Multiple Mass Spectrometry Fragmentation Trees Revisited: Boosting Performance and Quality

Mass spectrometry (MS) in combination with a fragmentation technique is the method of choice for analyzing small molecules in high throughput experiments. The automated interpretation of such data is highly non-trivial. Recently, fragmentation trees have been introduced for de novo analysis of tandem fragmentation spectra (MS), describing the fragmentation process of the molecule. Multiple-stage MS (MS) reveals additional information about the dependencies between fragments. Unfortunately, the computational analysis of MS data using fragmentation trees turns out to be more challenging than for tandem mass spectra. We present an Integer Linear Program for solving the Combined Colorful Subtree problem, which is orders of magnitude faster than the currently best algorithm which is based on dynamic programming. Using the new algorithm, we show that correlation between structural similarity and fragmentation tree similarity increases when using the additional information gained from MS. Thus, we show for the first time that using MS data can improve the quality of fragmentation trees.