Visual acuity interpretation based mass spectrometry (MS) and tandem mass spectrometry (MS / MS) spectrogram deconvolution

Abstract

In some examples, visual acuity interpretation based mass spectrometry (MS) and tandem mass spectrometry (MS / MS) spectrogram deconvolution may include receiving, for an ion to be identified, a plurality of expected ion spectrograms, where each of the plurality of expected ion spectrograms may include at least one expected peak profile; receiving an observed ion spectrum, the observed ion spectrum comprising at least one observed peak profile; and identifying characteristics of the expected ion spectrum and the observed ion spectrum. The characteristics may be analyzed to determine ion scores, where the highest ion score may be used to identify a corresponding expected ion spectrum. An indication and / or graphical user interface display of the expected ion spectrum corresponding to the highest ion score may be generated.

Description

[0001] Cross-reference to related applications This application claims priority to U.S. Provisional Patent Application No. 63 / 520,517, filed August 18, 2023, entitled “DECONVOLUTION BY VISUAL ACUITY-BASED INTERPRETATION OF MASS SPECTROMETRY (MS) AND TANDEM MASS SPECTROMETRY (MS / MS) SPECTRA,” and U.S. Provisional Patent Application No. 63 / 585,841, filed September 27, 2023, entitled “DECONVOLUTION BY VISUALACUITY-BASED INTERPRETATION OF MASS SPECTROMETRY (MS) AND TANDEM MASS SPECTROMETRY (MS / MS) SPECTRA,” both of which are incorporated herein by reference in their entirety. Background Technology

[0002] Regarding the annotation of fragment ions in mass spectra, the spectra can be derived from, for example, short peptides or complete proteins (such as monoclonal antibodies). Various types of deconvolution methods exist. Such deconvolution methods may be method-specific or require expert tuning when converting between low-noise peptide MS / MS spectra and highly complex top-down protein mass spectra. Furthermore, various types of software tools can be used to average the scans together to improve the signal-to-noise ratio. Attached Figure Description

[0003] The features of this disclosure are shown by way of example and are not limited to the following one or more figures, wherein like reference numerals indicate like elements, in the figures:

[0004] Figure 1 The layout of a mass spectrometry (MS) and tandem mass spectrometry (MS / MS) spectrum deconvolution device based on visual sensitivity interpretation is shown in the examples of this disclosure.

[0005] Figures 2A-2C The partitioning of visual acuity is illustrated to illustrate examples according to this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0006] Figure 3 The definition of an ion scoring binary classifier is shown to illustrate examples based on this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0007] Figures 4A-4C A consistent weighting scheme is demonstrated by applying it to all classifier metrics to illustrate the examples in this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0008] Figure 5 An exhaustive search for hydrogen transfer is demonstrated to illustrate examples based on this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0009] Figure 6 Annotated ions are shown to illustrate examples according to this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0010] Figure 7A and Figure 7B The settings for signal processing parameters are shown to illustrate examples based on this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0011] Figure 8A and Figure 8B Benchmark tests for ion scoring are presented to illustrate examples based on this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0012] Figure 9 The complete NIST monoclonal antibody (NIST mAb) is shown to illustrate the examples according to this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0013] Figure 10 Top-down deconvolution benchmarks are shown to illustrate the examples based on this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0014] Figure 11 User settings preferences are shown to illustrate examples based on this disclosure text. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0015] Figure 12The example demonstrates parameterless deconvolution to illustrate the concepts presented in this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0016] Figure 13 Deconvolution and decharge are demonstrated in ExDViewer to illustrate examples based on this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0017] Figure 14 Hybrid targeted / non-targeted deconvolution is demonstrated to illustrate the examples based on this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0018] Figure 15 Ion identification and monitoring of ion charge distribution and efficiency in MS / MS spectra under ECD conditions are shown to illustrate examples according to this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0019] Figure 16 Ion identification is demonstrated to illustrate examples according to this disclosure. Figure 1 Operation of visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution devices;

[0020] Figure 17 Example block diagrams are shown illustrating visual sensitivity-based interpretation of MS and MS / MS spectral deconvolution according to examples in this disclosure.

[0021] Figure 18 A flowchart illustrating an example method for deconvolution of MS and MS / MS spectral maps based on visual acuity interpretation, as shown in this disclosure, is presented; and

[0022] Figure 19 Further example block diagrams are shown for deconvolution of MS and MS / MS spectra based on visual acuity interpretation, according to another example of this disclosure. Detailed Implementation

[0023] For simplicity and illustrative purposes, this disclosure is described primarily through examples. Numerous specific details are set forth in the following description to provide a thorough understanding of this disclosure. However, it will be readily apparent that this disclosure can be practiced without being limited to these specific details. In other instances, some methods and structures have not been described in detail to avoid unnecessarily obscuring this disclosure.

[0024] Throughout this disclosure, the terms “a” and “an” are intended to mean at least one of a particular element. As used herein, the term “includes” means including but not limited to, and the term “including” means including but not limited to. The term “based on” means at least partially based on.

[0025] This paper discloses devices and methods for visual sensitivity-based interpretation of MS and MS / MS spectra deconvolution. The devices and methods disclosed herein provide, for example, averaging of profile spectra directly from raw vendor file formats, deconvolution of any unannotated spectra without expert tuning of input parameters, identification of fragment ions by precise isotopic distribution, manual processing of these identified ions, and generation of visualizations in easily disseminated, web-friendly applications. The devices and methods disclosed herein further provide an all-in-one tool for top-down and complete MS / MS deconvolution, which produces high-quality, interactive visualizations that can be shared in cloud environments.

[0026] The apparatus and methods disclosed herein, for example, provide greater than 80% sensitivity for all detectable ions to convert noisy centroid data into a resolved ion list. For instance, the apparatus and methods disclosed herein provide, for example, the generation of deconvolutioned ion lists with near-perfect accuracy (e.g., 99% accuracy) and extremely high interpretable ion sensitivity (e.g., >80%). This improvement in accuracy enables new automated (e.g., without human intervention) analytical techniques for complex MS and MS / MS spectra of intact proteins and large (>2 kDa) macromolecules, techniques that were previously impossible. One obstacle to the wider adoption of top-down and complete workflows is that achieving the desired high sequence coverage of ions may require expert tuning of instrument parameters. Instrument operators may typically investigate multiple instrument settings, such as collision energies, ion source conditions, and (in the case of a proprietary e-MSion ExD cell) up to six different voltage gradients to guide ions through and around an electron emission filament, which elicits desired reactions (such as charge reduction and molecular fragmentation). In this regard, the apparatus and methods disclosed herein enable near-perfect accuracy in automated ion identification, allowing for more robust autotuning of these instrument parameters, which currently relies on peak intensity measurements that can be susceptible to noise prevalent in the complex mass spectra described above. The analysis software can now examine a large set of putative ions in less than a second, determining which are detectable and reporting their cumulative scores. This information can be used as feedback from the apparatus and methods disclosed herein to the control software, allowing for real-time optimization of instrument parameters that maximizes coverage of the desired ions.

[0027] For the devices and methods disclosed herein, the identified ion information can be incorporated into a coverage plot, which can represent a medium for conveying which fragment ions were found for a given sequence. Users are typically interested in which amino acid (AA) residues and post-translational modifications (PTMs) are supported by fragment ions. For example, a biopharmaceutical company may be interested in whether its drug manufacturing platform is producing proteins / peptides with the desired AA sequences and PTMs. If an unexpected AA mutation or PTM is found, the Food and Drug Administration (FDA) may be concerned about the safety of the drug and suspend drug approval. If an unexpected mutation or PTM is found, the coverage plot may lack important fragment ion information because the expected ions from the unmodified sequence will not be detected. For some deconvolution methods, poor accuracy may mean that users cannot immediately trust these inconsistencies in the coverage and may therefore need to rely on alternative methods to generate MS / MS spectra with lower complexity, such as enzymatic digestion, depending on the specific protocol, which may significantly increase sample preparation time and cost. Such methods may also introduce their own PTMs, which may obscure the analysis of which PTMs are present in undigested samples.

[0028] In one example of the apparatus and methods disclosed herein, certain features can be implemented using software comprising a C++ server (e.g., for data processing) and a JavaScript front-end (e.g., for visualization) built on OpenMS, all available via a downloadable Windows installer or a browser-friendly web application. Data can be read from various vendor formats (e.g., the AGILENT vendor format) as well as from open-source mzML (the community standard for mass spectrometry data) and MGF (Mascot universal format) file types. When profile data is available, the data can be averaged together over multiple scans to improve the signal-to-noise ratio. Noise levels and peak selection parameters can be automatically adjusted (e.g., without human intervention) based on peak density. Deconvolution can utilize, for example, ab initio scoring methods that assign a score, for example, 0-15, to each predicted isotopic cluster based on how well the observed data matches theoretical data, taking into account profile peak data, ion centroid mass / charge number (m / z) error, hydrogen transfer, and overlap of adjacent peaks.

[0029] In one example of the apparatus and methods disclosed herein, certain features can be achieved using software to detect and visualize sequence coverage and associated MS or MS / MS fragment ions generated in top-down mass spectrometry, particularly based on electron fragmentation. In this regard, both electron capture dissociation (ECD) and electron transfer dissociation (ETD) can generate satellite ions (such as w-type ions) that distinguish between leucine / isoleucine (Leu / Ile) and aspartic acid / asparagine (Asp / Asn), but these are typically low in intensity and sometimes challenging to reliably identify in the presence of noise and adjacent peak complexity. Hydrogen transfer rearrangements can further complicate the analysis by broadening the expected isotopic distribution. Therefore, benchmarks can be performed using ECD and collision-induced dissociation (CID) spectra from different groups of samples containing peptides and proteins ranging from 1.4–29 kDa.

[0030] Regarding the apparatus and methods disclosed herein, the generated annotations for benchmarking include a high degree of accuracy. In this regard, the apparatus and methods disclosed herein provide top-down MS / MS analysis with a fully automated analysis workflow, thereby efficiently and rapidly matching experimental spectra with peptides and proteins by delivering results in real time (e.g., visualization of raw time-of-flight (TOF) data in < 30 seconds).

[0031] For the devices, methods, and non-transitory computer-readable media disclosed herein, the elements of the devices, methods, and non-transitory computer-readable media disclosed herein can be any combination of hardware and programming to implement the functionality of the respective elements. In some examples described herein, the combination of hardware and programming can be implemented in a variety of different ways. For example, the programming for the elements can be processor-executable instructions stored on a non-transitory machine-readable storage medium, and the hardware for the elements can include processing resources for executing those instructions. In these examples, a computing device implementing such elements can include a machine-readable storage medium storing instructions and processing resources for executing the instructions, or the machine-readable storage medium can be stored and accessed separately by the computing device and the processing resources. In some examples, some elements can be implemented in a circuit system.

[0032] Figure 1 The layout of an MS and MS / MS spectral deconvolution device (hereinafter also referred to as "device 100") based on visual sensitivity interpretation, as shown in the example of this disclosure, is illustrated.

[0033] refer to Figure 1 The device 100 may include an ion spectrometer 102, which is powered by at least one hardware processor (e.g., Figure 17 Hardware processor 1702 and / or Figure 19 The hardware processor 1904 performs the receiving of a plurality of expected ion spectra 106 for the ion to be identified 104. Each expected ion spectrum 108 in the expected ion spectrum 106 may include at least one expected peak profile 110. The ion spectrum analyzer 102 may receive observed ion spectra 112 for the ion to be identified 104, the observed ion spectrum including at least one observed peak profile 114. The ion spectrum analyzer 102 may identify the characteristics 116 of the expected ion spectrum 106 and the observed ion spectrum 112 based on the analysis of at least one expected peak profile 110 of each expected ion spectrum 106 and at least one observed peak profile of the observed ion spectrum 112.

[0034] Composed of at least one hardware processor (e.g., Figure 17 Hardware processor 1702 and / or Figure 19 The ion score generator 118, executed by the hardware processor 1904, can determine the ion score 120 based on the analysis of the identified characteristics 116 of the expected ion spectrum 106 and the observed ion spectrum 112.

[0035] Composed of at least one hardware processor (e.g., Figure 17 Hardware processor 1702 and / or Figure 19 The ion scoring analyzer 122, executed by the hardware processor 1904, can identify the highest ion score 124 from the ion score 120. The ion scoring analyzer 122 can identify the expected ion spectrum 126 corresponding to the highest ion score 124 based on the highest ion score 124.

[0036] Composed of at least one hardware processor (e.g., Figure 17 Hardware processor 1702 and / or Figure 19 The visual results generator 128, executed by the hardware processor 1904, can generate an indication 130 and / or a graphical user interface display 132 of the expected ion spectrum 126 corresponding to the highest ion score 124, based on the identification of the expected ion spectrum 126 corresponding to the highest ion score 124.

[0037] Ion score analyzer 122 can compare a highest ion score 124 with at least one threshold range 134 (e.g., a total threshold range of 0-15, which may include intermediate threshold ranges such as: greater than 0 to less than 5 for a “weak” match, greater than or equal to 5 and less than 11 for a “good” match, and greater than or equal to 11 and less than or equal to 15 for a “excellent” match). Ion score analyzer 122 can determine the match type 136 (e.g., “weak,” “good,” or “excellent” as disclosed herein) between the expected ion spectrum 126 corresponding to the highest ion score 124 and the observed ion spectrum 112 based on the comparison of the highest ion score 124 with at least one threshold range 134. Further, visual result generator 128 can generate another indication 138 or another graphical user interface display 140 for the determination of the match type 136 based on the determination of the match type 136 between the expected ion spectrum 126 corresponding to the highest ion score 124 and the observed ion spectrum 112.

[0038] Based on the examples disclosed herein, the identified characteristics 116 of the expected ion spectrum 106 may include the expected ion centroid mass / charge number (m / z) and intensity value, the expected profile signal, the m / z parts per million (ppm) error, the expected centroid standard deviation, and the noise threshold.

[0039] According to the examples disclosed herein, the identified characteristics 116 of the observed ion spectrum 112 may include at least one observed centroid value, observed profile signal, and ion mass / charge number (m / z) parts per million (ppm) error.

[0040] The ion score generator 118 can determine the ion score 120 by analyzing the identified characteristics 116 of the expected ion spectrum 106 and the observed ion spectrum 112, by defining the ion score 120 as a function of the chi-square p-value, Spearman correlation, Pearson p-value, and noise probability.

[0041] According to the examples disclosed herein, the ion score generator 118 can determine the ion score 120 as a function of the chi-square p-value by determining the probability of randomly observing a set of m / z and intensity values ​​that match the expected ion centroid mass / charge number (m / z) parts per million (ppm) error and intensity standard deviation.

[0042] Based on the examples disclosed herein, the ion score generator 118 can determine the ion score 120 as a function of Spearman correlation by determining the rank-based correlation coefficient between a specific expected ion spectrum and the observed ion spectrum in the expected ion spectrum.

[0043] According to the examples disclosed herein, the ion score generator 118 can determine the ion score 120 as a function of the Pearson p-value by determining the probability of randomly observing a set of ion centroid mass / charge number (m / z) and intensity values ​​of a particular expected ion spectrum that is linearly correlated with the expected value at a higher Pearson correlation coefficient for a specific expected ion spectrum in the expected ion spectrum.

[0044] Based on the examples disclosed herein, the ion score generator 118 can determine the ion score 120 as a function of the noise probability by determining the peak density as a function of the intensity rank between a specific expected ion spectrum and the observed ion spectrum in the expected ion spectrum.

[0045] Reference Figures 2A-16 The operation of device 100 is described in more detail.

[0046] Figures 2A-2C The partitioning of visual sensitivity is shown to illustrate the operation of device 100 according to an example of this disclosure.

[0047] refer to Figures 2A-2C Regarding the zoning of visual sensitivity, ion matching can be determined based on factors such as the alignment between the expected and observed peak profiles and factors that confuse signals below a subjective threshold. In this regard, the quality of peak matching can be based on subjective metrics. For device 100, each of these various metrics can be zoned and placed in clearly defined terms, and the levels can be adjusted until they are consistent with the consensus of expert annotations.

[0048] Figures 2A-2CThe matching between the expected ions and the data observed in a liquid chromatography-mass spectrometry (LC / MS) system (such as Agilent QTOF) equipped with an electron-based dissociation (ExD) cell, directly injected from carbonic anhydrase, is shown. “Excellent,” “Good,” and “Weak” matches are shown at 200, 202, and 204, respectively. Crosses (some indicated at 206, 208, and 210) can represent the expected m / z and intensity of each isotopic peak. The width of the boxes (some indicated, for example, at 212, 214, and 216) can represent the parts per million (ppm) error of the m / z around each peak, which can generally be well-defined. The height of each box (e.g., at 212, 214, and 216) can correspond to an estimate of the peak intensity standard deviation, which is a function of the expected centroid intensity and a subjective limit that can correspond to how much deviation is generally allowed between the observed intensity value and the expected intensity value. For example, the dashed lines at 218, 220, and 222 can correspond to an estimate of the noise threshold, which can represent the intensity cutoff of a function defined as m / z. Peaks below this threshold can be ignored as obfuscation. Obfuscation signals can be visually measured by examining the signal within the m / z range of the box (shown at 224, 226, and 228), which lies between the peak regions in the box, for example, at 212, 214, and 216. A large number of non-zero signals can degrade the quality of ion matching.

[0049] Figure 3 The definition of an ion scoring binary classifier is shown to illustrate the operation of device 100 according to examples in this disclosure.

[0050] refer to Figure 3 For inputs including the expected centroid m / z and intensity value (e.g., cross, such as a cross at 300), observed centroid value (e.g., line, such as a line at 302), expected profile signal (e.g., a dashed line at 304), observed profile signal (e.g., at 306), m / z ppm error, expected centroid standard deviation, and noise threshold, the output can include an ion score of 0 → 15.

[0051] about Figure 3 The ion score can be determined as follows: Ion score = (1.0 - Chi-square p-value) * Spearman correlation * log((1.0 - Pearson p-value) / noise probability) Equation (1) For equation (1), the chi-square p-value can represent the probability of randomly observing a set of centroid m / z and intensity values ​​that match the expected m / z ppm error and intensity standard deviation. The Spearman correlation can represent the rank-based correlation coefficient between the observed profile signal and the expected profile signal. The Pearson p-value can represent the probability of randomly observing a set of centroid m / z and intensity values ​​that are linearly correlated with the expected values ​​at a higher Pearson correlation coefficient. Furthermore, regarding the noise probability, assuming the independence of the centroids, regarding the probability of observing a peak with higher intensity, the noise probability can capture the peak density as a function of intensity rank in a broader spectrum.

[0052] Figures 4A-4C A consistent weighting scheme is applied to all classifier metrics to illustrate the operation of device 100 according to the examples in this disclosure.

[0053] refer to Figures 4A-4C The results show “excellent,” “good,” and “weak” matches at 400, 402, and 404, respectively. Regarding the application of uniform weighting, these measures can have greater applicability with respect to the most abundant isotope peaks. Less abundant isotope peaks may not be visible at all and may have a larger standard deviation. A straight line can be plotted between each expected centroid peak (the top of the polygon at 406). This line limits the weight of each measure as a function of m / z, which is applied... Figure 3 All metrics defined in the text. For example, a weighted chi-square p-value can be used when each weight is defined as the expected centroid strength. For 400... Figure 4A In the drawing, regions within polygonal shapes may not have overlapping signals, except in... Figure 4A The last triangle with the minimum weight on the right side of the orientation. For the triangle at 400... Figure 4A The plot shows that the expected isotopic ratios perfectly match the observed centroid. For the point at 402... Figure 4B The plot shows that the highest-weighted polygon at 408 has a sparse overlapping signal. The expected isotopic ratio matches the observed centroid perfectly, but only for the most abundant peak. For 404... Figure 4C The plot shows that the finite number of highest polygons at 410 have an acceptable overlap signal. The expected isotopic ratios also loosely match the observed centroids in those regions.

[0054] Figure 5 An exhaustive search for hydrogen transfer is demonstrated to illustrate the operation of device 100 according to the example in this disclosure.

[0055] refer to Figure 5An exhaustive search for hydrogen transfer allows each ion to be matched with any combination of + / -1 or + / -2 hydrogen transfers. In this regard, three possibilities exist. For the first two possibilities, since... Linear algebra can be used to determine the coefficients that minimize the error between the observed and expected centroid values. This applies to single hydrogen transfer events (e.g., Applying this to peak distribution can increase or decrease the mass (H+) / charge per m / z value. Regarding the desired isotopic peak distribution options, the option with the highest score can be used from the following: (1)

[0056] (2)

[0057] (3)

[0058] refer to Figure 5 ,for , and As shown at positions 500, 502, and 504, 51% H, 15% 2H, and 33% neutral content provide the best possible score for this match. The plot at position 500 can represent... The drawing at position 502 can represent The drawing at position 504 can represent .

[0059] Figure 6 Annotated ions are shown to illustrate the operation of device 100 according to examples in this disclosure.

[0060] refer to Figure 6At 600, a set of centroids above the noise threshold is displayed, and at 602, one or more annotated ions are displayed. In this regard, at 604, the ion spectrum analyzer 102 can perform matching with the isotopic distribution of known ions. At 606, the ion score generator 118 can maximize the ion score over all possible hydrogen transfers. At 608, the ion spectrum analyzer 102 can perform matching with the average isotopic distribution (unknown). At 610, the ion score generator 118 can determine the ion score. At 612, the ion score analyzer 122 can sort all ions by decreasing score. Further, at 614, if an ion does not contain a previously assigned centroid, the ion score analyzer 122 can iteratively select the ion with the highest score and add it to the output.

[0061] Figure 7A and Figure 7B The settings of signal processing parameters are shown to illustrate the operation of device 100 according to the example in this disclosure.

[0062] refer to Figure 7A and Figure 7B To set signal processing parameters, data (such as orbital trap profile data) can be cleansed using baseline and noise removal. In this regard, depending on the analyte, TOF data can include stray and overlapping signals. As shown at 700, short peptides can produce relatively simple spectra that can be annotated by matching centroids. Associated TOF data for a short peptide at 700 is shown at 702. Top-down fragmentation of large molecules (e.g., antibodies) may require tuning deconvolution parameters. Scoring preferences, signal-to-noise ratio thresholds, window widths, etc., can be specified based on the desired number of isotopic peaks and the presence of overlapping isotopic distributions.

[0063] Figure 8A and Figure 8B Benchmark tests for ion scoring are shown to illustrate the operation of device 100 according to examples in this disclosure.

[0064] refer to Figure 8A and Figure 8B Regarding benchmarking, to test sensitivity and accuracy, two real-world datasets can be used, each considering any ion with at least one isotopic peak found above a noise threshold. Based on validation of a set of putative ions from ubiquitine (analyzed by QTOF) (e.g., by multiple validators in the art), annotations with high accuracy regarding ions of specified m / z and charge are utilized. Figure 8A Results for multiple validators are shown at point 800, while Figure 8BResults for a single validator are shown at 802. At 90% sensitivity, for device 100, >99% of ions were correctly labeled by m / z and charge (scoring threshold = 1.21). At 85% sensitivity, for device 100, >99% of ions were correctly labeled by m / z and charge (scoring threshold = 1.15). Ions with scores close to the threshold (e.g., score = 2.0) can exhibit “weak” but “acceptable” matching quality, indicating that the aforementioned ion scoring technique is being performed accurately.

[0065] Figure 9 The complete NIST mAb is shown to illustrate the operation of device 100 according to the example in this disclosure.

[0066] refer to Figure 9 As shown at 900, the most abundant ions can be identified using a visual sensitivity classifier, where their contour signals are subtracted and further iterated. For the observed ion spectrum shown at 900, the results of ions identified by device 100 are shown at 902, 904, 906, 908, and 910. The associated region of interest is shown at 912. This paper discusses... Figure 5 Equations for calculating scores are disclosed. Visual acuity classifiers can operate by combining observed evidence from the raw profile signal with processed centroid peaks to test the assumption that the alignment between the observed and expected data is due to random chance. For example, Spearman correlation measures the deviation between the observed profile signal (e.g., the wavy line at 914) and the expected profile signal. The expected profile signal can be generated by first estimating the instrument resolution (e.g., peak width) by fitting a Gaussian curve to the profile signal around each centroid, measuring the peak width, and averaging this value over all observed centroids. Then, for each isotope distribution, the Gaussian curve can be scaled to each expected centroid m / z and intensity data point and summed across all expected isotope peaks, where the intensity is zero elsewhere. The remaining p-value measures the deviation between the observed centroid peak (the vertical line at 918) and the expected centroid peak (the cross at 920), as well as the amount of unexpected confounding centroid peaks within the isotope peak region. All these calculations are as described in this paper regarding... Figure 6 The data is weighted according to the publicly available data, so that evidence closer to the most abundant isotopes is more trustworthy than evidence from less abundant isotopes, where the classifier can be used for less abundant and partially overlapping isotope peak clusters.

[0067] Figure 10 A top-down deconvolution benchmark is shown to illustrate the operation of device 100 as an example according to this disclosure.

[0068] refer to Figure 10 The top-down deconvolution benchmark is shown at 1000, with 88% sequence coverage shown at 1002.

[0069] Figure 11 User settings preferences are shown to illustrate the operation of device 100 according to examples in this disclosure.

[0070] refer to Figure 11 User settings preferences are shown at 1100. In this regard, target information, m / z tolerance (ppm), minimum ion matching confidence, fragmentation, and iterative matching preferences are entered and modified as shown. Matching settings 1104 based on the user settings preferences at 1100 can be locked as shown at 1102. In this regard, matching settings such as mass tolerance and ion type can be set automatically (e.g., without human intervention).

[0071] Figure 12 Parameterless deconvolution is demonstrated to illustrate the operation of device 100 according to the example in this disclosure.

[0072] refer to Figure 12 Regarding the parameter-free deconvolution as shown at 1200, to analyze accuracy on larger datasets, based on the execution of ExDViewer with default settings (e.g., software providing a comprehensive overview of observed fragment ions matching a given protein sequence), all ions with scores of 1.5 or higher can be considered after three rounds of scoring to detect overlapping isotope distributions. For example, regarding more than six peptides (e.g., at 1202, 1204, etc.) and two intact proteins (e.g., at 1206 and 1208, both from direct injection and liquid chromatography separation) from multiple instruments, Device 100 achieved >99% m / z and charge accuracy from the resulting ion predictions. Based on the evaluation of >10,000 putative ions, Device 100 reliably identified constellations of satellite ion types through robust likelihood scores.

[0073] Figure 13 Deconvolution and decharge in ExDViewer are demonstrated to illustrate the operation of device 100 according to the example in this disclosure.

[0074] refer to Figure 13 Regarding deconvolution and decharge in ExDViewer, the original unlabeled spectrum is shown at 1300, the decharged (deconvolution) spectrum is shown at 1302, the labeled spectrum is shown at 1304, and the deisotope spectrum is shown at 1306.

[0075] Figure 14Hybrid targeted / non-targeted deconvolution is demonstrated to illustrate the operation of device 100 according to the examples in this disclosure.

[0076] refer to Figure 14 Regarding the mixed targeted / untargeted deconvolution as shown at 1400, for precursors with a charge of 20+, the ppm tolerance is greater than 1 / charge, each ppm tolerance box can contain at least one centroid, and the profile data can provide reliable identification. Y ions with a charge of 4+ can be matched to the exact isotopic distribution generated from the targeted search. Unassigned precursor products with a charge of 20+ can be assigned to the average isotopic distribution.

[0077] Figure 15 Ion identification in MS / MS spectra under ECD conditions and monitoring of ion charge distribution and efficiency are shown to illustrate the operation of device 100 according to the example of this disclosure.

[0078] refer to Figure 15 Regarding ion identification and monitoring of ion charge distribution and efficiency in MS / MS spectra under ECD conditions, tuning can be performed at 1500 to obtain full c / z ion coverage, where collision energies are added (e.g., CID+ECD). Tuning can be performed at 1502 to obtain additional efficiency and produce different charge state distributions and side-chain losses (ECD only).

[0079] Figure 16 Ion identification is demonstrated to illustrate the operation of device 100 according to an example of this disclosure.

[0080] refer to Figure 16 The device 100 can achieve, for example, >99% accuracy and >80% sensitivity on complex, complete mAb MS1 data. In this regard, based on spectrum reception, the device 100 can accurately identify a relatively high number of ions within fractions of a second.

[0081] Figures 17-19 The following diagrams illustrate example block diagram 1700, flowchart of example method 1800, and another example block diagram 1900 for MS and MS / MS spectral deconvolution based on visual acuity interpretation. Block diagrams 1700, 1800, and 1900 are shown above by way of example, not limitation. Figure 1 The described apparatus 100 is implemented. Block diagram 1700, method 1800, and block diagram 1900 can be implemented in other devices. In addition to block diagram 1700, Figure 17Hardware for device 100 that can execute the instructions of block diagram 1700 is shown. The hardware may include a processor 1702 and a memory 1704 storing machine-readable instructions that, when executed by the processor, cause the processor to execute the instructions of block diagram 1700. Memory 1704 may represent a non-transitory computer-readable medium. Figure 18 This can represent an example method for deconvolution of MS and MS / MS spectral maps based on visual acuity interpretation, as well as the steps of the method. Figure 19 A non-transitory computer-readable medium 1902, as exemplified by the example, stores machine-readable instructions for providing MS and MS / MS spectral deconvolution based on visual acuity interpretation. When executed, these machine-readable instructions cause processor 1904 to perform... Figure 19 The instructions in block diagram 1900 are shown in the figure.

[0082] Figure 17 Processor 1702 and / or Figure 19 The processor 1904 may include one or more processors or other hardware processing circuitry to perform the methods, functions, and other processes described herein. These methods, functions, and other processes may be embodied as machine-readable instructions stored on a computer-readable medium, which may be non-transitory (e.g., Figure 19 Non-transitory computer-readable media 1902, such as hardware storage devices (e.g., RAM (random access memory), ROM (read-only memory), EPROM (erasable programmable ROM), EEPROM (electrically erasable programmable ROM), hard disk drives, and flash memory). Memory 1704 may include RAM, in which machine-readable instructions and data for the processor can reside during operation.

[0083] refer to Figures 1-17 And especially refer to Figure 17 As shown in block diagram 1700, memory 1704 may include instructions 1706 to receive a plurality of expected ion spectra 106 for the ion 104 to be identified. Each expected ion spectrum 108 in the expected ion spectra 106 may include at least one expected peak profile 110.

[0084] The processor 1702 can acquire, decode, and execute instructions 1708 to receive observed ion spectra 112 for the ion to be identified 104, the observed ion spectra including at least one observed peak profile 114.

[0085] The processor 1702 can acquire, decode, and execute instructions 1710 to identify characteristics 116 of the expected ion spectrum 106 and the observed ion spectrum 112 based on analysis of at least one expected peak profile 110 of each expected ion spectrum 106 and at least one observed peak profile of the observed ion spectrum 112.

[0086] Processor 1702 can acquire, decode, and execute instructions 1712 to determine ion scores 120 based on analysis of identified characteristics 116 of the expected ion spectrum 106 and the observed ion spectrum 112.

[0087] Processor 1702 can extract, decode and execute instruction 1714 to identify the highest ion score 124 from ion score 120.

[0088] Processor 1702 can acquire, decode and execute instructions 1716 to identify the expected ion spectrum 126 corresponding to the highest ion score 124 based on the highest ion score 124.

[0089] The processor 1702 can acquire, decode, and execute instructions 1718 to generate an indication 130 and / or a graphical user interface display 132 of the expected ion spectrum 126 corresponding to the highest ion score 124, based on the identification of the expected ion spectrum 126 corresponding to the highest ion score 124.

[0090] refer to Figures 1-16 and Figure 18 And especially refer to Figure 18 For method 1800, at block 1802, the method may include receiving a desired ion spectrum for the ion to be identified 104, the desired ion spectrum including at least one desired peak profile.

[0091] At block 1804, method 1800 may include receiving an observed ion spectrum 112 for the ion 104 to be identified, the observed ion spectrum including at least one observed peak profile 114.

[0092] At box 1806, method 1800 may include identifying the characteristics of the expected ion spectrum and the observed ion spectrum 112 based on analysis of at least one expected peak profile of the expected ion spectrum and at least one observed peak profile of the observed ion spectrum 112.

[0093] At box 1808, method 1800 may include determining an ion score based on analysis of the identified characteristics of the expected ion spectrum and the observed ion spectrum.

[0094] Based on the examples disclosed herein, method 1800 may further include generating at least one of an indication or graphical user interface display of the expected ion spectrum and ion score.

[0095] Based on the examples disclosed herein, method 1800 may further include determining the matching type between the expected ion spectrum and the observed ion spectrum based on a comparison of the ion score with at least one threshold range.

[0096] refer to Figures 1-16 and Figure 19 And especially refer to Figure 19 For block diagram 1900, non-transitory computer-readable medium 1902 may include instructions 1906 to receive user setting preferences based on input at a graphical user interface display.

[0097] The processor 1904 can acquire, decode, and execute instructions 1908 to receive a plurality of expected ion spectra 106 for the ion to be identified 104. Each expected ion spectrum 108 in the expected ion spectra 106 may include at least one expected peak profile 110.

[0098] The processor 1904 can acquire, decode, and execute instructions 1910 to receive observed ion spectra 112 for the ion to be identified 104, the observed ion spectra including at least one observed peak profile 114.

[0099] The processor 1904 can acquire, decode, and execute instructions 1912 to determine an ion score 120 based on received user setting preferences and analysis of at least one expected peak profile 110 of each expected ion spectrum 108 in the expected ion spectrum 106 and at least one observed peak profile of the observed ion spectrum 112.

[0100] Processor 1904 can extract, decode and execute instruction 1914 to identify the highest ion score 124 from ion score 120.

[0101] The processor 1904 can acquire, decode, and execute instructions 1916 to identify the expected ion spectrum 126 corresponding to the highest ion score 124 based on the highest ion score 124.

[0102] The examples described and illustrated herein are examples and some variations thereof. The terminology, descriptions, and figures used herein are set forth by way of illustration only and are not intended to be limiting. Many variations are possible within the spirit and scope of the subject matter and are intended to be defined by the appended claims and their equivalents, wherein, unless otherwise stated, all terms shall be understood in their broadest reasonable sense.

Claims

1. An apparatus, the apparatus comprising: At least one hardware processor; An ion spectrum analyzer, the ion spectrum analyzer being executed by the at least one hardware processor to: For the ion to be identified, multiple expected ion spectra are received, each of the expected ion spectra including at least one expected peak profile. For the ion to be identified, an observed ion spectrum is received, the observed ion spectrum including at least one observed peak profile; and The characteristics of the expected ion spectra and the observed ion spectra are identified based on the analysis of at least one expected peak profile of each expected ion spectrum and at least one observed peak profile of the observed ion spectrum. An ion score generator, the ion score generator being executed by the at least one hardware processor to: Ion scores are determined based on the analysis of the expected ion spectra and the identified characteristics of the observed ion spectra. An ion scoring analyzer, the ion scoring analyzer being executed by the at least one hardware processor to: The highest ion score was identified from the ion scores; and Based on the highest ion score, identify the expected ion spectrum corresponding to the highest ion score; and A visual results generator, which is executed by the at least one hardware processor to: Based on the identification of the expected ion spectrum corresponding to the highest ion score, at least one of the following is generated: an indication or a graphical user interface display of the expected ion spectrum corresponding to the highest ion score.

2. The device according to claim 1, wherein: The ion scoring analyzer is executed by the at least one hardware processor to: The highest ion score is compared to at least one threshold range; and Based on the comparison between the highest ion score and the at least one threshold range, the matching type between the expected ion spectrum corresponding to the highest ion score and the observed ion spectrum is determined; and The visual result generator is executed by the at least one hardware processor to: Based on the determination of the matching type between the expected ion spectrum corresponding to the highest ion score and the observed ion spectrum, at least one of another indication or another graphical user interface display for the determination of the matching type is generated.

3. The device of claim 1, wherein the identified characteristics of the anticipated ion spectrum include the anticipated ion centroid mass / charge number (m / z) and intensity value, the anticipated profile signal, the m / z parts per million (ppm) error, the anticipated centroid standard deviation, and the noise threshold.

4. The device of claim 1, wherein the identified characteristics of the observed ion spectrum include at least one observed centroid value, an observed profile signal, and an ion mass / charge number (m / z) parts per million (ppm) error.

5. The device of claim 1, wherein the ion score generator is executed by the at least one hardware processor to determine the ion score based on the analysis of the identified characteristics of the expected ion spectrum and the observed ion spectrum in the following manner: The ion score is determined as a function of the chi-square p-value, Spearman correlation, Pearson p-value, and noise probability.

6. The device of claim 5, wherein the ion score generator is executed by the at least one hardware processor to determine the ion score as a function of the chi-square p-value by determining the probability of randomly observing a set of m / z and intensity values ​​that match the expected ion centroid mass / charge number (m / z) parts per million (ppm) error and intensity standard deviation.

7. The device of claim 5, wherein the ion score generator is executed by the at least one hardware processor to determine the ion score as a function of the Spearman correlation by determining a rank-based correlation coefficient between a specific expected ion spectrum in the expected ion spectrum and the observed ion spectrum.

8. The device of claim 5, wherein the ion score generator is executed by the at least one hardware processor to determine the ion score as a function of the Pearson p-value by determining the probability of randomly observing a set of ion centroid mass / charge number (m / z) and intensity values ​​of a particular expected ion spectrum that is linearly correlated with the expected value to a higher Pearson correlation coefficient for a specific expected ion spectrum in the expected ion spectrum.

9. The device of claim 5, wherein the ion score generator is executed by the at least one hardware processor to determine the ion score as a function of the noise probability by determining the peak density as a function of intensity rank between a specific expected ion spectrum in the expected ion spectrum and the observed ion spectrum.

10. A method for deconvolution of mass spectrometry (MS) and tandem mass spectrometry (MS / MS) spectra based on visual acuity interpretation, the method comprising: For the ion to be identified, a expected ion spectrum is received by at least a hardware processor, the expected ion spectrum including at least one expected peak profile. For the ion to be identified, the at least hardware processor receives an observed ion spectrum, the observed ion spectrum including at least one observed peak profile; The characteristics of the expected ion spectrum and the observed ion spectrum are identified by the at least hardware processor based on the analysis of at least one expected peak profile of the expected ion spectrum and at least one observed peak profile of the observed ion spectrum. as well as The ion score is determined by the hardware processor based on the analysis of the expected ion spectrum and the identified characteristics of the observed ion spectrum.

11. The method of claim 10, further comprising: The expected ion spectrum and an indication or graphical user interface display of the ion score are generated by the at least hardware processor.

12. The method of claim 10, further comprising: The matching type between the expected ion spectrum and the observed ion spectrum is determined by the at least hardware processor based on a comparison of the ion score with at least one threshold range.

13. The method of claim 10, wherein determining the ion score by the at least hardware processor based on the analysis of the identified characteristics of the anticipated ion spectrum and the observed ion spectrum further comprises: The ion score is determined by the at least hardware processor as a function of the chi-square p-value, Spearman correlation, Pearson p-value, and noise probability.

14. A non-transitory computer-readable medium storing machine-readable instructions that, when executed by at least one hardware processor, cause the at least one hardware processor to: User preferences are received based on input displayed on the graphical user interface. For the ion to be identified, multiple expected ion spectra are received, each of the expected ion spectra including at least one expected peak profile. For the ion to be identified, an observed ion spectrum is received, the observed ion spectrum including at least one observed peak profile; Ion scores are determined based on received user setting preferences and analysis of at least one expected peak profile of each expected ion spectrum and at least one observed peak profile of the observed ion spectrum. The highest ion score was identified from the ion scores; and Based on the highest ion score, the expected ion spectrum corresponding to the highest ion score is identified.

15. The non-transitory computer-readable medium of claim 14, wherein the machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to: Based on the identification of the expected ion spectrum corresponding to the highest ion score, at least one of the following is generated: an indication of generating the expected ion spectrum corresponding to the highest ion score or another graphical user interface display.

16. The non-transitory computer-readable medium of claim 14, wherein the machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to: The highest ion score is compared to at least one threshold range; and Based on the comparison between the highest ion score and the at least one threshold range, the matching type between the expected ion spectrum corresponding to the highest ion score and the observed ion spectrum is determined.

17. The non-transitory computer-readable medium of claim 16, wherein the machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to: Based on the determination of the matching type between the expected ion spectrum corresponding to the highest ion score and the observed ion spectrum, at least one of an indication of the determination of the matching type or another graphical user interface display is generated.

18. The non-transitory computer-readable medium of claim 14, wherein the machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to: Identify the characteristics of the expected ion spectrum, including the expected ion centroid mass / charge number (m / z) and intensity value, the expected profile signal, m / z parts per million (ppm) error, the expected centroid standard deviation, and the noise threshold.

19. The non-transitory computer-readable medium of claim 14, wherein the machine-readable instructions, when executed by the at least one hardware processor, further cause the at least one hardware processor to: Identify the characteristics of the observed ion spectrum, including at least one observed centroid value, observed profile signal, and ion mass / charge number (m / z) parts per million (ppm) error.

20. The non-transitory computer-readable medium of claim 14, wherein the machine-readable instructions for determining the ion score based on the analysis of the at least one expected peak profile of each expected ion spectrum in the expected ion spectra and the observed ion spectra, when executed by the at least one hardware processor, further cause the at least one hardware processor to: The ion score is determined as a function of the chi-square p-value, Spearman correlation, Pearson p-value, and noise probability.

Images