A parser for raw and identification mass-spectrometry data

Introduction

The r BiocStyle::Biocpkg("mzR") package aims at providing a common, low-level interface to several mass spectrometry data formats, namely mzData [@Orchard2007], mzXML [@Pedrioli2004], mzML [@Martens2010] for raw data, and mzIdentML [@Jones2012], somewhat similar to the Bioconductor package affyio for affymetrix raw data. No processing is done in r BiocStyle::Biocpkg("mzR"), which is left to packages such as r BiocStyle::Biocpkg("xcms") [@Smith:2006, Tautenhahn:2008] or r BiocStyle::Biocpkg("MSnbase") [@Gatto:2012]. These packages also provide more convenient, high-level interfaces to raw and identification. data

Most importantly, access to the data should be fast and memory efficient. This is made possible by allowing on-disk random file access, i.e. retrieving specific data of interest without having to sequentially browser the full content nor loading the entire data into memory.

The actual work of reading and parsing the data files is handled by the included C/C++ libraries or backends. The mzRramp RAMP parser, written at the Institute for Systems Biology (ISB) is a fast and lightweight parser in pure C. Later, it gained support for the mzData format. The C++ reference implementation for the mzML is the proteowizard library [@Kessner08] (pwiz in short), which in turn makes use of the boost C++ (http://www.boost.org/) library. RAMP is able to access mzML files by calling pwiz methods. More recently, the proteowizard (http://proteowizard.sourceforge.net/) [@Chambers2012] has been fully integrated using the mzRpwiz backend for raw data, and is not the default option. The mzRnetCDF backend provides support to CDF-based formats. Finally, the mzRident backend is available to access identification data (mzIdentML) through pwiz.

The r BiocStyle::Biocpkg("mzR") package is in essence a collection of wrappers to the C++ code, and benefits from the C++ interface provided through the Rcpp package [@Rcpp11].

IMPORTANT New developers that need to access and manipulate raw mass spectrometry data are advised against using this infrastucture directly. They are invited to use the corresponding MSnExp (with on disk mode) from ther BiocStyle::Biocpkg("MSnbase") package instead. The latter supports reading multiple files at once and offers access to the spectra data (m/z and intensity) as well as all the spectra metadata using a coherent interface. The MSnbase infrastructure itself used the low level classes in mzR, thus offering fast and efficient access.

Mass spectrometry raw data

All the mass spectrometry file formats are organized similarly, where a set of metadata nodes about the run is followed by a list of spectra with the actual masses and intensities. In addition, each of these spectra has its own set of metadata, such as the retention time and acquisition parameters.

Spectral data access

Access to the spectral data is done via the peaks function. The return value is a list of two-column mass-to-charge and intensity matrices or a single matrix if one spectrum is queried.

Chromatogram access

Access to the chromatogram(s) is done using the chromatogram (or chromatograms) function, that return one (or a list of) data.frames. See ?chromatogram for details. This functionality is only available with the pwiz backend.

Identification result access

The main access to identification result is done via psms, score and modifications. psms and score will return the detailed information on each psm and scores. modifications will return the details on each modification found in peptide.

Metadata access

Run metadata is available via several functions such as instrumentInfo() or runInfo(). The individual fields can be accessed via e.g. detector() etc.

Spectrum metadata is available via header(), which will return a list (for single scans) or a dataframe with information such as the basePeakMZ, peaksCount, ... or, for higher-order MS the msLevel and precursor information.

Identification metadatais available via mzidInfo(), which will return a list with information such as the software, ModificationSearched, enzymes, SpectraSource and other information for this identification result.

The availability of this metadata can not always be guaranteed, and depends on the MS software which converted the data.

Example

mzXML/mzML/mzData files

A short example sequence to read data from a mass spectrometer. First open the file.

library(mzR)
library(msdata)

mzxml <- system.file("threonine/threonine_i2_e35_pH_tree.mzXML",
                     package = "msdata")
aa <- openMSfile(mzxml)

We can obtain different kind of header information.

runInfo(aa)
instrumentInfo(aa)
header(aa,1)

Read a single spectrum from the file.

pl <- peaks(aa,10)
peaksCount(aa,10)
head(pl)
plot(pl[,1], pl[,2], type="h", lwd=1)

One should always close the file when not needed any more. This will release the memory of cached content.

close(aa)

mzIdentML files

You can use openIDfile to read a mzIdentML file (version 1.1), which use the pwiz backend.

library(mzR)
library(msdata)

file <- system.file("mzid", "Tandem.mzid.gz", package="msdata")
x <- openIDfile(file)

mzidInfo function will return general information about this identification result.

mzidInfo(x)

psms will return the detailed information on each peptide-spectrum-match, include spectrumID, chargeState, sequence. modNum and others.

p <- psms(x)
colnames(p)

The modifications information can be accessed using modifications, which will return the spectrumID, sequence, name, mass and location.

m <- modifications(x)
head(m)

Since different software will use different scoring function, we provide a score to extract the scores for each psm. It will return a data.frame with different columns depending on software generating this file.

scr <- score(x)
colnames(scr)

Future plans

Other file formats provided by HUPO, such as mzQuantML for quantitative data [@Walzer:2013] are also possible in the future.

Session information {#sec:sessionInfo}

toLatex(sessionInfo())


Try the mzR package in your browser

Any scripts or data that you put into this service are public.

mzR documentation built on Nov. 20, 2020, 2 a.m.