in the last column. Pictures of the derived alignment curves are given in the .pdf file.



Katharina Podwojski; 14:59 07.10.2008