Data-analysis-based resolution improvement method and system for mass spectrometer

Abstract

The present invention relates to the technical field of data processing. Disclosed are a data-analysis-based resolution improvement method and system for a mass spectrometer. The method comprises: training a resolution factor model; on the basis of an original mass spectrum and an original fragment map, determining basic information of a compound to be tested, determining a target signal-to-noise ratio of the original mass spectrum on the basis of a mass spectrum of a standard compound, and performing denoising processing on the original mass spectrum by means of a denoising method; on the basis of the denoised original mass spectrum and the mass spectrum of the standard compound, determining a baseline reference region; identifying features of peaks in the mass spectrum, separating overlapped peaks by means of the features of the peaks, and performing clustering analysis on all the peaks on the basis of a resolution factor; and performing dimension reduction processing on a class of peaks having data dimensions exceeding a dimension threshold value. In this way, data inconsistency caused by baseline drifts is reduced or avoided. Finally, the resolution of a mass spectrometer is improved by means of the separation of overlapped peaks, the clustering of peaks, and dimension reduction processing of the peaks, thereby ensuring the efficiency of resolution improvement.

Description

Mass spectrometer resolution enhancement method and system based on data analysis TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, more particularly, to a mass spectrometer resolution enhancement method and system based on data analysis. BACKGROUND

[0002] A mass spectrometer is an analytical instrument used to measure the mass and quantity of ions in an ionized gas sample. Its working principle is based on converting the compounds in the sample into gaseous ions, and then separating and detecting them according to their mass-to-charge ratio (m / z). Mass spectrometers have a wide range of applications in chemistry, biochemistry, drug development, environmental science, forensic science, and many other scientific fields. Through the data generated by mass spectrometers, scientists can gain a deep understanding of the composition and properties of matter, thereby promoting the development of scientific research and applications.

[0003] In the prior art, the resolution of the mass spectrometer is often improved by improving the hardware and software, but this improvement method has high cost and long cycle, resulting in poor adaptability and low efficiency of resolution improvement. The data related to the mass spectrometer can directly or indirectly reflect the resolution, and the resolution of the mass spectrometer can be improved through data analysis and processing.

[0004] Therefore, how to provide a mass spectrometer resolution enhancement method through data analysis is a technical problem to be solved at present. SUMMARY

[0005] The purpose of the present application is to solve the technical problems of high cost and low efficiency of resolution improvement in the prior art, and to provide a mass spectrometer resolution enhancement method based on data analysis.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0007] A mass spectrometer resolution enhancement method based on data analysis, comprising:

[0008] Training a resolution factor model through data of mass analyzers, ion optics systems, ion transmission, and ion flight paths and collision processes, the resolution factor describing the resolution potential of the mass spectrometer under different conditions;

[0009] Determining the basic information of the test compound according to the original mass spectrum and the original fragment spectrum, finding the closest standard compound in the standard compound library according to the basic information of the test compound, determining the target signal-to-noise ratio of the original mass spectrum based on the mass spectrum of the standard compound, and performing denoising processing on the original mass spectrum to achieve the target signal-to-noise ratio through a denoising means;

[0010] A baseline reference region is determined according to the denoised original mass spectrum and the mass spectrum of the standard compound, and the baseline reference region is optimized by an ALS algorithm to obtain a mass spectrum after removing baseline drift;

[0011] Each feature of the peaks in the mass spectrum is identified, and the overlapping peaks are separated by the peak features to obtain all the peaks, and the similar peaks are classified by cluster analysis according to the resolution factor;

[0012] The data dimension of each type of peak is determined, and the peaks of one type whose data dimension exceeds a dimension threshold are processed by dimension reduction to improve the resolution of the mass spectrometer;

[0013] The original fragment map is an ion fragment map after activation.

[0014] In some embodiments of the present application, a resolution factor model is trained by data of mass analyzers, ion optical systems, ion transmission and flight paths and collision processes of ions, including:

[0015] The data of mass analyzers, ion optical systems, ion transmission and flight paths and collision processes of ions and corresponding resolution data within a period of time are collected, and the corresponding time series data of each part is generated in chronological order to determine the influence degree of each part of the time series data on the resolution data;

[0016] The hierarchical structure is defined as ion source layer-ion transmission layer-mass analysis layer-detector layer in turn, and the time series data of mass analyzers, ion optical systems, ion transmission and flight paths and collision processes of ions are input correspondingly according to the ion source layer, ion transmission layer, mass analysis layer and detector layer;

[0017] The training ratio is determined according to the ratio of the influence degrees of the multi-part time series data on the resolution data, each part of the time series data is cut according to the training ratio, each part of the time series data is cut into a training set and a test set, and the training set and the test set of each part of the time series data are integrated;

[0018] The multi-layer model of the hierarchical structure is trained by the training set, and the multi-layer model is adjusted and optimized according to the test set.

[0019] In some embodiments of the present application, the basic information of the to-be-tested compound is used to find the closest standard compound in the standard compound library, including:

[0020] The basic information includes structural component information and peak characteristics, and the structural component information of the to-be-tested compound is compared with the structural component information of each standard compound in the standard compound library to screen a plurality of to-be-determined standard compounds;

[0021] The difference between the to-be-tested compound and each structural component information of the to-be-determined standard compound is normalized, the comprehensive structural component information difference index is obtained after integrating the difference of each structural component information, and the comprehensive structural component information difference is used for screening in a plurality of to-be-determined standard compounds, so as to screen out the first to-be-determined standard compound and the second to-be-determined standard compound;

[0022] The similarity of the peak characteristics of the first to-be-determined standard compound, the second to-be-determined standard compound and the to-be-tested compound is compared in sequence, so as to determine the basic information similarity index in combination with the comprehensive structural component information difference index;

[0023] ;

[0024] Wherein, The basic information similarity index of the to-be-tested compound x and the first to-be-determined standard compound or the second to-be-determined standard compound y, n is the number of peak characteristics, The weight corresponding to the i th peak characteristic, The size of the i th peak characteristic of the to-be-tested compound, The size of the i th peak characteristic of the first to-be-determined standard compound or the second to-be-determined standard compound, The similarity of the peak characteristics, exp is the exponential function, The comprehensive structural component information difference index of the first to-be-determined standard compound or the second to-be-determined standard compound, The corresponding constant of the first to-be-determined standard compound or the second to-be-determined standard compound;

[0025] The first to-be-determined standard compound or the second to-be-determined standard compound with the minimum basic information similarity index is taken as the standard compound closest to the to-be-tested compound.

[0026] In the following embodiments of the present application, the target signal-to-noise ratio of the original mass spectrum is determined based on the mass spectrum of the standard compound, which comprises:

[0027] The mass spectrum of the standard compound is obtained, several representative peaks are determined on the mass spectrum, the areas of the representative peaks are calculated, and the peak threshold region is determined according to the types of the representative peaks. The signal-to-noise ratio of the peak is determined by the area of the peak and the peak threshold region, the signal-to-noise ratios of all representative peaks are integrated, and the signal-to-noise ratio of the mass spectrum of the standard compound is obtained;

[0028] A target percentage is obtained by mapping the basic information similarity index of the to-be-tested compound and the standard compound, and the target signal-to-noise ratio of the original mass spectrum is determined according to the target percentage and the signal-to-noise ratio of the mass spectrum of the standard compound.

[0029] In some embodiments of the present application, the baseline reference region is determined according to the denoised original mass spectrum and the mass spectrum of the standard compound, including:

[0030] The minimum signal value in the starting point peripheral region on the denoised original mass spectrum is determined, and the maximum signal value on the denoised original mass spectrum is determined, and the extension distance is determined by the ratio of the minimum signal value to the maximum signal value;

[0031] The point of the minimum signal value in the starting point peripheral region is taken as the baseline starting point, and the baseline starting point is extended according to the extension distance to determine the first baseline region, and the second baseline region is determined in the mass spectrum of the standard compound;

[0032] If the first baseline region and the second baseline region have an intersection, the intersection baseline region is taken as the baseline reference region;

[0033] Otherwise, the region between the first baseline region and the second baseline region is taken as the baseline reference region.

[0034] In some embodiments of the present application, the overlapping peaks are separated by peak characteristics, and all peaks are obtained, including:

[0035] All peak characteristics are standardized, and IMDA or ICA is used to identify and separate overlapping peaks to obtain all single peaks.

[0036] In some embodiments of the present application, all peaks are subjected to cluster analysis according to the resolution factor, so that similar peaks are classified, including:

[0037] Several representative peaks on the mass spectrum are determined, and the area of each representative peak is calculated, and the minimum distance in the cluster algorithm is determined in combination with the resolution factor;

[0038] ;

[0039] Wherein, L is the minimum distance in the cluster algorithm, is the resolution factor, m is the number of types of representative peaks, is the area ratio of the jth peak, is the area of the jth peak, is the initial minimum distance, indicates the initial minimum distance obtained by area average value mapping;

[0040] According to the minimum distance, all peaks are subjected to cluster analysis, and similar peaks are classified.

[0041] In some embodiments of the present application, the data dimension in each type of peak is determined, and one type of peak with a data dimension exceeding a dimension threshold is subjected to dimension reduction processing, including:

[0042] The data dimension of each type of peak is evaluated, the position and retention time of each type of peak are calculated, the dimension threshold is set by the standard deviation of the position and retention time, and the dimension reduction processing is performed on the type of peak whose data dimension exceeds the dimension threshold.

[0043] Correspondingly, the application also provides a mass spectrometer resolution improvement system based on data analysis, comprising:

[0044] The training module is configured to train a resolution factor model by data of the mass analyzer, the ion optical system, ion transmission, and the flight path and collision process of ions, the resolution factor describing the resolution potential of the mass spectrometer under different conditions.

[0045] The determination module is configured to determine the basic information of the to-be-tested compound according to the original mass spectrum and the original fragment spectrum, find a standard compound closest to the to-be-tested compound in a standard compound library, determine the target signal-to-noise ratio of the original mass spectrum based on the mass spectrum of the standard compound, and perform denoising processing on the original mass spectrum by a denoising method to achieve the target signal-to-noise ratio.

[0046] The removal module is configured to determine a baseline reference region according to the denoised original mass spectrum and the mass spectrum of the standard compound, and optimize the baseline reference region by an ALS algorithm to obtain a mass spectrum after removal of baseline drift.

[0047] The classification module is configured to identify various features of peaks in the mass spectrum, separate the overlapping peaks by the peak features, thereby obtaining all the peaks, perform cluster analysis on all the peaks according to the resolution factor, and thereby classify similar peaks.

[0048] The dimension reduction module is configured to judge the data dimension of each type of peak, and perform dimension reduction processing on the type of peak whose data dimension exceeds the dimension threshold, thereby improving the resolution of the mass spectrometer.

[0049] The original fragment spectrum is an ion fragment spectrum after activation.

[0050] By applying the above technical solution, a resolution factor model is trained by data of a mass analyzer, an ion optical system, ion transmission, and a flight path and collision process of ions, the resolution factor describing resolution potential of the mass spectrometer under different conditions; basic information of the to-be-tested compound is determined according to an original mass spectrum and an original fragment spectrum, a standard compound closest to the to-be-tested compound is found in a standard compound library according to the basic information of the to-be-tested compound, a target signal-to-noise ratio of the original mass spectrum is determined based on a mass spectrum of the standard compound, and the original mass spectrum is denoised to reach the target signal-to-noise ratio through a denoising means; a baseline reference region is determined according to the denoised original mass spectrum and the mass spectrum of the standard compound, and the baseline reference region is optimized through an ALS algorithm to obtain a mass spectrum after removal of baseline drift; each feature of a peak in the mass spectrum is identified, and the overlapping peaks are separated through the peak features to obtain all peaks, and all the peaks are subjected to cluster analysis according to the resolution factor, so that similar peaks are classified; data dimensions in each class of peaks are determined, and a class of peaks with a data dimension exceeding a dimension threshold is subjected to dimension reduction processing, so as to improve resolution of the mass spectrometer. The resolution factor model is trained to help subsequent cluster analysis of the peaks, the target signal-to-noise ratio of the original mass spectrum is determined based on the mass spectrum of the standard compound, the mass spectrum is reasonably denoised, and problems such as excessive or insufficient noise and distortion are avoided, so that data quality of the mass spectrum is improved. The baseline reference region is determined according to the denoised original mass spectrum and the mass spectrum of the standard compound, and the baseline is optimized, so as to reduce or avoid data inconsistency caused by baseline drift. Finally, the resolution of the mass spectrometer is improved through separation of the overlapping peaks, cluster analysis of the peaks, and dimension reduction processing of the peaks, and resolution improvement efficiency is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0051] FIG. 1 is a flowchart of a mass spectrometer resolution improvement method based on data analysis according to the present application;

[0052] FIG. 2 is a structural diagram of a mass spectrometer resolution improvement system based on data analysis according to the present application. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0054] The embodiments of the present application provide a mass spectrometer resolution improvement method based on data analysis. The working process and principle of the mass spectrometer can be summarized as follows:

[0055] 1. Sample introduction:

[0056] The sample is first converted into gaseous ions through various ionization techniques. Common ionization techniques include electron impact ionization (EI), electrospray ionization (ESI), atmospheric pressure chemical ionization (APCI), etc. Each technique has its applicable sample types and analysis purposes.

[0057] 2. Ion Transmission:

[0058] The ionized sample enters the ion transmission system, which usually includes components such as ion mirrors, lenses, etc., to guide and focus ions, ensuring they effectively enter the mass analyzer.

[0059] 3. Mass Analysis:

[0060] The mass analyzer is the core part of the mass spectrometer, which separates ions according to their mass / charge ratio (m / z). Common types of mass analyzers include magnetic mass spectrometers, quadrupole rods, time-of-flight (TOF), ion traps, and Fourier transform ion cyclotron resonance (FT-ICR), etc. Each analyzer works differently, but the goal is to separate ions according to their mass / charge ratio.

[0061] 4. Ion Detection:

[0062] Separated ions are sent to the detector, which converts ions into electrical signals. The detector can be an electron multiplier, microchannel plate (MCP), or other types of photodetectors.

[0063] 5. Signal Recording and Processing:

[0064] The signals generated by the detector are recorded and processed through the data system. These data include the mass and relative intensity of each ion, ultimately forming a mass spectrum.

[0065] 6. Data Analysis:

[0066] Mass spectra usually need to be analyzed by professional software to identify and quantify compounds in the sample. Data analysis may include peak detection, baseline correction, isotope analysis, fragmentation analysis, etc.

[0067] 7. Result Interpretation:

[0068] Based on the information provided by the mass spectrum, the composition, structure, relative concentration, etc. of the sample can be interpreted. This may involve comparison with databases, confirmation of known compounds, or structure inference of unknown compounds.

[0069] A mass spectrum is produced by a mass spectrometer and shows the mass-to-charge ratio (m / z) of different ions in a sample and their relative intensities. In a mass spectrum, the horizontal axis represents the mass-to-charge ratio (m / z) of the ions, and the vertical axis represents the intensity or number of ions corresponding to each m / z value. Fragmentation and mass spectrum are conceptually related but not exactly the same thing. Fragmentation usually refers to the ion fragmentation pattern detected by the mass spectrometer after performing collision-induced dissociation (CID) or other forms of activation. This graph shows the mass and relative intensity of fragment ions produced by the original ion (parent ion) under the action of impact or electron bombardment, etc. Mass spectrum is a broader concept that includes any type of mass spectrum data showing the mass and relative intensity of all ions in the sample, whether they are intact molecular ions, fragment ions, isotope ions, or other forms of ions. Mass spectrum can be the ion spectrum detected directly without any activation, or the fragment ion spectrum obtained after some form of activation (such as CID).

[0070] As shown in FIG. 1, the method includes the following steps:

[0071] Step S101, training a resolution factor model by data of the mass analyzer, ion optics system, ion transmission, and flight path and collision process of ions, which describes the resolution potential of the mass spectrometer under different conditions.

[0072] In this embodiment, the parameters of these processes all affect the resolution, so they are considered to train a resolution factor model. Operating parameters of the mass analyzer: in a magnetic mass spectrometer, adjusting the magnetic field strength can affect the degree of ion deflection in the magnetic field, thereby affecting the resolution. In a quadrupole mass spectrometer, adjusting the radio frequency can optimize the focusing and separation of ions in the quadrupole, thereby affecting the resolution. Ion optics system: the design and configuration of ion mirrors and lenses have a significant impact on the transmission and focusing of ions. The parameters of these components can affect the transmission efficiency and focusing effect of ions, thereby changing the resolution. Ion source and transmission efficiency: the transmission efficiency of ions in the mass spectrometer directly affects the resolution. By optimizing the ion source parameters and transmission system, the ion transmission efficiency can be improved, thereby affecting the resolution. Flight path and collision process of ions: the flight path of ions in the mass analyzer directly affects the resolution. Flight path parameters, such as the distance between the ion source and the mass analyzer, the length of the flight path, etc., can affect the resolution. Mass deviation may occur during the collision process of ions, affecting the resolution.

[0073] In some embodiments of the present application, a resolution factor model is trained by data of the mass analyzer, ion optics system, ion transmission, and flight path and collision process of ions, including:

[0074] Data of the mass analyzer, ion optical system, ion transmission, and flight path and collision process of ions in a period of time and corresponding resolution data are collected, and corresponding time series data of each part are generated in chronological order, and the influence degree of each part of the time series data on the resolution data is determined;

[0075] The defined hierarchy is ion source layer-ion transmission layer-mass analysis layer-detector layer in turn, and the time series data of the mass analyzer, ion optical system, ion transmission, and flight path and collision process of ions are input correspondingly according to the ion source layer, ion transmission layer, mass analysis layer, and detector layer;

[0076] The training ratio is determined according to the ratio of the influence degrees of the multi-part time series data on the resolution data, each part of the time series data is cut according to the training ratio, and each part of the time series data is cut into a training set and a test set, so as to integrate the training set and the test set of each part of the time series data;

[0077] The multi-layer model of the hierarchy of ion source layer-ion transmission layer-mass analysis layer-detector layer is trained by the training set, and the multi-layer model is adjusted and optimized according to the test set.

[0078] In this embodiment, the defined hierarchy is ion source layer-ion transmission layer-mass analysis layer-detector layer in turn, because the occurrence sequence is like this, so as to completely simulate the specific condition. The hierarchy of the mass spectrometer is defined:

[0079] Ion source layer: contains parameters of the ion source, such as temperature, voltage, etc.

[0080] Ion transmission layer: contains parameters of the ion transmission system, such as voltage of ion mirror, lens, etc.

[0081] Mass analysis layer: contains parameters of the mass analyzer, such as voltage, magnetic field strength, etc.

[0082] Detector layer: contains parameters of the detector, such as gain, voltage, etc.

[0083] The parameter adjustment of each layer is transmitted to the next layer through connection, and affects the resolution of the next layer.

[0084] In this embodiment, the influence degree of each part of the time series data on the resolution data can be determined by an influence amount calculation method (Pearson correlation coefficient, etc.) to determine the influence degree on the resolution data. The training ratio is determined according to the ratio of the influence degrees of the multiple parts of the time series data on the resolution data, and the relative influence degree of each part of the time series data is determined, so as to determine the training ratio. Different relative influence degrees correspond to different training ratios. Each part of the time series data is cut according to the training ratio. The training ratio refers to the ratio in time. The data in the previous part of time is used as the training set, and the data in the latter part of time is used as the test set.

[0085] It should be noted that the specific model training and adjustment optimization process is not the focus of protection of the present scheme, and as long as the above requirements can be met, it is within the protection scope of the present application.

[0086] In step S102, the basic information of the to-be-tested compound is determined according to the original mass spectrum and the original fragment spectrum, a standard compound closest to the to-be-tested compound is found in a standard compound library, the target signal-to-noise ratio of the original mass spectrum is determined based on the mass spectrum of the standard compound, and the original mass spectrum is denoised to achieve the target signal-to-noise ratio.

[0087] In this embodiment, by comparing the peaks in the original mass spectrum and the original fragment spectrum, the basic information (structural component information and peak characteristics) of the to-be-tested compound can be determined. Here, a preliminary judgment is made to find a standard compound closest to the to-be-tested compound. The target signal-to-noise ratio and the baseline region are set by the standard compound to preprocess the original mass spectrum.

[0088] In some embodiments of the present application, a standard compound closest to the to-be-tested compound is found in a standard compound library based on the basic information of the to-be-tested compound, which includes:

[0089] The basic information includes structural component information and peak characteristics. The structural component information of the to-be-tested compound is compared with the structural component information of each standard compound in the standard compound library to screen a plurality of to-be-determined standard compounds.

[0090] The difference between the to-be-tested compound and each structural component information of the to-be-determined standard compound is normalized, and the comprehensive structural component information difference index is obtained after integrating the difference of each structural component information. The first to-be-determined standard compound and the second to-be-determined standard compound are screened out by screening the comprehensive structural component information difference in the plurality of to-be-determined standard compounds.

[0091] The similarity of the peak characteristics of the first to-be-determined standard compound, the second to-be-determined standard compound and the to-be-tested compound is compared in order, so as to determine the basic information similarity index in combination with the comprehensive structural component information difference index.

[0092] ;

[0093] wherein, is the basic information similarity index of the test compound x and the first or second tentative standard compound y, n is the number of peak characteristics, is the weight corresponding to the ith peak characteristic, is the size of the ith peak characteristic of the test compound, is the size of the ith peak characteristic of the first or second tentative standard compound, represents the similarity of the peak characteristics, exp is the exponential function, is the comprehensive structural component information difference index of the first or second tentative standard compound, is the corresponding constant of the first or second tentative standard compound;

[0094] The first or second tentative standard compound with the minimum basic information similarity index is taken as the standard compound closest to the test compound.

[0095] In this embodiment, the structural component information includes chemical structure, molecular formula, molecular weight, retention time, concentration, etc., and the peak characteristics include width, area, shape, position, intensity ratio, etc. By comparing the structural component information of the test compound with the structural component information of each standard compound in the standard compound library, a plurality of tentative standard compounds are screened out, and here the similar or close tentative standard compounds are screened out by a single structural component information. After integrating the differences of each structural component information, the comprehensive structural component information difference index (weighted sum) is obtained. The comprehensive structural component information difference index of the first tentative standard compound is smaller than that of the second tentative standard compound, and the smaller the comprehensive structural component information difference index, the smaller the comprehensive difference, and the closer to the test compound. The similarity of the peak characteristics of the first and second tentative standard compounds and the test compound is compared in order, and here the peak characteristic similarity is described by the Euclidean distance, and the smaller the value, the more similar. The value range is between 1-1.34, and the basic information similarity index is represented by combining the structural component information and the peak characteristic similarity.

[0096] In the following embodiments of the present application, the target signal-to-noise ratio of the original mass spectrum is determined based on the mass spectrum of the standard compound, comprising:

[0097] The mass spectrum of the standard compound is obtained, several representative peaks are determined on the mass spectrum, the areas of the representative peaks are calculated, and the peak margin area is determined according to the types of the representative peaks, the signal-to-noise ratio of the peaks is determined by the area of the peaks and the peak margin area, the signal-to-noise ratios of all the representative peaks are integrated, and the signal-to-noise ratio of the mass spectrum of the standard compound is obtained.

[0098] A target percentage is obtained by mapping the basic information of the standard compound and the compound to be tested, and the target signal-to-noise ratio of the original mass spectrum is determined according to the target percentage and the signal-to-noise ratio of the mass spectrum of the standard compound.

[0099] In this embodiment, the several representative peaks are the peaks with larger areas, because they usually represent compounds with higher concentrations, which is particularly important for quantitative analysis. Peaks with clear shapes are selected, which usually means that the contrast between signal and noise is higher, and the signal-to-noise ratio is better. The retention time of the peak, which is similar to the compound to be tested, helps to determine the identity of the compound.

[0100] In this embodiment, the types of the representative peaks determine the peak margin area, different types correspond to different sizes of the peak margin area, and the peak margin area refers to the distance around the peak. The signal-to-noise ratio of the peak is determined by the area of the peak and the peak margin area, and the signal-to-noise ratio is obtained by the ratio of the area to the noise level near the peak. Here, the signal-to-noise ratio of the peak is used to represent the signal-to-noise ratio of the mass spectrum, because the peak can better reflect the composition or other effective information. According to the target percentage * the signal-to-noise ratio of the mass spectrum of the standard compound = the target signal-to-noise ratio of the original mass spectrum, because the standard compound and the compound to be tested are different, the target signal-to-noise ratio is set according to the proportion.

[0101] It should be noted that the original mass spectrum is denoised by denoising means, and here it is not limited to using what means to denoise, as long as the target signal-to-noise ratio can be achieved. If the noise is too large, it will cause errors in analysis, and if the noise is too small, it may cause signal loss.

[0102] In step S103, the baseline reference area is determined according to the denoised original mass spectrum and the mass spectrum of the standard compound, and the baseline reference area is optimized by the ALS algorithm to obtain the mass spectrum after removing the baseline drift.

[0103] In this embodiment, the baseline will automatically change due to some factors, including changes in laboratory environment (such as temperature, humidity, electromagnetic interference, etc.), aging of instruments, differences in sample preparation process, etc. These factors can cause baseline drift, affecting the accuracy and repeatability of mass spectrometry data, and thus affecting the resolution. In order to remove the baseline drift, two commonly used baseline correction algorithms, Asymmetric Least Squares (ALS) and AirPLS, can be used, which calculate the baseline value at each time point by iteration and subtract it from the original signal to remove the baseline drift. Therefore, a baseline reference region needs to be selected to facilitate subsequent baseline optimization.

[0104] In some embodiments of the present application, the baseline reference region is determined according to the denoised original mass spectrum and the mass spectrum of the standard compound, including:

[0105] The minimum signal value in the starting point peripheral region of the denoised original mass spectrum is determined, and the maximum signal value in the denoised original mass spectrum is determined, and the extension distance is determined by the ratio of the minimum signal value to the maximum signal value;

[0106] The point of the minimum signal value in the starting point peripheral region is taken as the baseline starting point, and the baseline starting point is extended according to the extension distance to determine the first baseline region, and the second baseline region is determined in the mass spectrum of the standard compound;

[0107] If the first baseline region and the second baseline region have an intersection, the intersection baseline region is taken as the baseline reference region;

[0108] Otherwise, the region between the first baseline region and the second baseline region is taken as the baseline reference region.

[0109] In this embodiment, the determination of the baseline on the denoised original mass spectrum is usually based on the ordinate, i.e. the y-axis in the mass spectrum. The baseline represents the signal level when no ions pass through the detector, and a low value is usually set to facilitate the display of the intensity of the ion signal. The reason for selecting the low point part of the ion signal as the baseline starting point is that these regions usually have no significant ion signal peaks, so they can be used as the starting point of the baseline. When the mass spectrometer starts collecting data, there may be some preheating or preparation stage, which may cause the signal intensity at the starting point to be non-zero, so the relative minimum signal value in the peripheral region is found. The extension distance is the distance of the y-axis, and the ratio of the minimum signal value to the maximum signal value corresponds to different extension distances, which ensures that the baseline region is relatively complete. If the first baseline region and the second baseline region do not have an intersection, the region between the first baseline region and the second baseline region is taken as the baseline reference region. In general, even if the two baseline regions do not intersect, they will not differ much.

[0110] Step S104, identifying respective features of the peaks in the mass spectrum, separating the overlapping peaks by the peak features, obtaining all the peaks, performing cluster analysis on all the peaks according to the resolution factor, and classifying similar peaks.

[0111] In some embodiments of the present application, the overlapping peaks are separated by the peak features, and all the peaks are obtained, including:

[0112] The peak features are standardized, and IMDA or ICA is used to identify and separate the overlapping peaks to obtain all the single peaks.

[0113] In the present embodiment, intuitive discriminant analysis (IMDA) is used to calculate the similarity or distance between each peak according to the peak features, and the similarity or distance matrix can help to distinguish different peaks. According to the results of IMDA analysis, the overlapping peaks are identified and separated. Independent component analysis (ICA) is a signal processing technology that can separate signals mixed by multiple sources.

[0114] In some embodiments of the present application, cluster analysis is performed on all the peaks according to the resolution factor, and similar peaks are classified, including:

[0115] Several representative peaks on the mass spectrum are determined, and the area of each representative peak is calculated. The minimum distance in the cluster algorithm is determined in combination with the resolution factor;

[0116] ;

[0117] wherein L is the minimum distance in the cluster algorithm, is the resolution factor, m is the number of types of representative peaks, is the area ratio of the jth peak, is the area of the jth peak, is the initial minimum distance, denotes the initial minimum distance obtained by mapping the area average value;

[0118] According to the minimum distance, cluster analysis is performed on all the peaks, and similar peaks are classified.

[0119] In this embodiment, the percentage of peak area (area ratio) can help determine the minimum distance to some extent. The resolution factor can help determine the minimum distance in the clustering algorithm. The minimum distance refers to the minimum number of points that a core point must contain in the DBSCAN algorithm. The resolution factor can be used to guide the selection of the minimum distance, because high resolution usually means that finer clustering is needed. If the resolution factor indicates that the resolution is high, a smaller minimum distance can be selected to improve the granularity of clustering. Through clustering analysis, mass spectrometry peaks with similar characteristics can be distinguished, which may come from the same compound or homologues of the compound. Once these peaks are identified and separated, each type of peak can be analyzed separately, thereby improving the resolution.

[0120] In step S105, the data dimension in each type of peak is determined, and the dimension reduction processing is performed on the peak of one type whose data dimension exceeds the dimension threshold, so as to improve the resolution of the mass spectrometer. The original fragment graph is the ion fragment spectrum after activation.

[0121] In some embodiments of the present application, determining the data dimension in each type of peak and performing dimension reduction processing on the peak of one type whose data dimension exceeds the dimension threshold, includes:

[0122] In some embodiments of the present application, the data dimension of each type of peak is evaluated, the position and retention time of each type of peak are calculated, the dimension threshold is set by the standard deviation of the position and retention time, and the dimension reduction processing is performed on the peak of one type whose data dimension exceeds the dimension threshold.

[0123] In this embodiment, the position of the peak is the mass-to-charge ratio of the mass spectrum, and the retention time refers to the time required for the compound to pass through the chromatographic column in the gas chromatograph. A dimension threshold is set by the standard deviation of the position and retention time, and the position and retention time of the peak can reflect the separation of the compound. The dimension reduction technique can reduce the dimension of the data and remove the redundant information in the data. Reducing the dimension of the data can more clearly show the mass spectrum, thereby improving the resolution.

[0124] It should be noted that the specific clustering analysis process and dimension reduction process are not the focus of protection of the present scheme, and any scheme that can achieve the above requirements is acceptable. The corresponding relationship or mapping relationship involved in the entire text of the present application can be obtained from historical experience or mathematical relationship, and all of them are within the protection scope of the present application.

[0125] By applying the above technical solutions, a resolution factor model is trained by data of a mass analyzer, an ion optical system, ion transmission and a flight path and collision process of ions, the resolution factor describing resolution potential of the mass spectrometer under different conditions; basic information of the to-be-tested compound is determined according to an original mass spectrum and an original fragment spectrum, a standard compound closest to the to-be-tested compound is found in a standard compound library according to the basic information of the to-be-tested compound, a target signal-to-noise ratio of the original mass spectrum is determined based on a mass spectrum of the standard compound, and the original mass spectrum is denoised to reach the target signal-to-noise ratio through a denoising means; a baseline reference region is determined according to the denoised original mass spectrum and the mass spectrum of the standard compound, and the baseline reference region is optimized through an ALS algorithm to obtain a mass spectrum after removal of baseline drift; each feature of peaks in the mass spectrum is identified, and the overlapping peaks are separated through the peak features to obtain all the peaks, and all the peaks are subjected to cluster analysis according to the resolution factor, so that similar peaks are classified; data dimensions in each class of peaks are determined, and a class of peaks with a data dimension exceeding a dimension threshold is subjected to dimension reduction processing, so as to improve resolution of the mass spectrometer. The resolution factor model is trained to help subsequent cluster analysis of the peaks, the target signal-to-noise ratio of the original mass spectrum is determined based on the mass spectrum of the standard compound, the mass spectrum is reasonably denoised, and problems such as distortion caused by excessively large or small noise are avoided, and data quality of the mass spectrum is improved. The baseline reference region is determined according to the denoised original mass spectrum and the mass spectrum of the standard compound, and the baseline is optimized, so as to reduce or avoid data inconsistency caused by baseline drift. Finally, the resolution of the mass spectrometer is improved through separation of the overlapping peaks, cluster analysis of the peaks and dimension reduction processing of the peaks, and resolution improvement efficiency is ensured.

[0126] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by hardware, or by means of software and a necessary general hardware platform. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.), and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various implementation scenarios of the present application.

[0127] In order to further illustrate the technical idea of the present application, the technical solutions of the present application will be described in conjunction with specific application scenarios.

[0128] Correspondingly, the present application also provides a mass spectrometer resolution improvement system based on data analysis, as shown in FIG. 2, which comprises:

[0129] a training module configured to train a resolution factor model by data of the mass analyzer, the ion optics system, ion transmission and ion flight path and collision process, the resolution factor describing resolution potential of the mass spectrometer under different conditions;

[0130] a determination module configured to determine basic information of the to-be-tested compound according to the original mass spectrum and the original fragment spectrum, find a standard compound closest to the basic information of the to-be-tested compound in a standard compound library, determine a target signal-to-noise ratio of the original mass spectrum based on a mass spectrum of the standard compound, and perform denoising processing on the original mass spectrum by a denoising method to achieve the target signal-to-noise ratio;

[0131] a removal module configured to determine a baseline reference region according to the denoised original mass spectrum and the mass spectrum of the standard compound, and optimize the baseline reference region by an ALS algorithm to obtain a mass spectrum after removal of baseline drift;

[0132] a classification module configured to identify various features of peaks in the mass spectrum, separate overlapping peaks by the peak features, and obtain all peaks, and perform cluster analysis on the all peaks according to the resolution factor, so as to classify similar peaks;

[0133] a dimension reduction module configured to judge a data dimension of peaks in each class, and perform dimension reduction processing on a class of peaks whose data dimension exceeds a dimension threshold, so as to improve resolution of the mass spectrometer;

[0134] The original fragment spectrum is a spectrum of ion fragments after activation.

[0135] Those skilled in the art can understand that the modules in the system in the implementation scenario can be distributed in the system in the implementation scenario according to the description of the implementation scenario, or can be changed and located in one or more systems different from the implementation scenario. The modules in the above implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0136] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not drive the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for improving resolution of a mass spectrometer based on data analysis, characterized in that, Comprising: training a resolution factor model by data of a mass analyzer, data of an ion optical system, data of ion transmission, data of a flight path of ions and data of a collision process, the resolution factor describing resolution potential of a mass spectrometer under different conditions; determining basic information of a to-be-tested compound according to an original mass spectrum and an original fragment spectrum, finding a standard compound closest to the basic information of the to-be-tested compound in a standard compound library, determining a target signal-to-noise ratio of the original mass spectrum based on a mass spectrum of the standard compound, and performing denoising processing on the original mass spectrum by a denoising means to achieve the target signal-to-noise ratio; determining a baseline reference region according to the denoised original mass spectrum and the mass spectrum of the standard compound, and optimizing the baseline reference region by an ALS algorithm to obtain a mass spectrum after removing baseline drift; identifying various features of peaks in the mass spectrum, separating overlapping peaks by the peak features, thereby obtaining all peaks, and performing cluster analysis on all peaks according to the resolution factor, thereby classifying similar peaks; judging data dimensions in each class of peaks, and performing dimension reduction processing on a class of peaks with a data dimension exceeding a dimension threshold, thereby improving resolution of the mass spectrometer; the training process of the resolution factor model is as follows: collecting data of a mass analyzer, data of an ion optical system, data of ion transmission, data of a flight path of ions and data of a collision process and corresponding resolution data within a period of time, and generating time series data corresponding to each part in chronological order, and determining an influence degree of each part of the time series data on the resolution data; defining a hierarchical structure as ion source layer-ion transmission layer-mass analysis layer-detector layer in turn, and inputting time series data of the mass analyzer, the ion optical system, ion transmission and the flight path and collision process of ions according to the ion source layer, the ion transmission layer, the mass analysis layer and the detector layer; determining a training ratio according to a ratio of the influence degrees of the multiple parts of the time series data on the resolution data, cutting each part of the time series data according to the training ratio, cutting each part of the time series data into a training set and a test set, and thereby integrating the training set and the test set of each part of the time series data; training a multi-layer model of the hierarchical structure as ion source layer-ion transmission layer-mass analysis layer-detector layer in turn by the training set, and adjusting and optimizing the multi-layer model according to the test set. The original fragment spectrum is an ion fragment spectrum after activation, the data of the mass analyzer is an operating parameter of the mass analyzer, the data of the ion optical system is a design parameter of an ion mirror and a lens, the data of ion transmission is an ion transmission efficiency, the flight path data of ions is a flight path parameter of ions, and the data of the collision process is a parameter generated in the ion collision process.

2. The method of claim 1, wherein the method is based on data analysis. Finding a standard compound closest to the basic information of the to-be-tested compound in the standard compound library includes: The basic information includes structural component information and peak features, and a plurality of to-be-determined standard compounds are screened out by comparing the structural component information of the to-be-tested compound with the structural component information of each standard compound in the standard compound library. The difference between the to-be-tested compound and each structural component information of the to-be-determined standard compound is normalized, and a comprehensive structural component information difference index is obtained after integrating the difference of each structural component information, and the comprehensive structural component information difference is used for screening in a plurality of to-be-determined standard compounds, so as to screen out two types of first to-be-determined standard compounds and second to-be-determined standard compounds; The similarity of the peak characteristics of the first to-be-determined standard compound, the second to-be-determined standard compound and the to-be-tested compound is compared in sequence, so as to determine a basic information similarity index in combination with the comprehensive structural component information difference index; ； wherein Similarity index of basic information of the compound x to be tested and the first standard compound to be determined or the second standard compound to be determined, n is the number of peak characteristics, weight corresponding to the i-th peak characteristic, the size of the i-th peak characteristic of the test compound, the size of the i-th peak characteristic in the first or second pending standard compound, similarity of peak characteristics, exp is the exponential function, a difference index of the integrated structural component information for the first or second pending standard compound, The corresponding constant of the first to-be-determined standard compound or the second to-be-determined standard compound; The first to-be-determined standard compound or the second to-be-determined standard compound with the minimum basic information similarity index is taken as the standard compound closest to the to-be-tested compound.

3. The method of claim 2, wherein the method is based on data analysis. Determine the target signal-to-noise ratio of the original mass spectrum based on the mass spectrum of the standard compound, comprising: Obtain the mass spectrum of the standard compound, and determine several representative peaks on the mass spectrum, calculate the area of the representative peaks, and determine the peak vicinity area according to the type of the representative peaks, determine the signal-to-noise ratio of the peak by the area of the peak and the peak vicinity area, integrate the signal-to-noise ratios of all representative peaks to obtain the signal-to-noise ratio of the mass spectrum of the standard compound; A target percentage is obtained by mapping the basic information similarity index of the to-be-tested compound and the standard compound, and the target signal-to-noise ratio of the original mass spectrum is determined according to the target percentage and the signal-to-noise ratio of the mass spectrum of the standard compound.

4. The method for improving the resolution of a mass spectrometer based on data analysis according to claim 1, wherein, Determine the baseline reference area according to the denoised original mass spectrum and the mass spectrum of the standard compound, comprising: Determine the minimum signal value in the starting point vicinity area on the denoised original mass spectrum, and determine the maximum signal value on the denoised original mass spectrum, and determine the extension distance by the ratio of the minimum signal value to the maximum signal value; Take the point of the minimum signal value in the starting point vicinity area as the baseline starting point, extend the baseline starting point according to the extension distance, determine the first baseline area, and determine the second baseline area in the mass spectrum of the standard compound; If the first baseline area and the second baseline area have an intersection, the intersection baseline area is taken as the baseline reference area; Otherwise, the area between the first baseline area and the second baseline area is taken as the baseline reference area.

5. The method for improving the resolution of a mass spectrometer based on data analysis according to claim 1, wherein, And separate the overlapping peaks by peak characteristics, so as to obtain all the peaks, comprising: Standardize all peak characteristics, and use IMDA or ICA to identify and separate overlapping peaks to obtain all single peaks.

6. The method for improving the resolution of a mass spectrometer based on data analysis according to claim 1, wherein, Cluster analysis is performed on all peaks according to the resolution factor, so as to classify similar peaks, comprising: Determine several representative peaks on the mass spectrum, and calculate the area of each representative peak, and determine the minimum distance in the clustering algorithm in combination with the resolution factor; ； wherein L is the minimum distance in the clustering algorithm, for the resolution factor, m is the number of species of representative peaks, Aarea proportion of the jth peak, Aarea for the jth peak, for the initial minimum distance, The initial minimum distance obtained by mapping the area average value is represented as: According to the minimum distance, the similar peaks are classified by clustering analysis.

7. The method of claim 1, wherein the method is based on data analysis. Determine the data dimension of each type of peak, and perform dimension reduction processing on the peak with a data dimension exceeding a dimension threshold, comprising: The data dimension of each type of peak is evaluated, the position and retention time of each type of peak are calculated, a dimension threshold is set by the standard deviation of the position and the retention time, and the dimension of a type of peak exceeding the dimension threshold is reduced.

8. A data analysis based mass spectrometer resolution enhancement system for implementing a data analysis based mass spectrometer resolution enhancement method according to any one of claims 1-7, characterized by The system comprises: A training module for training a resolution factor model by data of a mass analyzer, data of an ion optical system, data of ion transmission, data of a flight path of ions, and data of a collision process, the resolution factor describing the resolution potential of the mass spectrometer under different conditions; A determination module for determining basic information of the to-be-tested compound according to an original mass spectrum and an original fragment spectrum, finding a standard compound closest to the to-be-tested compound in a standard compound library, determining a target signal-to-noise ratio of the original mass spectrum based on a mass spectrum of the standard compound, and performing denoising processing on the original mass spectrum by a denoising method to achieve the target signal-to-noise ratio; A removal module for determining a baseline reference region according to the denoised original mass spectrum and the mass spectrum of the standard compound, and optimizing the baseline reference region by an ALS algorithm to obtain a mass spectrum after removal of baseline drift; A classification module for identifying various features of peaks in the mass spectrum, separating overlapping peaks by the peak features, obtaining all peaks, and performing cluster analysis on the peaks according to the resolution factor to classify similar peaks; A dimension reduction module for judging the data dimension of each type of peak, and reducing the dimension of a type of peak exceeding a dimension threshold, so as to improve the resolution of the mass spectrometer. The original fragment spectrum is a spectrum of ion fragments after activation, the data of the mass analyzer are operation parameters of the mass analyzer, the data of the ion optical system are design parameters of ion mirrors and lenses, the data of ion transmission are ion transmission efficiency, the flight path data of ions are flight path parameters of ions, and the data of the collision process are parameters generated in the ion collision process.

Images