XFlow: An algorithm for extracting ion chromatograms

Abstract

13Mass spectrometry is a fundamental tool for modern proteomics. The increasing availability of 14 mass spectrometry data paired with the increasing sensitivity and fidelity of the instruments 15 necessitates new and more potent analytical methods. To that end, we have created and present 16 XFlow, a feature detection algorithm for extracting ion chromatograms from MS1 LC-MS data. 17 XFlow is a parameter-free procedurally agnostic feature detection algorithm that utilizes the latent 18 properties of ion chromatograms to resolve them from the surrounding noise present in MS1 data. 19 XFlow is designed to function on either profile or centroided data across different resolutions and 20 instruments. This broad applicability lends XFlow strong utility as a one-size-fits-all method for 21 MS1 analysis or target acquisition for MS2. XFlow is written in Java and packaged with JS- MS, 22 an open-source mass spectrometry analysis toolkit. 23 24 25Mass spectrometry is a popular approach for measuring the sample-bound content and quantity of 26 a variety of classes of molecules across a broad range of applications including pharmaceuticals, 2 27 forensics, biochemistry, and food science. All applications of mass spectrometry have a common 28 problem: the instrument itself does not provide measurements of molecules nor their identities, but 29 rather produces raw data that must be rendered human-interpretable through the application of data 30 processing algorithms. 31 32 According to community perceptions, advancements in software have lagged behind the steady 33 pace of instrumentation advancements. 1 Unlike other computational science fields (such as 34 genomics) where several foundational computational problems are regarded as solved, most mass 35 spectrometry users feel that significant problems in computational mass spectrometry remain 36 unsolved 1 despite (in some cases) dozens of published algorithms designed to address them. 2 37 Beyond user sentiment, the experimental influence of algorithm selection suggests that the analysis 38 and advancement of computational mass spectrometry algorithms is a valuable pursuit. 3 39 40 Mass spectrometry systems generate datasets that quantify counts of charged particles at specific 41 mass-to-charge (m/z) values. In liquid chromatography-mass spectrometry (LC-MS) systems, 42 these measurements are taken over the time (retention time or RT) required for the molecules to 43 elute from a chromatography column designed to slow or speed the migration of the molecules 44 depending on particular physico-chemical properties such as size, or polarity.45 46 Mapping the raw LC-MS data points to particular classes of molecule (say, a particular peptide at 47 gap present between isotopic-specific sub-signals (extracted ion chromatograms or XICs) in the 51 molecule's signal (see Figure 1). 52 Fig 1. 3d Isotopic Envelope. In this figure are five extracted ion chromatograms (XIC) bounded 53 by yellow rectangles. Each XIC is composed of points, each with m/z, RT and intensity...