Mass spectroscopy apparatus and method

Abstract

To prevent fine mass information from being greatly damaged when creating a data set for multivariable analysis reduced or compressed based on a plurality of mass spectra.SOLUTION: A KMD analyzer 28 generates n KMD analysis results based on n mass spectra. An integrator 34 generates n integrated value distributions based on the n KMD analysis results. A principal component analyzer 42 applies principal component analysis to the n integrated value distributions (data sets). A component-reflected KMD plot is generated based on a loading map (component distribution) generated by the principal component analysis.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a mass spectrometry processing apparatus and method, and particularly to a technique for mutually comparing a plurality of mass spectra obtained from a plurality of samples.

Background Art

[0002] A mass spectrometry system generally consists of a mass spectrometer and an information processing device. The mass spectrometer performs mass spectrometry on a sample. The information processing device applies various processes to the mass spectrum obtained by mass spectrometry. The information processing device can be referred to as a mass spectrometry processing device.

[0003] A mass spectrometry system is used in the analysis of polymers. As an analysis method for a mass spectrum obtained from a polymer, Kendrick mass defect (KMD) analysis is known (see Patent Document 1).

[0004] As methods for statistically analyzing a data set consisting of a plurality of data, principal component analysis (PCA: Principal Component Analysis), vertex component analysis (VCA: Vertex Component Analysis) (see Non-Patent Document 1), hierarchical cluster analysis, etc. are known. All of these methods are multivariable analysis methods. In addition, Patent Document 2 discloses the UMAP (Uniform Manifold Approximation and Projection) method, which is a dimensionality reduction method, as a mass spectrum analysis method. In this specification, following general Japanese notation, two terms, analysis and parsing, are used, but they have the same meaning.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

[0006] [Non-Patent Document 1] Jose MP Nascimento, Vertex Component Analysis: A Fast Algorithm to Unmix Hyperspectral Data, IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 43, NO. 4, APRIL 2005 [Overview of the project] [Problems that the invention aims to solve]

[0007] In comparative or difference analyses involving multiple samples, multiple mass spectra obtained from multiple samples are compared with each other, or differences between multiple mass spectra are identified. Therefore, it is conceivable to apply multivariate analysis to a dataset consisting of multiple mass spectra (mass spectral data).

[0008] If multiple mass spectra are acquired using a mass spectrometer with very high mass spectrometry accuracy, each mass spectrum will consist of, for example, 500,000 ion intensities corresponding to 500,000 m / z values. Constructing a dataset using such multiple mass spectra is impractical because the computational load for multivariate analysis based on that dataset would be enormous.

[0009] Therefore, it is conceivable to reduce or compress multiple mass spectra and then construct a dataset for multivariate analysis from these reduced or compressed mass spectra. Specifically, first, multiple integration intervals are set on the m / z axis. Next, for each mass spectrum, multiple ion intensities belonging to each integration interval are integrated. A reduced dataset for multivariate analysis is constructed from multiple series of integrated values ​​calculated from multiple series of ion intensities.

[0010] However, while increasing the integration interval size increases the reduction rate (compression degree) and thus the computational load in multivariate analysis, it also increases the loss of decimal mass information (decimal values ​​of each mass). For example, if the processing range from 500 to 4000 on the m / z axis is divided into 5000 integration intervals, the size of each integration interval becomes 0.7 (equivalent to 0.7u). Mass information smaller than 0.7 is lost during the integration process. Even if a mass spectrometer with very high mass resolution is used, the full benefits of this resolution cannot be realized.

[0011] The object of the present invention is to ensure that detailed mass information is not significantly lost when creating a dataset for multivariate analysis that has been reduced or compressed based on multiple mass spectra. Alternatively, the object of the present invention is to utilize the mass resolution function of KMD analysis when creating a dataset for multivariate analysis. Alternatively, the object of the present invention is to provide a mechanism that reflects a second analysis result with low mass resolution in relation to a first analysis result with high mass resolution. [Means for solving the problem]

[0012] The mass spectrum processing apparatus according to the present invention is characterized by comprising: a KMD analyzer that applies Kendrick mass defect (KMD) analysis to n (where n is an integer of 2 or more) mass spectra obtained by mass spectrometry of a plurality of samples, thereby generating n KMD analysis results; an integrator that applies integration processing according to a plurality of integration intervals on the nominal Kendrick mass (NKM) axis and a plurality of integration intervals on the KMD axis to the n KMD analysis results, thereby generating n integrated value distributions as n reduced KMD analysis results; and a multivariate analyzer that applies multivariate analysis to a dataset constructed based on the n integrated value distributions.

[0013] The mass spectrum processing method according to the present invention is characterized by comprising the steps of: applying Kendrick mass defect (KMD) analysis to n (where n is an integer of 2 or more) mass spectra obtained by mass spectrometry of a plurality of samples, thereby generating n KMD analysis results; applying integration processing to the n KMD analysis results according to a plurality of integration intervals on the nominal Kendrick mass (NKM) axis and a plurality of integration intervals on the KMD axis, thereby generating n integrated value distributions as n reduced KMD analysis results; and applying multivariate analysis to a dataset constructed based on the n integrated value distributions. [Effects of the Invention]

[0014] According to the present invention, when creating a dataset for multivariate analysis that has been reduced or compressed based on multiple mass spectra, it is possible to prevent significant loss of detailed mass information. Alternatively, according to the present invention, when creating a dataset for multivariate analysis, the mass resolution function provided by KMD analysis can be utilized. Alternatively, according to the present invention, a mechanism can be provided that reflects a second analysis result having a lower mass resolution in relation to a first analysis result having a higher mass resolution. [Brief explanation of the drawing]

[0015] [Figure 1]It is a block diagram showing a configuration example of a mass spectrometry system according to an embodiment. [Figure 2] It is a diagram showing a plurality of mass spectra obtained from a plurality of samples. [Figure 3] It is a diagram showing a peak list including KMD analysis results. [Figure 4] It is a diagram showing a KMD plot. [Figure 5] It is a diagram showing a multiple KMD plot. [Figure 6] It is a diagram showing a matrix for integration processing. [Figure 7] It is a diagram showing an integrated value distribution (reduced KMD plot). [Figure 8] It is a diagram showing a dataset. [Figure 9] It is a diagram showing two score plots. [Figure 10] It is a diagram showing a loading map corresponding to the first principal component. [Figure 11] It is a diagram showing a loading map corresponding to the second principal component. [Figure 12] It is a diagram showing a loading map corresponding to the third principal component. [Figure 13] It is a diagram showing a positive emphasis section and a negative emphasis section. [Figure 14] It is a diagram showing a first example of a component reflection KMD plot. [Figure 15] It is a diagram showing a second example of a component reflection KMD plot. [Figure 16] It is a diagram showing a third example of a component reflection KMD plot. [Figure 17] It is a block diagram showing another configuration example of an information processing apparatus according to an embodiment. [Figure 18] It is a diagram showing a component map (component spectrum map) generated by vertex component analysis. [Figure 19] It is a diagram showing a component map (phase map) generated by cluster analysis. [Figure 20] It is a flowchart showing a mass spectrum processing method according to an embodiment. [Figure 21] This figure shows the analysis region set up on a two-dimensional coordinate system. [Modes for carrying out the invention]

[0016] The embodiments will be described below with reference to the drawings.

[0017] (1) Outline of the Embodiment The mass spectrum processing apparatus according to this embodiment includes a KMD analyzer, an integrator, and a multivariate analyzer. The KMD analyzer applies Kendrick Mass Defect (KMD) analysis to n mass spectra (where n is an integer of 2 or more) obtained by mass spectrometry of multiple samples, thereby generating n KMD analysis results. The integrator applies integration processing to the n KMD analysis results according to multiple integration intervals on the nominal Kendrick Mass (NKM) axis and multiple integration intervals on the KMD axis, thereby generating n integrated value distributions as n reduced KMD analysis results. The multivariate analyzer applies multivariate analysis to a dataset constructed based on the n integrated value distributions. The processor described later functions as the KMD analyzer, integrator, and multivariate analyzer.

[0018] KMD analysis has a mass resolution function. That is, in KMD analysis, each m / z, i.e., each mass on the m / z axis is converted into integer mass (NKM) and fractional mass (KMD). Even if multiple integration intervals are set on the NKM axis, the fractional mass information is not lost because the NKM axis represents integer mass. On the other hand, if multiple integration intervals are set on the KMD axis, some fractional mass information is lost, but the fractional mass information as a whole is not lost. The decrease in mass resolution is limited.

[0019] With the above configuration, even when a reduced or compressed dataset is created, some fractional mass information is preserved. Therefore, it becomes possible to utilize the preserved fractional mass information after multivariate analysis. Examples of multivariate analysis include principal component analysis (PCA), vertex component analysis (VCA), and hierarchical cluster analysis. The above n is generally an integer of 3 or greater.

[0020] The mass spectral processing apparatus according to this embodiment includes a generator that generates at least one plot of both analysis results based on at least one of n KMD analysis results and the results of a multivariate analysis. A processor, described later, functions as the generator.

[0021] Both analysis result plots reflect both the KMD analysis results and the multivariate analysis results. By observing or evaluating both analysis result plots, it is possible to identify differences between multiple samples and the characteristics of each sample. In the embodiment, the KMD analysis result is an ion intensity distribution with a first mass resolution, and the multivariate analysis result is a feature quantity distribution (component distribution) with a second mass resolution lower than the first mass resolution.

[0022] In the embodiment, the n KMD analysis results are ion intensity distributions on a two-dimensional coordinate system defined by the NKM axis and the KMD axis, respectively. The multivariate analysis is a component analysis. The results of the multivariate analysis include component distributions on the same two-dimensional coordinate system as described above. The generator generates at least one double analysis result plot based on at least one ion intensity distribution and component distribution. The at least one double analysis result plot is a component-reflecting KMD plot having emphasized portions identified based on the component distribution.

[0023] For example, in a component-reflecting KMD plot, the highlighted portion is the area corresponding to a region with components exceeding a threshold, or the area corresponding to a region with a specific component. The highlighted portion is a portion that is represented in a way that distinguishes it from other portions, or a portion that is displayed only to a limited extent. The highlighted portion may be distinguished from other portions by changes in brightness or hue. The component distribution corresponds to the distribution of features calculated by component analysis.

[0024] In this embodiment, KMD analysis is performed based on the mass of repeating units in the sample under analysis. The width of each integration interval on the NKM axis is determined based on the mass of the repeating units. Since multiple plot positions (ionic intensity plot positions) on the NKM axis are arranged at intervals corresponding to the mass of the repeating units, according to the above configuration, multiple integration intervals are evenly allocated to multiple plot positions. The sample under analysis is the sample that produced the mass spectrum subject to KMD analysis.

[0025] In this embodiment, the n KMD analysis results are each ion intensity distributions on a two-dimensional coordinate system defined by the NKM axis and the KMD axis. The integrator applies an integration process to the n ion intensity distributions, which are the n KMD analysis results, according to a matrix defined by multiple integration intervals on the NKM axis and multiple integration intervals on the KMD axis, thereby generating n integrated value distributions.

[0026] Each ion intensity distribution corresponds to an ion intensity matrix or KMD plot. The ion intensity distribution may be configured as a peak list containing KMD analysis results. Each integrated value distribution corresponds to an integrated value matrix or integrated KMD plot. The number of data points (number of integrated values) that make up the integrated value distribution is less than the number of data points (number of ion intensities) that make up the ion intensity distribution. When displaying the integrated value distribution, each integrated cell in the matrix may be a filled area, or display elements may be plotted against coordinates representing each integrated cell.

[0027] The mass spectral processing apparatus according to the embodiment includes a preprocessor. Prior to multivariate analysis, the preprocessor applies a smoothing process to each of the n integrated value distributions in the direction of the KMD axis, and / or applies an intensity correction to each of the n integrated value distributions in the direction of the NKM axis.

[0028] In the integrated value distribution, random errors are more likely to affect the plot coordinates in the direction of the KMD axis; in other words, the plot coordinates are more prone to shifting. Therefore, smoothing is applied to the integrated value distribution in the direction of the KMD axis as needed. Smoothing is not applied in the direction of the NKM axis.

[0029] In mass spectrometry, increasing ion mass can lead to a decrease in ionization efficiency, a decrease in ion detection sensitivity, and other issues. This is called mass discrimination. To mitigate or eliminate the effects of such phenomena, intensity correction is applied to each integrated value distribution in the direction of the NKM axis, as needed. In this case, multiple integrated value sequences parallel to the NKM axis may be multiplied by a correction coefficient that increases with increasing NKM. Intensity correction may also be applied to each ion intensity distribution in the direction of the NKM axis, yielding the same result.

[0030] The mass spectral processing device according to this embodiment includes a normalizer that applies a normalization process to a dataset. Multivariate analysis is applied to the dataset after the normalization process. A processor, described later, functions as the normalizer. The normalization process can optimize the multivariate analysis.

[0031] In this embodiment, the dataset consists of n parts extracted from n integrated value distributions. This configuration allows for narrowing down the target of multivariate analysis. The n parts may be identified according to a region set by the user or automatically in the above two-dimensional coordinate system.

[0032] In this embodiment, the n KMD analysis results are, each, ion intensity distributions on a two-dimensional coordinate system defined by the NKM axis and the KMD axis. The multivariate analysis is principal component analysis. The results of the multivariate analysis include a loading distribution on the same two-dimensional coordinate system as described above. The loading map, also called a loading plot, is a component quantity plot or feature quantity plot. Loading is the weight assigned to a variable, corresponding to the contribution of the variable to the principal component. For each coordinate on the two-dimensional coordinate system, n integrated values ​​are calculated. The n integrated values ​​associated with each coordinate correspond to one variable.

[0033] The mass spectrum processing method according to this embodiment comprises a first analysis step, an integration step, and a second analysis step. In the first analysis step, Kendrick mass defect (KMD) analysis is applied to n mass spectra (where n is an integer of 2 or more) obtained by mass spectrometry of multiple samples, thereby generating n KMD analysis results. In the integration step, integration processing is applied to the n KMD analysis results according to multiple integration intervals on the nominal Kendrick mass (NKM) axis and multiple integration intervals on the KMD axis, thereby generating n integrated value distributions as n reduced KMD analysis results. In the second analysis step, multivariate analysis is applied to a dataset constructed based on the n integrated value distributions.

[0034] According to the above configuration, KMD analysis and multivariate analysis are applied stepwise to n mass spectra. More specifically, KMD analysis is applied to n mass spectra, and then multivariate analysis is applied to the n KMD analysis results generated. Since integration processing is applied to the n KMD analysis results prior to multivariate analysis, the loss of fractional mass information is suppressed, and the computational load in multivariate analysis is also reduced.

[0035] The mass spectral processing method according to the embodiment further comprises a generation step and an evaluation step. In the generation step, at least one dual analysis result plot is generated based on at least one of the n KMD analysis results and the results of the multivariate analysis. In the evaluation step, the differences between multiple samples or the characteristics of each of the multiple samples are evaluated based on at least one dual analysis result plot.

[0036] A program for executing the mass spectrum processing method according to the embodiment is installed on the information processing device via a network or a portable storage medium. The information processing device has a non-temporary storage medium on which the program is stored.

[0037] (2)KMD analysis Before describing the details of the embodiment, let's explain Kendrick Mass Defect (KMD) analysis.

[0038] Mass spectrometry of a sample involves an ion source to ionize the sample, a mass spectrometer to separate and detect ions at each ion mass divided by valence (i.e., m / z), and an information processing device to create mass spectra and process data. When the sample is a polymer, an ion source following a soft ionization method such as MALDI (Matrix-Assisted Laser Desorption / Ionization) or ESI (Electrospray Ionization) is typically used. When using MALDI, mainly monovalent ions are produced. Therefore, MALDI is frequently used in the analysis of polymers with molecular weight distributions. In general, in mass spectrometry, 12 A unit system is used in which the mass of carbon (C) is expressed as 12U.

[0039] Polymer molecules contain multiple linked repeating units (also known as monomer units). The chemical composition of these repeating units determines the type of polymer. For example, in polyethylene, the repeating units are C2H4; in polypropylene, they are C3H6; in polystyrene, they are C8H8; and in polyethylene glycol, they are C2H4O.

[0040] The mass of a polymer molecule is determined by the composition of its repeating units, the number of repeating units (degree of polymerization), and the composition of its end groups. Mass spectrometry requires the ionization of the polymer molecule, and depending on the type of cationizing agent used, adduct ions may be attached to the polymer molecule ions. Examples of such adduct ions include proton adduct ions, sodium adduct ions, potassium adduct ions, and silver adduct ions.

[0041] Kendrick mass defect analysis (KMD analysis) is used in the analysis of polymers. KMD analysis is also sometimes used in the analysis of samples other than polymers.

[0042] For a given polymer mass M, the Kendrick mass (KM) is defined as follows:

[0043] KM = M × Mri / Mr (1)

[0044] Here, Mri is the integer mass of the repeating unit, and Mr is the precise mass of the repeating unit. The former integer mass can be determined from the latter precise mass. The precise mass is the mass including the decimal part.

[0045] The integer part of KM is defined as the nominal Kendrick mass (NKM). The Kendrick mass defect (KMD) is defined as the difference between NKM and KM, as follows:

[0046] KMD = NKM - KM (2)

[0047] On the other hand, the mass M of a polymer can generally be expressed as follows:

[0048] M = Mr × n + Me + Mc (3)

[0049] Here, n is the degree of polymerization, Me is the mass of the terminal group (the total mass of the two terminal groups), and Mc is the mass of the adduction (the total mass of multiple adductions if there are several). In the right-hand side of equation (3) above, (Mr × n) is the mass of the main chain portion, and (Me + Mc) is the mass of the portion other than the main chain portion (non-main chain portion).

[0050] Substituting the right-hand side of equation (3) for M in equation (1) above, KM can be expressed as follows.

[0051] KM = Mri × n + ( Me + Mc ) Mri / Mr (4)

[0052] Since the first term on the right-hand side of equation (4) above is an integer, it does not contribute to KMD. Taking this into account, from equations (2) and (4) above, KMD can be expressed as follows.

[0053] KMD = Round { ( Me + Mc ) Mri / Mr} - ( Me + Mc ) Mri / Mr (5)

[0054] The first term on the right-hand side of equation (5) above is NKM. Round{} means rounding. According to equation (5) above, KMD takes a value in the range of -0.5 to +0.5. KMD does not depend on the degree of polymerization n. More specifically, KMD does not depend on the mass of the main chain portion but depends on the mass of the non-main chain portion. In KMD analysis, the above coefficient (Mri / Mr) is used to obtain a feature that does not depend on the mass of the main chain portion.

[0055] For each peak in the polymer's mass spectrum, an NKM-KMD pair is calculated. That is, multiple NKM-KMD pairs corresponding to multiple peaks are obtained. Multiple NKM-KMD pairs are represented as multiple display elements on a two-dimensional coordinate system with NKM and KMD axes. This generates a KMD plot (more precisely, an NKM-KMD plot).

[0056] In a KMD plot, multiple display elements corresponding to multiple peaks arising from a given polymer are arranged parallel to the horizontal axis at equal intervals. For example, each display element is a circle, in which case the diameter of each circle is determined according to the area (ionic intensity) of each peak. KMDs with values ​​in the range of 0 to 1.0 are also known. Such KMDs can also be used in embodiments.

[0057] (3) Details of the embodiment Figure 1 discloses a mass spectral processing system according to an embodiment. The mass spectral processing system is used, for example, to identify the differences between multiple mass spectra obtained from multiple samples, the characteristics of each mass spectrum obtained from multiple samples, and so on.

[0058] The mass spectrum processing system consists of a measurement unit (measuring device) 10 and an information processing unit (information processing device) 12. The measurement unit 10 is composed of a mass spectrometer, and the information processing unit 12 is composed of a computer. The information processing unit 12 corresponds to a mass spectrum processing device.

[0059] The measurement unit 10 performs mass spectrometry on the sample. The measurement unit 10 consists of an ion source 16, a mass spectrometer 18, and a detector 20. The ion source 16 is, for example, an ion source that follows the MALDI method. The mass spectrometer 18 is, for example, a time-of-flight mass spectrometer. The detector 20 detects individual ions that have passed through the mass spectrometer 18. The detection signal output from the detector 20 is converted into detection data in a signal processing circuit (not shown), and this detection data is sent to the information processing unit 12.

[0060] In the embodiment, i samples S1 to Si are subjected to mass spectrometry. In the embodiment, i is 6. More specifically, the i samples S1 to Si are six types of ethylene oxide-propylene oxide (EO-PO) copolymer polymers. Each of these EO-PO copolymer polymers is subjected to j times (four times in the embodiment) of mass spectrometry. This yields 24 mass spectra. The repeating unit in ethylene oxide is C2H4O, and the repeating unit in propylene oxide is C3H6O.

[0061] The information processing unit 12 includes a processor 22, an input device 50, and a display device 52. The processor 22 performs multiple functions. In Figure 1, these multiple functions are represented by multiple blocks.

[0062] The mass spectrum generator 24 generates mass spectra (mass spectral data) based on the input detection data. As described above, in this embodiment, four mass analyses are performed for each of the six types of samples. The mass spectrum generator 24 generates 24 mass spectra based on 24 detection data points.

[0063] The list generator 26 generates a peak list for each mass spectrum based on the mass spectrum. The peak list is a list containing multiple ionic intensities corresponding to multiple peaks included in the mass spectrum, and more specifically, a list containing multiple ionic intensities corresponding to multiple m / z values. When generating the peak list, for example, for each peak, the m / z corresponding to the centroid point of that peak is identified, and the area of ​​that peak is identified as the ionic intensity. For m / z values ​​where no peak exists, 0 is associated as the ionic intensity. A concrete example of the peak list will be shown later. Deisotope processing may be applied to the mass spectrum or the peak list.

[0064] The KMD analyzer 28 applies KMD analysis to each mass spectrum. Prior to the KMD analysis, the mass (Mr) of one of the repeating units contained in the sample to be analyzed is specified. In the embodiment, C2H4O or C3H6O is specified as the repeating unit.

[0065] In KMD analysis, for each peak, the m / z corresponding to the peak is converted into a combination of NKM as an integer mass and KMD as a decimal mass. In practice, each m / z in the peak list is converted into a combination of NKM and KMD. This generates a peak list containing the KMD analysis results for each mass spectrum. The peak list containing the KMD analysis results corresponds to the ion intensity distribution on a two-dimensional coordinate system defined by the NKM axis and the KMD axis, or to a KMD plot having said two-dimensional coordinate system. From this perspective, the KMD analyzer 28 can be called an ion intensity distribution generator and a KMD plot generator. In Figure 1, the symbol 30 indicates the ion intensity distribution output from the KMD analyzer 28. In practice, the KMD analyzer 28 sequentially outputs 24 ion intensity distributions 30. Note that the display processor 32, described later, may generate KMD plots, etc.

[0066] The integrator 34 applies integration processing to individual KMD analysis results (specifically, individual ion intensity distributions 30) according to the matrix defined in the two-dimensional coordinate system described above. Specifically, in the two-dimensional coordinate system, multiple integration intervals are set on the NKM axis and multiple integration intervals are set on the KMD axis. This defines a matrix. The matrix consists of multiple integration cells arranged in a two-dimensional array. A concrete example of the matrix will be shown later.

[0067] The integrator 34 adds up the multiple ion intensities belonging to each integration cell, thereby obtaining an integrated value. An integrated value distribution is constructed from multiple integrated values ​​on a two-dimensional coordinate system defined by the NKM axis and the KMD axis. The integrated value distribution can also be called an integrated value plot (integrated KMD plot). A peak list having multiple coordinates on the two-dimensional coordinate system and multiple integrated values ​​associated with them may also be generated as the integrated value distribution.

[0068] The size ΔNKM of each integration interval on the NKM axis and the size ΔKMD of each integration interval on the KMD axis are specified by the user or are automatically specified depending on the situation. ΔNKM may be determined based on the mass (Mr) of the repeating unit. For example, ΔNKM may be determined by Round { Mr} × k or by Round { Mr} × (1 / k). Such a determination makes it possible to evenly distribute multiple ionic intensities discretely arranged on the NKM axis to multiple integration intervals on the NKM axis.

[0069] Considering that random mass errors occurring in the measurement unit (mass spectrometer) 10 will appear on the KMD axis, ΔKMD may be determined based on random mass errors. For example, ΔKMD may be twice the random mass error. Random mass errors can be easily identified experimentally.

[0070] The integrator 34 applies preprocessing to each generated integrated value distribution as needed. In other words, the integrator 34 functions as a preprocessor. For example, first preprocessing and second preprocessing may be used as preprocessing steps.

[0071] In the first preprocessing step, a smoothing process along the KMD axis is applied to each integrated value distribution. Each integrated value distribution consists of multiple integrated value columns (multiple vertical columns) parallel to the KMD axis. For example, a one-dimensional smoothing filter is applied to each integrated value column. In integrated value distributions, the effects of random errors are easily introduced along the KMD axis, in other words, plot coordinate shifts are likely to occur. The first preprocessing step can mitigate the effects of random errors. Generally, smoothing along the NKM direction is unnecessary.

[0072] In the second preprocessing step, an intensity correction along the NKM axis is applied to each integrated value distribution. Each integrated value distribution consists of multiple integrated value columns (multiple horizontal columns) parallel to the NKM axis. A correction coefficient is multiplied by each integrated value column. The correction coefficient increases with increasing NKM. Alternatively, a correction coefficient greater than 1 is set within the high-mass range on the NKM axis. In mass spectrometry, as ion mass increases, a decrease in ionization efficiency, a decrease in ion detection sensitivity, etc., may occur. The second preprocessing step is performed to mitigate or eliminate the effects of such phenomena. In Figure 1, reference numeral 36 indicates the integrated value distribution output from the integrator 34. In reality, 24 integrated value distributions are output from the integrator 34.

[0073] The dataset generator 38 constructs a dataset based on multiple integrated value distributions 36, and in this embodiment, based on 24 integrated value distributions. This dataset can be described as a reduced or compressed dataset. More specifically, the dataset generator 38 generates a one-dimensional integrated value sequence from individual two-dimensional integrated value distributions. Subsequently, the dataset is constructed by arranging the 24 one-dimensional integrated value sequences in order of spectral number.

[0074] The normalizer 40 applies a normalization process to the dataset prior to multivariate analysis (principal component analysis in the example shown in Figure 1). Examples of normalization processes include a first normalization process and a second normalization process. The dataset consists of multiple horizontally integrated value columns arranged vertically, or conversely, multiple vertically integrated value columns arranged horizontally. In the first normalization process, each horizontally integrated value column is normalized. In the second normalization process, each horizontally integrated value column is normalized. The first and second normalization processes are usually performed selectively. The first and second normalization processes will be described in detail later.

[0075] The principal component analyzer 42 performs principal component analysis based on the dataset. The principal component analysis identifies multiple principal components, and for each principal component, a score is calculated for the individual mass spectrum (i.e., individual samples). During this process, a loading (loading value) is calculated for each principal component and for each coordinate in the two-dimensional coordinate system described above. The loading is a weight for the variable, representing its contribution to the principal component. In this embodiment, one variable corresponds to one coordinate (more precisely, 24 integrated values ​​associated with that coordinate).

[0076] Each cumulative value that makes up the dataset is associated with a pair of NKM (specifically, an NKM interval identifier) ​​and KMD (specifically, a KMD identifier). Each loading corresponding to each cumulative value is also associated with a pair of NKM and KMD.

[0077] The score plot generator 44 generates one or more score plots based on the results of the principal component analysis. Generally, a score plot has two principal component axes. One principal component axis represents a specific principal component, and the other principal component axis represents another specific principal component.

[0078] The loading map generator 46 generates one or more loading maps corresponding to one or more principal components based on the results of principal component analysis. Each loading map consists of multiple loadings represented on a two-dimensional coordinate system defined by the NKM axis and the KMD axis. Loading maps may also be called loading distributions or loading plots. From another perspective, a loading map is a component map, a component quantity map, or a feature quantity map.

[0079] Generally, from the perspective of comparative or difference analysis of multiple samples, regions with large positive or negative loading in a loading map are more important regions. Conversely, regions with small positive or negative loading are less important regions.

[0080] In this embodiment, data representing individual loading maps is sent to the display processor 32 and the post-processor 48. The principal component analyzer 42 or the loading map generator 46 may provide the post-processor 48 with a list or table containing multiple loadings and their corresponding coordinate information for each principal component.

[0081] The post-processor 48 applies post-processing based on a specific loading map corresponding to the selected principal component to a specific KMD plot (ion intensity distribution) corresponding to the selected mass spectrum, thereby generating a component-reflecting KMD plot as both analysis result plots. Specifically, the post-processor 48 identifies enhanced regions based on a specific loading map and generates a KMD plot having enhanced portions corresponding to these enhanced regions. This KMD plot is the component-reflecting KMD plot.

[0082] For example, a region having positive loading above a positive threshold or negative loading above a negative threshold may be identified as an emphasis region. In this embodiment, the post-processor 48 also has a color calculation function. The emphasis region may be distinguished from other parts by luminance or hue.

[0083] KMD plots have a first mass resolution, while loading maps have a second mass resolution lower than the first resolution. Since component-reflecting KMD plots are generated by post-processing of KMD plots, they essentially have a first mass resolution. Furthermore, in component-reflecting KMD plots, important parts are broadly highlighted. Component-reflecting KMD plots provide extremely useful information in comparative or difference analyses involving multiple samples.

[0084] The user may select or sequentially select the mass spectrum and principal component of interest. In this case, the post-processor 48 selects a specific KMD plot corresponding to the mass spectrum of interest from among the multiple KMD plots generated. The post-processor 48 also selects a specific loading map corresponding to the principal component of interest from among the multiple loading maps generated. A component-reflecting KMD plot may be automatically generated for each combination defined based on the multiple KMD plots and loadings generated. A component-reflecting polymerized KMD plot may be generated based on a polymerized KMD plot consisting of multiple KMD plots.

[0085] The display processor 32 receives input data representing multiple mass spectra, multiple KMD plots, multiple component-reflecting KMD plots, multiple integrated value plots, multiple score plots, and multiple loading maps. The display processor 32 selects one or more images to display on the display 52 according to the user's instructions.

[0086] For example, based on the displayed KMD plots reflecting multiple components, the user may identify differences between multiple mass spectra, or identify the relative characteristics of each of the multiple mass spectra. Through such evaluation, the best sample suitable for a particular purpose may be selected from among multiple samples. The best sample may be a material constituting a specific chemical product.

[0087] The display processor 32 may also function as an evaluator 53. The evaluator 53 identifies differences between multiple mass spectra or identifies the relative characteristics of each of the multiple mass spectra based on multiple KMD plots after post-processing. User instructions are received via the input device 50. The input device 50 consists of a keyboard or a pointing device. The display device 52 consists of a liquid crystal display.

[0088] The mass spectrometry system and mass spectrum processing method according to the embodiment will be described in more detail below.

[0089] Figure 2 shows the six mass spectra (A1) to (A6) obtained from the six types of EO-PO copolymer polymers described above. In each mass spectrum, the horizontal axis is the m / z axis, and the vertical axis is the intensity axis.

[0090] Figure 3 shows an example of a peak list including KMD analysis results. Reference numeral 54 indicates the portion corresponding to the peak list generated by the list generator. This portion 54 has multiple m / z values ​​on the m / z axis and multiple ionic intensities corresponding to them. Peak list 54A includes the KMD analysis results. That is, for each m / z value, NKM is associated with the m / z value as an integer mass and KMD as a decimal mass. In this embodiment, i × j (specifically 24) peak lists 54A are generated. The calculations for NKM and KMD may be performed in advance, and the calculation results may be registered in advance.

[0091] Figure 4 shows an example of a KMD plot (ionic intensity distribution). Prior to generating the KMD plot, the mass of the repeating unit is specified. In this embodiment, for example, 58.042, which is the mass of the repeating unit C3H6O in polypropylene oxide, is specified. In Figure 4, the KMD plot 56 includes multiple display elements corresponding to multiple peaks contained in the mass spectrum. Each display element 56a is a circle as a geometric figure, and its diameter represents the ionic intensity. In the KMD plot 56, the horizontal axis is the KMD axis representing NKM, and the vertical axis is the KMD axis representing KMD.

[0092] Figure 5 shows the polymerization KMD plot 58. The polymerization KMD plot 58 consists of six KMD plots generated from the six mass spectra shown in Figure 3. The content of the polymerization KMD plot 58 is complex, and it is very difficult to perform comparative analysis and difference analysis using the polymerization KMD plot 58.

[0093] Figure 6 shows an example of a matrix. A two-dimensional coordinate system is defined by the horizontal axis (NKM axis) and the vertical axis (KMD axis). Matrix 60 is defined on this two-dimensional coordinate system. Multiple integration intervals are set for the NKM axis. The width of each integration interval is ΔNKM. Similarly, multiple integration intervals are set for the KMD axis. The width of each integration interval is ΔKMD. Multiple integration cells 62 are defined by the multiple integration intervals on the NKM axis and the multiple integration intervals on the KMD axis.

[0094] Based on the ion intensity distribution (KMD plot), the multiple ion intensities belonging to each individual integration cell 62 are integrated to obtain an integrated value. The integrated value is associated with the coordinates of each cell (specifically, a combination of interval coordinates on the NKM axis and interval coordinates on the KMD axis). The integrated value distribution is constructed from the multiple integrated values ​​corresponding to multiple integration cells. In the matrix 60 shown in Figure 6, the coordinates at the beginning are (x1, y1) and the coordinates at the end are (xmax, ymax).

[0095] In this embodiment, for example, ΔNKM is 58 and ΔKMD is 0.02. In this case, 60 integration intervals are set on the NKM axis and 50 integration intervals are set on the KMD axis. The number of integration cells 62 is 3000. A high compression ratio is achieved while avoiding a large loss of fractional mass information.

[0096] By converting the ion intensity distribution to an integrated value distribution, the amount of data in the dataset targeted for principal component analysis can be significantly reduced. Moreover, since integer mass information and decimal mass information are resampled independently, the decimal mass information, which is important in comparative and difference analysis, is not significantly lost.

[0097] For the integrated value distribution, a first preprocessing step and / or a second preprocessing step are applied as necessary. The first preprocessing step is a smoothing step along the KMD axis, as previously described (see reference numeral 64). The second preprocessing step is an intensity correction step along the KNM axis, as previously described (see reference numeral 66).

[0098] Figure 7 shows an example of an integrated value distribution (integrated KMD plot). The illustrated integrated value distribution 68 consists of multiple display elements. Each display element corresponds to each integrated cell and has a rectangular shape. The density (actually hue) of each display element represents the magnitude of the integrated value.

[0099] Figure 8 shows an example of a dataset. The illustrated dataset 70 consists of multiple subsets 72-1 to 72-ij arranged in order of sample number, where ij is 24 in this embodiment. Each subset 72-1 to 72-ij is a one-dimensional integrated value sequence generated by transforming the integrated value distribution. The one-dimensional integrated value sequence consists of multiple integrated values ​​71. In each subset 72-1 to 72-ij, the starting coordinate corresponds to the above coordinate (x1, y1), and the ending coordinate corresponds to the above coordinate (xmax, ymax).

[0100] The first normalization process described above normalizes the individual subsets 72-1 to 72-ij. Specifically, multiple integrated values ​​are modified so that the sum Σ1 of the multiple integrated values ​​constituting each integrated value column (each horizontal column) becomes a predetermined value. The multiple total ion intensities corresponding to multiple mass spectra vary. The first normalization process can suppress the effects of this variation.

[0101] Dataset 70 consists of multiple vertical columns arranged horizontally. In the second normalization process, the sum Σ2 of the multiple integrated values ​​constituting each vertical column is calculated, and the multiple integrated values ​​are modified so that the sum Σ2 becomes a predetermined value. When the total ion intensity varies between multiple coordinates, the second normalization process can mitigate the effects of this variation. By applying the normalization process to the dataset, it becomes possible to perform principal component analysis appropriately.

[0102] Figure 9 shows two score plots, 73 and 74. In score plot 73, the horizontal axis is the first principal component axis, and the vertical axis is the second principal component axis. Each point 73a represents the combination of the first principal component score and the second principal component score corresponding to each sample.

[0103] In score plot 74, the horizontal axis represents the first principal component axis, and the vertical axis represents the third principal component axis. Each point 74a represents the combination of the first principal component score and the third principal component score corresponding to each sample. Through observation or evaluation of such score plots, the relationships between multiple spectra, i.e., the relationships between multiple samples, can be visually recognized or evaluated.

[0104] Figure 10 shows the loading map 76 corresponding to the first principal component. The loading map 76 is a loading distribution on a two-dimensional coordinate system defined by the NKM axis and the KMD axis. The loading map 76 is composed of multiple elements 76a representing multiple loadings, with each element 76a corresponding to an integration cell. The density of each element 76a represents the magnitude of the loading. In practice, the magnitude of positive and negative loadings are represented by changes in hue. Through observation or evaluation of the loading map 76, it is possible to identify which part of the multiple integration value distribution as a whole contributes most significantly to the first principal component.

[0105] Figure 11 shows the loading map 78 corresponding to the second principal component. Figure 12 shows the loading map 80 corresponding to the third principal component.

[0106] Next, post-processing will be described. In this embodiment, for example, as shown in Figure 13, a first emphasis interval 82A and a second emphasis interval 82B are set on the loading axis. The first emphasis interval 82A is the interval exceeding a positive threshold, and the second emphasis interval 82B is the interval exceeding a negative threshold. When the loading of interest belongs to the first emphasis interval 82A or the second emphasis interval 82B, the display element representing the ionic intensity (more precisely, the ionic intensity of the peak in the mass spectrum) corresponding to the loading of interest is highlighted in the KMD plot. The emphasis portion may be defined by other methods. Alternatively, instead of highlighting, the display may be toggled on and off.

[0107] Figure 14 shows a first example of a component-reflecting KMD plot. In the illustrated component-reflecting KMD plot 84, for example, the first part 84A consists of multiple display elements corresponding to multiple positive loadings belonging to the first emphasis interval. For convenience, each such display element is referred to as a display element that satisfies the first emphasis condition. Display elements that satisfy the first emphasis condition have a first color. Also, for example, the second part 84B consists of multiple display elements corresponding to multiple negative loadings belonging to the second emphasis interval. For convenience, each such display element is referred to as a display element that satisfies the second emphasis condition. These display elements have a second color different from the first color.

[0108] Reference numeral 86 indicates a display element corresponding to loading that does not belong to any of the emphasis intervals. Such display elements are displayed at low brightness. Such display elements may be hidden.

[0109] Figure 15 shows a second example of a component-reflecting KMD plot. In the illustrated component-reflecting KMD plot 88, for example, the background of the first region 90A, where multiple display elements satisfying the first emphasis condition exist, is represented in the first color. The background of the second region 90B, where multiple display elements satisfying the second emphasis condition exist, is represented in the second color. Reference numeral 92A indicates the outer edge of the first region 90A, and reference numeral 92B indicates the outer edge of the second region 90B. Display elements that do not satisfy the first or second emphasis condition are not emphasized or are hidden (see reference numeral 86).

[0110] Figure 16 shows a third example of a component-reflecting KMD plot. In the illustrated component-reflecting KMD plot 94, for example, the outer edge 96A of the first region containing multiple display elements that satisfy the first emphasis condition is represented in the first color. The outer edge 96B of the second region containing multiple display elements that satisfy the second emphasis condition is represented in the second color. Display elements that do not satisfy the first or second emphasis condition are not emphasized or are hidden (see reference numeral 86).

[0111] As described above, according to the embodiment, it is possible to generate a special KMD plot in which important parts are highlighted from the viewpoint of comparative analysis or difference analysis. Since this special KMD plot has the same mass resolution as the first mass resolution of the original KMD plot, precise evaluation can be performed in a two-dimensional mass coordinate system.

[0112] Figure 17 shows another configuration example of the information processing apparatus according to the embodiment. In Figure 17, elements similar to those shown in Figure 1 are denoted by the same reference numerals, and their descriptions are omitted.

[0113] The information processing device 12A shown in Figure 17 has an analyzer 42A. The analyzer 42A performs vertex component analysis or hierarchical cluster analysis. The component analysis results from the analyzer 42A are sent to the map generator 46A and also to the display processor 32. The map generator 46A generates a component map (component distribution) that represents the component analysis results.

[0114] When vertex component analysis is performed in analyzer 42A based on the dataset, map generator 46A generates a component map 100, for example, as shown in Figure 18. The component map 100 shows the distribution of multiple components (component spectra).

[0115] In the illustrated example, the first component (first component spectrum) is dominant in region F1, the second component (second component spectrum) is dominant in region F2, the third component (third component spectrum) is dominant in region F3, and the fourth component (fourth component spectrum) is dominant in region F4. Reference numeral 102 indicates the display element corresponding to the integration cell.

[0116] A component-reflected KMD map is generated by applying the component map 100 to the KMD map. For example, a specific component (specific region) may be specified in the component map 100, and the part corresponding to that specific region in the KMD map may be designated as the highlighted part.

[0117] In the analyzer 42A shown in Figure 17, when hierarchical cluster analysis is performed based on the dataset, the map generator 46A generates, for example, the component map 104 shown in Figure 19. Region G1 is the part classified as the first component (first phase), region G2 is the region classified as the second component (second phase), region G3 is the region classified as the third component (third phase), and region G4 is the region classified as the fourth component (fourth phase).

[0118] A component-reflected KMD map is generated by applying the component map 104 to the KMD map. For example, a specific component (specific region) may be specified in the component map 104, and the part corresponding to that specific region in the KMD map may be highlighted.

[0119] Figure 20 shows a mass spectrometry processing method according to an embodiment. In S10, multiple samples S to be compared are prepared. In S12, mass spectrometry is performed sequentially on the multiple samples S. This generates n mass spectra, where n is an integer greater than or equal to 2, and is generally an integer greater than or equal to 3.

[0120] In S14, KMD analysis is sequentially performed on n mass spectra, generating n KMD plots (ion intensity distributions) 106 as the result of the n KMD analysis. In the KMD analysis, each m / z is decomposed into NKM, where the mass is an integer, and KMD, where the mass is a decimal. Also in S14, a polymerized KMD plot 108 is generated by combining all or part of the n KMD plots 106.

[0121] In S16, n integrated value distributions are generated by integrating n KMD plots 106. A reduced dataset 109 is constructed from these n integrated value distributions. In S18, component analysis is applied to the dataset 109. This generates a component map or component quantity map (including a loading map) 113.

[0122] In S22, post-processing based on the component map 113 is applied to the selected KMD plot 106, thereby generating a KMD plot (component-reflecting KMD plot) 110 with emphasized portions. Multiple component-reflecting KMD plots 110 may be generated based on multiple KMD plots 106, or a component-reflecting polymerized KMD plot 112 may be generated based on a polymerized KMD plot 108.

[0123] In S24, the user observes one or more component-reflecting KMD plots 110,112, evaluates the differences between multiple samples, and assesses the characteristics of each sample. Based on the results of such evaluations, a sample suitable for a particular purpose may be selected from among the multiple samples. In comparative or difference analysis, other information 111 representing the results of component analysis may be provided to the user.

[0124] Figure 21 shows the integrated value distribution 118. A portion belonging to a specified specific region 114 may be extracted from the integrated value distribution 118. For example, a dataset for multivariate analysis may be constructed from n portions extracted from n integrated value distributions 118. The specific region 114 may be specified by the user, or it may be set automatically. A portion 116 may be specified in the integrated value distribution 118 to be excluded from multivariate analysis. In the ion intensity distribution (i.e., KMD map), the portion to be included in multivariate analysis may be specified, or portions to be excluded from multivariate analysis may be specified. [Explanation of Symbols]

[0125] 10 Measurement unit, 12 Information processing unit, 22 Processor, 28 KMD analyzer, 34 Integrated unit, 38 Data set creator, 40 Normalizer, 42 Principal component analyzer, 44 Score plot generator, 46 Loading map generator, 32 Display processor.

Claims

1. A KMD analyzer that applies Kendrick mass defect (KMD) analysis to n mass spectra (where n is an integer of 2 or more) obtained by mass spectrometry of multiple samples, thereby generating n KMD analysis results, An integrator that applies integration processing according to multiple integration intervals on the nominal Kendrick mass (NKM) axis and multiple integration intervals on the KMD axis to the aforementioned n KMD analysis results, thereby generating n integrated value distributions as n reduced KMD analysis results, A multivariate analyzer that applies multivariate analysis to a dataset constructed based on the aforementioned n cumulative value distributions, A mass spectral processing apparatus characterized by including

2. In the mass spectral processing apparatus according to claim 1, Includes a generator that generates at least one plot of both analysis results based on at least one of the n KMD analysis results and the results of the multivariate analysis, A mass spectral processing apparatus characterized by the following:

3. In the mass spectral processing apparatus according to claim 2, The n KMD analysis results are, respectively, ion intensity distributions on a two-dimensional coordinate system defined by the NKM axis and the KMD axis. The aforementioned multivariate analysis is a component analysis, The results of the multivariate analysis include the component distributions on the same two-dimensional coordinate system as the aforementioned two-dimensional coordinate system. The generator generates at least one analysis result plot based on at least one ion intensity distribution and the component distribution. A mass spectral processing apparatus characterized by the following:

4. In the mass spectral processing apparatus according to claim 3, The at least one of the analysis result plots is a component-reflecting KMD plot having an emphasis portion identified based on the component distribution. A mass spectral processing apparatus characterized by the following:

5. In the mass spectral processing apparatus according to claim 1, The KMD analysis described above is performed based on the mass of repeating units in the sample being analyzed. The width of each integration interval on the NKM axis is determined based on the repeating unit. A mass spectral processing apparatus characterized by the following:

6. In the mass spectral processing apparatus according to claim 1, The n KMD analysis results are, respectively, ion intensity distributions on a two-dimensional coordinate system defined by the NKM axis and the KMD axis. The integrator applies the integration process to the n ion intensity distributions, which are the n KMD analysis results, according to a matrix defined by the plurality of integration intervals on the NKM axis and the plurality of integration intervals on the KMD axis, thereby generating the n integrated value distributions. A mass spectral processing apparatus characterized by the following:

7. In the mass spectral processing apparatus according to claim 6, Prior to the multivariate analysis, a preprocessor is included that applies a smoothing process to each of the n integrated value distributions in the direction of the KMD axis, A mass spectral processing apparatus characterized by the following:

8. In the mass spectral processing apparatus according to claim 6, Prior to the multivariate analysis, a preprocessor is included that applies an intensity correction in the direction of the NKM axis to each of the n integrated value distributions. A mass spectral processing apparatus characterized by the following:

9. In the mass spectral processing apparatus according to claim 1, Includes a normalizer that applies a normalization process to the aforementioned dataset, The multivariate analysis is applied to the dataset after the normalization process. A mass spectral processing apparatus characterized by the following:

10. In the mass spectral processing apparatus according to claim 1, The aforementioned dataset is composed of n parts extracted from the aforementioned n integrated value distributions. A mass spectral processing apparatus characterized by the following:

11. In the mass spectral processing apparatus according to claim 1, The n KMD analysis results are, respectively, ion intensity distributions on a two-dimensional coordinate system defined by the NKM axis and the KMD axis. The aforementioned multivariate analysis is principal component analysis, The results of the multivariate analysis include a loading distribution on the same two-dimensional coordinate system as the aforementioned two-dimensional coordinate system. A mass spectral processing apparatus characterized by the following:

12. The process involves applying Kendrick Mass Defect (KMD) analysis to n mass spectra (where n is an integer of 2 or more) obtained by mass spectrometry of multiple samples, thereby generating n KMD analysis results. The process involves applying an integration process to the n KMD analysis results according to multiple integration intervals on the nominal Kendrick mass (NKM) axis and multiple integration intervals on the KMD axis, thereby generating n integrated value distributions as n reduced KMD analysis results. The process of applying multivariate analysis to a dataset constructed based on the aforementioned n cumulative value distributions, A mass spectral processing method characterized by including the following.

13. In the mass spectral processing method according to claim 12, Furthermore, the process includes generating at least one plot of both analysis results based on at least one of the n KMD analysis results and the results of the multivariate analysis, A step of evaluating the differences between the plurality of samples or the characteristics of each of the plurality of samples based on the at least one of the analysis result plots, A mass spectral processing method characterized by including the following.

14. A program for executing a mass spectrum processing method in an information processing device, The mass spectral processing method described above is: The process involves applying Kendrick Mass Defect (KMD) analysis to n mass spectra (where n is an integer of 2 or more) obtained by mass spectrometry of multiple samples, thereby generating n KMD analysis results. The process involves applying an integration process to the n KMD analysis results according to multiple integration intervals on the nominal Kendrick mass (NKM) axis and multiple integration intervals on the KMD axis, thereby generating n integrated value distributions as n reduced KMD analysis results. The process of applying multivariate analysis to a dataset constructed based on the aforementioned n cumulative value distributions, A program characterized by including the following.

Images