Minifying files with RaMS

As of version 1.1.0, RaMS also has functions that allow irrelevant data to be removed from the file to reduce file sizes. Like grabMSdata, there’s one wrapper function minifyMSdata that accepts mzML or mzXML files, plus a vector of m/z values that should either be kept (mz_include) or removed (mz_exclude). The function then opens up the provided MS files and removes data points in the MS¹ and MS² spectra that fall outside the accepted bounds. mz_include is useful when only a few masses are of interest, as in targeted metabolomics. mz_exclude is useful when many masses are known to be contaminants or interfere with peakpicking/plotting abilities. This minification can shrink a file over three orders of magnitude, decreasing both processing time and memory allocation later in the pipeline.

This is also very useful for creating demo MS files - RaMS uses these functions to produce the sample data in extdata, with 6 MS files taking up less than 5 megabytes of disk space. Many other programs provide the ability to shrink files, but none (known to me) shrink files by excluding m/z values and instead can only remove certain retention times.

Below, we begin with a large MS file containing both MS¹ and MS² data and extract only the data corresponding to valine/glycine betaine and homarine.

library(RaMS)
msdata_files <- list.files(
  system.file("extdata", package = "RaMS"), full.names = TRUE, pattern = "mzML"
)[1:4]

initial_filename <- msdata_files[1]
output_filename <- gsub(x=paste0("minified_", basename(initial_filename)), "\\.gz", "")

masses_of_interest <- c(118.0865, 138.0555)
minifyMSdata(files = initial_filename, output_files = output_filename, 
             mz_include = masses_of_interest, ppm = 10, warn = FALSE)

Then, when we open the file up (with RaMS or other software) we are left with the data corresponding only to those compounds:

init_msdata <- grabMSdata(initial_filename)
msdata <- grabMSdata(output_filename)

knitr::kable(head(msdata$MS1, 3))

knitr::kable(head(msdata$MS2, 3))

par(mfrow=c(2, 1), mar=c(2.1, 2.1, 1.1, 0.1))
plot(init_msdata$BPC$rt, init_msdata$BPC$int, type = "l", main = "Initial BPC")
plot(msdata$BPC$rt, msdata$BPC$int, type = "l", main = "New BPC")

The minifyMSdata function is vectorized so the exact same syntax can be used for multiple files:

dir.create("mini_mzMLs/")
output_files <- paste0("mini_mzMLs/", basename(msdata_files))
output_files <- gsub(x=output_files, "\\.gz", "")

minifyMSdata(files = msdata_files, output_files = output_files, verbosity = 0,
             mz_include = masses_of_interest, ppm = 10, warn = FALSE)

mini_msdata <- grabMSdata(output_files, verbosity = 0)

library(ggplot2)
ggplot(mini_msdata$BPC) + geom_line(aes(x=rt, y=int, color=filename)) + theme_bw()

These new files are valid according to the validator provided in MSnbase, which means that most programs should be able to open them, but this feature is still experimental and may break on quirky data. If that happens, please feel free to submit a bug report at https://github.com/wkumler/RaMS/issues.

As an example of how I use this minification function, here’s the code used to create the minified files in the \extdata folder that ships with the package. This was especially useful because the package can’t be more than 5MB but it’s incredibly useful to include some standalone MS data for demos and vignettes like this one. I don’t actually run this code in the vignette itself to save compilation time but it will run if you test it yourself.

First, we identify the m/z values we’d like to keep in the minified files. For the demo data, I’ll use the Ingalls Lab list of targeted compounds - those we have authentic standards for.

raw_stans <- read.csv(paste0("https://raw.githubusercontent.com/",
                             "IngallsLabUW/Ingalls_Standards/",
                             "b098927ea0089b6e7a31e1758e7c7eaad5408535/",
                             "Ingalls_Lab_Standards_NEW.csv"))

mzs_to_include <- as.numeric(unique(raw_stans[raw_stans$Fraction1=="HILICPos",]$m.z))
# Include glycine betaine isotopes for README demo
mzs_to_include <- c(mzs_to_include, 119.0899, 119.0835)

Then, we download the raw MS data from the online repository into which it’s been deposited.

if(!dir.exists("vignettes/data"))dir.create("vignettes/data")
base_url <- "ftp://ftp.ebi.ac.uk/pub/databases/metabolights/studies/public/MTBLS703/FILES/"
chosen_files <- paste0(base_url, "170223_Smp_LB12HL_", c("AB", "CD", "EF"), "_pos.mzXML")
new_names <- gsub(x=basename(chosen_files), "170223_Smp_", "")

mapply(download.file, chosen_files, paste0("vignettes/data/", new_names), 
       mode = "wb", method = "libcurl")

MW_url <- paste0(
  "https://www.metabolomicsworkbench.org/data/file_extract_7z.php?",
  "A=ST002830_rawdata.zip&F=TCR_081023_Fu_WorkBench%252FS30657.mzXML"
)
download.file(MW_url, destfile = "vignettes/data/S30657.mzXML")

library(RaMS)

if(!dir.exists("inst/extdata"))dir.create("inst/extdata", recursive = TRUE)
init_files <- list.files("vignettes/data/", full.names = TRUE)
out_files <- paste0("inst/extdata/", basename(init_files))
minifyMSdata(files = init_files, output_files = out_files, warn = FALSE,
             mz_include = mzs_to_include, ppm = 20)

Now we have four minified mzXML files in our inst/extdata folder. However, we’d like to be able to demo the mzML functionality as well as that of mzXMLs, so we can use Proteowizard’s msconvert tool because RaMS can’t convert between mzML and mzXML or vice versa. You’ll need to install msconvert and add it to your path for this step.

We also use msconvert to trim the files by retention time, keeping data between 4 and 15 minutes.

Finally, we gzip the files to get them as small as possible, also using msconvert.

system("msconvert inst/extdata/*.mzXML -o inst/extdata/temp --noindex")
system("msconvert --mzXML inst/extdata/*.mzXML -o inst/extdata/temp --noindex")
system('msconvert inst/extdata/temp/*.mzML --filter \"scanTime [240,900]\" -o inst/extdata -g')
system('msconvert inst/extdata/temp/*.mzXML --mzXML --filter \"scanTime [240,900]\" -o inst/extdata -g')

And then for the last few steps, we again rename the files (since msconvert expands them to their full .raw names) and remove the ones we don’t need for the demos.

init_files <- list.files("inst/extdata", full.names = TRUE)
new_names <- paste0("inst/extdata/", gsub(x=init_files, ".*(Smp_|Extracts_)", ""))
file.rename(init_files, new_names)

unlink("inst/extdata/temp", recursive = TRUE)
file.remove(list.files("inst/extdata", pattern = "mzXML$", full.names = TRUE))
file.remove(paste0("inst/extdata/", c("LB12HL_CD.mzXML.gz", "LB12HL_EF.mzXML.gz")))

To check that the new files look ok, we can see if we can read them with RaMS and MSnbase.

MSnbase::readMSData(list.files("inst/extdata", full.names = TRUE)[1], msLevel. = 1)
RaMS::grabMSdata(new_names[1])

unlink("vignettes/data", recursive = TRUE)