MSnbase contributors 2010 - 2016 2017

4 minute read

MSnbase (or here on GitHub) (Gatto and Lilley, 2010) is one of my main software projects. I started working on the package when I moved to Cambridge in 2010. It offers a way to import, manipulate and process raw mass spectrometry and quantitative proteomics data in R.

Since then, I benefited from quite a few contributions (which I already briefly highlighted here). In this post, I want to give a few more details about and credit to the contributors.

The figure below summarises the contributors over time.

MSnbase contributors

The first data point on the plot is Mon Oct 4 23:35:23 2010, and corresponds to the very first git commit (with typo) in the GitHub repository:

commit 7cb6b1a598d0b2ed55234f75229b925ceb26afaa
Author: Laurent Gatto <>
Date:   Mon Oct 4 23:35:23 2010 +0100

    Inital commit - package dir structure and R code
    All the R code files, some with roxygen incode documentation
    as well as the R package structure are committed. Included
    are also DESCRIPTION, NAMESPACE, NEWS (empty) and
    A dataset in mzXML format, dummyiTRAQ.mzXML in inst/extdata
    will serve as testing file.

Since then, the commits have been quite regular, except for 2012.

GitHub commits

Guangchuang Yu

Guangchuang contributed the plotMzDelta function in June 2011. The function produces a figure used as a quality for MS2 spectra, as detailed in Foster et al. 2011. All differences between neighbouring peaks in MS2 spectra are calculated and plotted as a histogram. Assuming good peptide fragmentation and absence of contamination, the histogram should feature peaks corresponding to amino acids.

plotMzDelta output

He is also the author of the first implementation of the readMgfData that, as the name implies, reads mgf data (thanks for the reminder).

Samuel Wieczorek and Cosmin Lazar

The contributions of Sam and Cosmin in February 2014 stem from our work on missing value imputation in quantitative proteomics (Lazar et al. 2016) and have materialised in improvements in the impute function.

Vlad Petyuk

Vlad’s main contrition was in the combineFeatures function that aggregates low level features, in March 2014. He contributed the redundancy handler, that defines how to handle peptides that can be associated to multiple higher-level features (proteins).

The ggplot2-based implementation of image, that produces a simple heatmap of the quantitative data also come from him, and is based on his own vp.misc::image_msnset implementation.

Facetted image

Thomas Naake

Thomas visited the group as an Erasmus student from April to June 2014 and implemented the first version of the pRolocGUI package (and here on GitHub). During this work, we discussed about features that would be needed for the interactive visualisation that ended up being implemented/added to MSnbase and then used in the GUIs. The main one I can remember is the FreaturesOfInterest class, that stores an arbitrary set of features (proteins) that can then conveniently highlighted on a PCA plot using the highlightOnPlot function from the pRoloc package.

x <- FeaturesOfInterest(description = "A test set of features of interest",
                        fnames = featureNames(tan2009r1)[1:10],
                        object = tan2009r1)
highlightOnPlot(tan2009r1, x)
highlightOnPlot(tan2009r1, x, labels = TRUE, pos = 3)

Highlight features of interest

Sebastian Gibb

Sebastian visited the group for 3 months in 2014. He did a lot of work on MSnbase and synapter (here) and is still active. Among his many contributions are the addIdentificationData, that adds identification data from mzid files to raw (MSnExp objects) and quantitative (MSnSet objects) data. He also added various raw data processing functions (such as smoothing, peak picking) by leveraging existing code in his MALDIQuant package and support for label-free MS2 quantitation. He also worked on spectra comparison, annotation and visualisation, as illustrated below.

Spectum annotation and comparison

Richie Cotton

Richie contributed in supporting mzTab version 1.0, as described in issue #41 from June 2015. I updated his code to fit into the MSnbase infrastructure and annotate some the ontology controlled parameters (using the rols package).

Martina Fischer

In June 2015, Martina contributed a whole new method for feature aggregation, termed iPQF (for Isobaric Protein Quantification based on Features)). iPQF is a new peptide-to-protein summarisation method using peptide spectra characteristics to improve protein quantification. All details in Fischer and Renard, 2016, readily available using combineFeatures(..., method = "iPQF").

Johannes Rainer

Johannes has been instrumental in the recent (October 2016) release of MSnbase version 2.0. During summer 2016, we worked on a new backend for raw data. Instead of loading spectra into memory, as in the original MSnExp implementation, the alternative implementation accesses the raw data from the hard drive on-the-fly only when it is needed. This is made possible by the fast on-disk access provided by the mzR package (here on GitHub) that uses the proteowizard C/C++ code base under the hood. For more details and a direct comparison, see the benchmarking vignette.

This joint work with Johannes aims at providing a common and efficient infrastructure for mass spectrometry data that can be used by the proteomics and metabolomics developers.

Arne Smits

Arne developes the DEP package for differential enrichment analysis of proteomics data and has contributed MSnSet to/from SummarizedExperiment methods, to facilitate the inter-operability between his package and MSnbase.

Seeing that MSnbase is used and attracts attention from other developers is a great reward for me. Thank you all for your valuable contributions!

Updates Since its first publication (2016-11-27), this post has been ammended to add Arne Smits’ contribution (2018-01-04).

Leave a Comment