When doing mass spectrometry on highly post-translationally modified proteins, the interpretation of the mass data is performed through a perpetual cycling in the following three steps :

• formulation of a structural hypothesis;
• calculation of the mass of the molecular species that embodies that specific hypothesis;
• comparison of the calculated mass with the masses actually measured by mass spectrometry.

The first scientific project that I was involved in was about the polyglutamylation of neuronal tubulins. I was tasked with the deciphering of the structure of the polyglutamylated C-terminal tail of these proteins. That work went through hundreds of theoretical mass computations which were done by hand.

Later I developed a project about the molecular diversity of another protein, telokin, which boasted the following molecular characteristics :

• differential translation initiation sites with removal or conservation of the initiating methionine residue;
• acetylation of the N-terminal residue;
• variety of C-terminal deglutamylation reactions (up to six glutamates were removed from the C-terminus of the protein);
• phosphorylation of one serine residue;
• genetic polymorphism at one position.

In both the projects above, I could clearly envision a software that would allow biochemists to easily predict the mass of molecular species corresponding to structural hypotheses made on top of a protein sequence. That software had to make this easily performed : ---"what would be the mass of this peptide if it were acetylated on its N-terminal residue and phosphorylated on that serine residue at the same time, or not."

massXpert version 1

I had the opportunity to develop that software during my post-doc at the Ecole polytechnique-X in the Institut européen de Chimie et Biologie de Bordeaux (IECB-X), Talence. It was published in : Rusconi, F. and Belghazi, M. 2002. Desktop prediction/analysis of mass spectrometric data in proteomic projects by using massXpert. Bioinformatics, 18:644-645.

GNU polyXmass

I later switched from the Microsoft Windows platform to the GNU/Linux platform as an everyday computing setup and thus had to port massXpert to the new environment. I settled on the Gtk+ graphics library to provide the windowing environment. During the port, a rather large amount of features were added to the software. One of the main improvements was certainly the fact that the chemical data defining the polymer chemistry were no more hard-coded in the software. Indeed, while massXpertV1 had all the atoms and protein chemistry entities hard-coded in it, GNU polyXmass allowed the user define all the entities required to fully characterize the polymer chemistry at hand. Once the polymer chemistry had been defined, it became available to edit polymer sequences and to perform biochemical and mass spectrometric simulations. This software has been published in: Rusconi, F. 2006. GNU polyxmass: a software framework for mass spectrometric simulations of linear (bio-)polymeric analytes. BMC Bioinformatics, 7:226-242.

massXpert version 2

Following the publication of the GNU polyXmass software suite, I received a number of messages of colleagues alledging the non-portability of the software as one cause of not using it. I also started questioning the usability of the object-oriented features in libgobject, the library that served as the basis for the pseudo-object-oriented code in Gtk+. For the development of the project, I needed to use C++ and I discovered at that time that the Qt libraries had been made Free Software even on Windows. Indeed, I was looking for good libraries that were fully portable and that were in an object-oriented language that I could use : C++. I settled on the Qt libraries by Trolltech (now Nokia), which were at that time already excellently designed, truly portable and robust (the GNU/Linux KDE desktop environment is based onto these libraries).

In two or three weeks, all the non-GUI (graphical user interface) code was ported to C++ using Qt. That was very exciting to see that the very same code would compile on Debian GNU/Linux (my development platform), on MS-Windows and on MacOSX. It took much more to fully port all the GUI stuff. As before, the port was an opportunity to redesign some architectural flaws present in the GNU polyXmass implementation and to really exploit the C++ features of the Qt libraries. The new software environment was published in: Rusconi, F. 2009. massXpert 2: a cross-platform software environment for polymer chemistry modelling and simulation/analysis of mass spectrometric data. Bioinformatics, 25: 2741-2742 and is located at http://www.massxpert.org.