A mass spectrum recognition method and system for liquid chromatography-mass spectrometry

Abstract

The present invention relates to the field of mass spectrum recognition, and more particularly to a mass spectrum recognition method and system for liquid chromatography-mass spectrometry. The method comprises: obtaining a mass spectrum and a mass spectrum baseline, performing preliminary denoising on the mass spectrum using an LMS algorithm to obtain an initial mass spectrum, adjusting an initial step size factor in the LMS algorithm to obtain an optimal step size factor, and further denoising the mass spectrum for use in mass spectrum recognition. The present invention adjusts the initial step size factor using a peak side fluctuation coefficient and a burst noise characteristic coefficient to obtain an optimal step size factor, and finally denoises the mass spectrum using the LMS algorithm, thereby improving the denoising effect of the mass spectrum and facilitating mass spectrum recognition.

Description

Technical Field

[0001] The present invention relates to the field of mass spectrum recognition, and in particular to a mass spectrum recognition method and system for liquid chromatography-mass spectrometry. Background Art

[0002] Liquid chromatography-mass spectrometry (LC-MS / MS) is an analytical instrument that combines liquid chromatography and mass spectrometry. It is used in fields such as chemistry, pharmacy, environmental science, and food safety for both qualitative and quantitative analysis of complex samples. Mass spectrometry, a method for analyzing the composition of substances based on the mass-to-charge ratio of ions, has become a key detection technology in food safety monitoring systems due to its high sensitivity and specificity.

[0003] The Chinese patent application document with publication number CN116500118A discloses a method and device for identifying ion fragment peak areas of Vocs mass spectra based on deep learning. The method includes: obtaining a target Vocs mass spectrum, inputting the target Vocs mass spectrum into a trained target DeepGCMSPeak model, the target DeepGCMSPeak model including a first CNN network and a second CNN network; identifying a target ion fragment peak area from the target Vocs mass spectrum through the first CNN network; inputting the target ion fragment peak area into the second CNN network, and identifying a target area area of ​​the target ion fragment peak area through the second CNN network; and determining the target ion fragment peak area and the target area area as the target recognition result of the target Vocs mass spectrum.

[0004] The accuracy of the test results is affected by the mass spectrum, and the quality of the mass spectrum is affected by many factors during the detection process, which leads to problems such as low signal-to-noise ratio, peak overlap, and decreased resolution in the mass spectrum. When conventional mass spectrometry analysis schemes deal with samples with complex matrix interference, their detection efficiency is often difficult to meet the needs of accurate identification. Therefore, it is necessary to denoise the mass spectrum to improve its quality. Summary of the Invention

[0005] In order to denoise a mass spectrum and improve its quality, the present invention provides a mass spectrum recognition method and system for liquid chromatography-mass spectrometry.

[0006] In a first aspect, the present invention provides a mass spectrum recognition method for liquid chromatography-mass spectrometry, which adopts the following technical solution:

[0007] Obtaining a mass spectrum and a mass spectrum baseline, performing preliminary denoising on the mass spectrum using the LMS algorithm to obtain an initial mass spectrum, adjusting the initial step size factor in the LMS algorithm to obtain an optimal step size factor, and performing denoising on the mass spectrum again for use in mass spectrum identification;

[0008] The optimal step size factor is calculated as follows: peak detection is performed on the initial mass spectrum to obtain multiple peak points, including the first peak point and the second peak point, and whether the first peak point is an abnormal peak point is determined. If so, the average of the absolute value of the difference between the amplitude of the first peak point and the second peak points on both sides is used as the abnormal factor. Otherwise, 0 is used as the abnormal factor, and the sum of the abnormal factors of all the first peak points is used as the peak side fluctuation coefficient.

[0009] The drift degree of the mass spectrometry baseline and the burst noise characteristic coefficient of the mass spectrum are calculated. The burst noise characteristic coefficient is positively correlated with the baseline drift degree. The preset initial step factor is corrected using the ratio of the burst noise characteristic coefficient to the peak side fluctuation coefficient to obtain the optimal step factor.

[0010] The peak side fluctuation coefficient can reflect the oscillation of the LMS algorithm during the denoising process, and the burst noise characteristic coefficient can reflect the characteristics of the sudden pulse signal in the mass spectrum. The peak side fluctuation coefficient and the burst noise characteristic coefficient are used to adjust the initial step factor to obtain the optimal step factor. Finally, the LMS algorithm is used to denoise the mass spectrum, which improves the denoising effect of the mass spectrum and facilitates the identification of the mass spectrum.

[0011] Preferably, the first peak point is obtained by comparing the amplitude of the peak point with the amplitude of the adjacent peak points, and taking the peak point whose amplitude is greater than the amplitudes on the left and right sides as the first peak point.

[0012] Preferably, the method for determining whether the first peak point is an abnormal peak point is: calculating the absolute value of the mass difference between the first peak point and the adjacent first peak point. If the absolute value of the mass difference is within a preset interval, the first peak point is an abnormal peak point; otherwise, the first peak point is a normal peak point.

[0013] Identifying the first peak point can preliminarily determine whether the corresponding first peak point is abnormal, providing a theoretical basis for subsequent adjustment of the step factor.

[0014] Preferably, the method further comprises calculating the sharpness factor of the peak point: obtaining the horizontal coordinates of the troughs on the left and right sides of the peak point, taking the absolute value of the difference between the horizontal coordinates of the troughs on both sides as the eigenvalue, and taking the ratio of the amplitude of the peak point to the eigenvalue as the sharpness factor.

[0015] By calculating the sharp factor of the peak point, the sharp factor reflects the characteristics of the sudden pulse signal in the mass spectrum, which facilitates the calculation of the sudden noise characteristic coefficient.

[0016] Preferably, the burst noise characteristic coefficient is calculated by calculating the range of sharp factors of multiple peak points, and taking the product of the range and the degree of baseline drift as the burst noise characteristic coefficient.

[0017] Preferably, the expression of the optimal step size factor is:

[0018]

[0019] Where, is the optimal step size factor of the LMS algorithm, is the initial step size factor of the LMS algorithm, is the burst noise characteristic coefficient of the mass spectrum, is the peak side fluctuation characteristic coefficient of the mass spectrum, is the logical mapping function.

[0020] The initial step size factor is adjusted using the burst noise characteristic coefficient and the peak side fluctuation characteristic coefficient to obtain the optimal step size factor, which can quickly denoise the mass spectrum and reduce the oscillation of the LMS algorithm during the denoising process.

[0021] Preferably, the method further comprises the step of normalizing the abscissa and the amplitude.

[0022] Preferably, the mass spectrum is peak detected using the AMPD algorithm to obtain multiple peak points.

[0023] Preferably, the position change of the mass spectrum baseline is taken as the drift degree of the mass spectrum baseline.

[0024] The baseline drift can be used to further determine whether there is sudden noise in the mass spectrum, thereby improving the accuracy of the calculation results of the sudden noise characteristic coefficient.

[0025] In a second aspect, the present invention provides a mass spectrum recognition system for liquid chromatography-mass spectrometry, which adopts the following technical solution:

[0026] A mass spectrum recognition system for liquid chromatography-mass spectrometry instrument comprises: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the mass spectrum recognition method for liquid chromatography-mass spectrometry instrument is implemented.

[0027] The mass spectrum recognition method for liquid chromatography-mass spectrometry is generated into a computer program and stored in a memory so as to be loaded and executed by a processor. Thus, a system is produced based on the memory and the processor for easy use.

[0028] The present invention has the following technical effects:

[0029] The peak side fluctuation coefficient can reflect the oscillation of the LMS algorithm during the denoising process, and the burst noise characteristic coefficient can reflect the characteristics of the sudden pulse signal in the mass spectrum. The peak side fluctuation coefficient and the burst noise characteristic coefficient are used to adjust the initial step factor to obtain the optimal step factor. Finally, the LMS algorithm is used to denoise the mass spectrum, which improves the denoising effect of the mass spectrum and facilitates the identification of the mass spectrum. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The present invention is a flow chart of a mass spectrum recognition method for liquid chromatography-mass spectrometry. DETAILED DESCRIPTION

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0032] The embodiment of the present invention discloses a mass spectrum recognition method for liquid chromatography-mass spectrometry, referring to Figure 1 , including the following steps:

[0033] S1: Obtain a mass spectrum and mass spectrum baseline, and use the LMS algorithm to perform preliminary denoising on the mass spectrum to obtain an initial mass spectrum.

[0034] The food to be tested is pretreated, and the solvent is selected according to the polarity of the target object. For example, when detecting the residues of non-compliant ingredients in food, acidified acetonitrile is combined with SPE column purification to remove protein and fat interference. Then, a liquid chromatography-mass spectrometer is used to obtain the mass spectrum of the food to be tested and the mass spectrum baseline is obtained. In the mass spectrum, the horizontal axis of the mass spectrum represents the mass-to-charge ratio in units of m / z, and the vertical axis represents the relative abundance of ions. The LMS algorithm is used to perform preliminary denoising on the mass spectrum to obtain an initial mass spectrum. At this time, the step size factor in the LMS algorithm uses the preset initial step size factor.

[0035] S2: Adjust the initial step size factor in the LMS algorithm to obtain the optimal step size factor.

[0036] The LMS algorithm minimizes the mean square error between the signal and the noise by dynamically adjusting the filter weights, effectively suppressing noise interference. In mass spectra, the noise in the mass spectra is usually random noise introduced by the instrument. The LMS algorithm has a good noise suppression effect, thereby improving the recognition effect of the mass spectra. In the LMS algorithm, when the step size factor is large, the convergence speed is fast, and the optimal solution is quickly approached, which may cause algorithm oscillation, such as fluctuations on both sides of the ion peak in the mass spectrum, resulting in increased residual noise energy. When the step size factor is small, the convergence is slow, and more iterations are required to achieve a stable effect. Otherwise, the sudden noise caused by instrument pulse interference will not be effectively suppressed. Therefore, it is necessary to calculate the optimal step size factor to adapt the LMS algorithm to filter and denoise the mass spectrum.

[0037] S21: Perform peak detection on the initial mass spectrum to obtain multiple peak points.

[0038] The AMPD algorithm is used to perform peak detection on the initial mass spectrum to obtain multiple peak points. The amplitude of the peak point is compared with that of the adjacent peak points. The peak point with an amplitude greater than the amplitudes on the left and right sides is taken as the first peak point, and the remaining peak points are taken as the second peak points. The mass value and amplitude of each peak point are obtained. It can be understood that the mass value is the horizontal coordinate value of the peak point, and the amplitude is the vertical coordinate value of the peak point.

[0039] When the mass spectrum shows abnormal algorithm oscillation during the denoising process, fluctuations will appear on both sides of the ion peak, and the amplitude of the fluctuation is smaller than the amplitude of the ion peak. When the average of the absolute value of the difference between the amplitude of the first peak point and the adjacent second peak point is larger, it indicates that the LSM algorithm is more likely to have abnormal oscillation during the denoising process.

[0040] S22: Determine whether the first peak point is an abnormal peak point. If so, take the average of the absolute values ​​of the differences between the first peak point and the second peak points on its left and right sides as the abnormal factor. Otherwise, take 0 as the abnormal factor.

[0041] The absolute value of the quality difference between the first peak point and the adjacent first peak point is calculated. If the absolute value of the quality difference is within a preset interval, the first peak point is an abnormal peak point; otherwise, the first peak point is a normal peak point.

[0042] The fluctuating peaks in the mass spectrum are related to a certain regularity in mass number. Under normal circumstances, the mass difference between the molecular ion peak and its adjacent molecular ion peaks in the mass spectrum conforms to a certain regularity. For example, the mass difference is 15 (alkyl breakage), 18 (alcohol, carboxylic acid dehydration), and 28 (carbonyl breakage). However, when the mass difference between the molecular ion peak and its adjacent peaks is between 4 and 14, or between 20 and 25, it cannot correspond to typical chemical bond breakage or isotope effect. At this time, when the residual noise energy in the mass spectrum increases, the mass difference between the molecular ion peak and the molecular ion peaks on both sides is within an abnormal range.

[0043] Therefore, when the absolute value of the mass difference between the first peak point and the first peak points adjacent to it on the left and right is in the interval (4, 14) and (20, 25), the corresponding first peak is an abnormal peak point; when the absolute value of the mass difference between the first peak point and the first peak points adjacent to it on the left and right is not in the interval (4, 14) and (20, 25), the corresponding first peak is a normal peak point.

[0044] For example, for the second first peak point, the quality difference with the first first peak point is 5, and the quality difference with the third first peak point is 15, then the second first peak point is an abnormal peak point, that is, if one of the two quality differences falls into the corresponding interval, the corresponding first peak point will be judged as an abnormal peak point.

[0045] If the first peak point is an abnormal peak point, the amplitudes of the first peak point and the second peak point are normalized using the Z-Score algorithm. The absolute value of the difference between the amplitudes of the first peak point and the second peak point on the adjacent left side, as well as the absolute value of the difference between the amplitudes of the first peak point and the second peak point on the adjacent right side are calculated. The average of the two absolute values ​​of the difference is used as the abnormal factor of the corresponding first peak point. Conversely, if the first peak point is a normal peak point, the abnormal factor of the corresponding first peak point is 0.

[0046] The larger the anomaly factor, the greater the possibility of abnormal oscillation in the LMS algorithm during the initial denoising of the mass spectrum using the LMS algorithm. At this time, the step size factor of the LMS algorithm should be reduced to reduce the steady-state error.

[0047] S23: The sum of the abnormal factors of all first peak points is taken as the peak side fluctuation coefficient.

[0048] For the initial mass spectrum, the sum of the anomaly factors of all the first peak points is calculated, and the sum of the anomaly factors is used as the peak side fluctuation coefficient of the mass spectrum.

[0049] S24: Calculate the sharpness factor of the peak point.

[0050] Obtain the horizontal coordinates of the troughs on the left and right sides of the peak point, as well as the vertical coordinate of the peak point. These coordinates are normalized using the Z-Score algorithm. The absolute value of the difference between the horizontal coordinates of the two troughs is used as the eigenvalue, and the ratio of the peak point's amplitude to the eigenvalue is used as the sharpness factor. When a peak point exhibits the characteristics of a sharp signal, the larger the peak point's amplitude and the smaller the absolute value of the difference between the horizontal coordinates of the troughs closest to the peak point, the greater the sharpness factor of the peak point, and the more consistent it is with the characteristics of a burst pulse signal, indicating a greater likelihood that the corresponding peak point is a burst noise.

[0051] S25: Calculate the burst noise characteristic coefficient.

[0052] In one embodiment, the drift degree of the mass spectrum baseline is calculated, the range of sharpness factors of multiple peak points is calculated, and the product of the range and the baseline drift degree is used as the burst noise characteristic coefficient.

[0053] The instrument pulse interference causes sudden noise in the mass spectrum, and at the same time causes the mass spectrum baseline to drift. The drift degree of the mass spectrum baseline is usually within ±10%. The method for calculating the drift degree of the mass spectrum baseline is the existing technology. For example, in the mass spectrum, the mass spectrum baseline changes from 1.0×10 6 counts drifted to 1.2×10 6 counts, the drift of the mass spectrometer baseline is +20%.

[0054] Burst noise appears as pulses, appearing on the mass spectrum as sharp pulses that differ significantly from normal molecular ion peak shapes. Therefore, the burst noise characteristic coefficient comprehensively reflects whether the noise in the mass spectrum meets the characteristics of burst noise. A larger burst noise characteristic coefficient indicates a greater likelihood of burst noise in the mass spectrum. In this case, it is necessary to increase the step size factor of the LMS algorithm to accelerate algorithm convergence.

[0055] S26: using the ratio of the burst noise characteristic coefficient to the peak side fluctuation coefficient to modify the preset initial step length factor to obtain the optimal step length factor.

[0056] For mass spectra with more obvious peak burst noise characteristics, the corresponding burst noise characteristic coefficient is larger. In order to quickly denoise the mass spectrum, the step size factor of the LMS algorithm needs to be increased. The larger the peak side fluctuation characteristic coefficient is, the more fluctuations occur on both sides of the molecular ion peak in the mass spectrum, resulting in increased residual noise energy and algorithm oscillation. In this case, the step size factor of the LMS algorithm should be reduced.

[0057] Therefore, the expression for the optimal step size factor is:

[0058]

[0059] Where, is the optimal step size factor of the LMS algorithm, is the initial step size factor of the LMS algorithm, is the burst noise characteristic coefficient of the mass spectrum, is the peak side fluctuation characteristic coefficient of the mass spectrum, is the logical mapping function.

[0060] In the process of using LMS algorithm to denoise the mass spectrum, the initial step factor is in the interval [0.01, 1]. In this embodiment, the median of the interval 0.505 is selected as the initial step factor of the LMS algorithm. The implementer can select other values ​​in the interval [0.01, 1] as the initial step factor and modify the initial step in the interval based on the peak side fluctuation characteristic coefficient and the burst noise characteristic coefficient. The logical mapping function is to convert The value of is mapped to between 0 and 1, so that the initial step factor remains in the interval [0.01, 1] after correction.

[0061] S3: The mass spectrum is denoised again using the LMS algorithm for identification.

[0062] After obtaining the optimal step size factor, the LMS algorithm is used to denoise the mass spectrum again to obtain a denoised mass spectrum, and then component analysis is performed based on the denoised mass spectrum.

[0063] An embodiment of the present invention also discloses a mass spectrum recognition system for a liquid chromatography-mass spectrometer, comprising a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, a mass spectrum recognition method for a liquid chromatography-mass spectrometer according to the present invention is implemented.

[0064] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0065] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A mass spectrum recognition method for liquid chromatography-mass spectrometry, characterized in that: Including steps: Obtaining a mass spectrum and a mass spectrum baseline, performing preliminary denoising on the mass spectrum using the LMS algorithm to obtain an initial mass spectrum, adjusting the initial step size factor in the LMS algorithm to obtain an optimal step size factor, and performing denoising on the mass spectrum again for use in mass spectrum identification; The optimal step size factor is calculated as follows: peak detection is performed on the initial mass spectrum to obtain multiple peak points, including the first peak point and the second peak point, and whether the first peak point is an abnormal peak point is determined. If so, the amplitudes of the first peak point and the second peak point are normalized using the Z-Score algorithm, and the average of the absolute values ​​of the difference between the amplitudes of the first peak point and the second peak points on both sides is used as the abnormal factor. Otherwise, 0 is used as the abnormal factor, and the sum of the abnormal factors of all first peak points is used as the peak side fluctuation coefficient. The drift degree of the mass spectrometry baseline and the burst noise characteristic coefficient of the mass spectrum are calculated. The burst noise characteristic coefficient is positively correlated with the baseline drift degree. The ratio of the burst noise characteristic coefficient to the peak side fluctuation coefficient is used to correct the preset initial step size factor to obtain the optimal step size factor. The first peak point is obtained by comparing the amplitude of the peak point with the amplitude of the adjacent peak points, and taking the peak point with an amplitude greater than the amplitudes on the left and right sides as the first peak point; The method for determining whether the first peak point is an abnormal peak point is: calculating the absolute value of the mass difference between the first peak point and the adjacent first peak point. If the absolute value of the mass difference is within a preset range, the first peak point is an abnormal peak point; otherwise, the first peak point is a normal peak point.

2. The mass spectrum recognition method for liquid chromatography-mass spectrometry according to claim 1, characterized in that: The method also includes calculating the sharpness factor of the peak point: obtaining the horizontal coordinates of the troughs on the left and right sides of the peak point, taking the absolute value of the difference between the horizontal coordinates of the troughs on both sides as the eigenvalue, and taking the ratio of the amplitude of the peak point to the eigenvalue as the sharpness factor.

3. The mass spectrum recognition method for liquid chromatography-mass spectrometry according to claim 2, characterized in that: The burst noise characteristic coefficient is calculated by calculating the range of sharp factors of multiple peak points and taking the product of the range and the baseline drift as the burst noise characteristic coefficient.

4. The mass spectrum recognition method for liquid chromatography-mass spectrometry according to claim 1, characterized in that: The expression for the optimal step size factor is: Where, is the optimal step size factor of the LMS algorithm, is the initial step size factor of the LMS algorithm, is the burst noise characteristic coefficient of the mass spectrum, is the peak side fluctuation characteristic coefficient of the mass spectrum, is the logical mapping function.

5. The mass spectrum recognition method for liquid chromatography-mass spectrometry according to claim 2, characterized in that: The method further comprises the step of normalizing the abscissa and the amplitude.

6. The mass spectrum recognition method for liquid chromatography-mass spectrometry according to claim 1, characterized in that: The AMPD algorithm was used to perform peak detection on the mass spectrum to obtain multiple peak points.

7. The mass spectrum recognition method for liquid chromatography-mass spectrometry according to claim 1, characterized in that: The position change of the mass spectrum baseline is regarded as the drift degree of the mass spectrum baseline.

8. A mass spectrum recognition system for liquid chromatography-mass spectrometry, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a mass spectrum recognition method for liquid chromatography-mass spectrometry according to any one of claims 1 to 7 is implemented.

Images