Category: Automation and High-Throughput Technologies
Mass resolution is the critical feature of a mass spectrometer that defines its ability to distinguish and quantify individual compounds with closely related masses. Even marginal improvements of the resolution come at a much higher instrument costs that, however, does not always translate into higher analytical power: mass spectrometric analysis reports more individual peaks, but a small fraction of those actually belong to genuine analytes.
Cutting off peaks by their intensity is not practical since, along with noise, it eliminates biologically interesting low abundant compounds and undermines the impact of increased resolution.
Instead we propose to use the increased amount of peaks to measure peak reliability and cut off peaks that are not reliable.
To evaluate the peak reliability we first do a statistical analysis of peak detectability given a comparable set of preconditions, such as collision energy and mode, then we evaluate the reoccurrence of these peaks across multiple scans (Schuhmann et al, 2017).
The peak reliability measure is lowered if peaks do not repeat, given the same preconditions. We should note that while the peaks are grouped they are not averaged and the original peak information is maintained.
In contrast to many other filtering approaches this one does not rely on signal intensity, therefore low intensity signals are not discriminated against. And given a more reliable set of peaks, identification can be targeted instead of matching against an arbitrary collection of reference spectra. An example of targeted identification is with LipidXplorer where, given a description of lipid fragmentation pathways, peaks of informative molecular fragments can be recognized in highly convoluted spectra to matched the chemical structure of target molecules (Herzog et al, 2011)
The algorithm implemented in PeakStrainer software (Schuhmann et al, 2017) could eliminate >95% of noise peaks and drastically reduces time and memory load required to further process high resolution spectra of diverse analytical contexts. Data processing is no longer hampered by the apparent complexity of spectra, this allows us to take full analytical advantage of the high and ultra-high resolution delivered by modern mass spectrometers.
Eduardo Miranda Ackerman– Postdoc at Shevchenko Lab,, Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Sachsen, Germany
Postdoc at Shevchenko Lab,
Max Planck Institute of Molecular Cell Biology and Genetics
Dresden, Sachsen, Germany
I am Jacobo Miranda originally from Mexico, I completed my masters and PhD at TU Dresden, in the Computer Networks Chair at the Informatics faculty. I joined the Shevchenko lab as a postdoc with a focus on information extraction from Mass Spectrometry. I am interested in data analysis, machine learning and efficient processing of large datasets.