Method and system for rapidly identifying and analyzing traditional Chinese medicine compound components

Abstract

The invention relates to the technical field of traditional Chinese medicine complex system component analysis, in particular to a rapid identification and analysis method and system for traditional Chinese medicine compound components. Performing step-by-step expansion on the spectral signals to construct a multi-level feature structure; meanwhile, trajectory tracking is conducted on mass spectrum fragments, and a fracture path network is established. And establishing an interactive verification channel between the two to transmit a matching degree parameter, and screening candidate compounds according to the matching degree parameter. And calculating the characteristic contribution degree of each candidate in the spectrum structure and the path integrity of each candidate in the mass spectrum network, merging to generate a confidence index, and finally completing component identification. According to the method, through deep coupling of spectrum multi-scale decomposition and a mass spectrum fracture network, cross validation of multi-dimensional information is achieved, and the accuracy and reliability of component identification of a complex traditional Chinese medicine system are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of traditional Chinese medicine complex system component analysis, and in particular to a traditional Chinese medicine compound component rapid identification analysis method and system. BACKGROUND

[0002] Traditional Chinese medicine compound components are complex, and various chemical components in the system interfere with each other, bringing great challenges to rapid and accurate identification. The current mainstream method mainly relies on chromatography and mass spectrometry combined technology, and the identification is realized by comparing the mass spectrum fragment information of the sample with the standard compound spectrum library. However, in the face of complex matrix, the traditional method has limited processing capacity for the interference of spectral baseline drift, peak overlap and other interferences caused by coexisting components, and usually relies on simple pretreatment or intuitive characteristic peak matching, which is difficult to fully excavate and utilize the fine chemical information contained in high-dimensional spectral data. This leads to insufficient discrimination of trace components or structurally similar compounds, and the identification results are easily affected by background noise and co-elution.

[0003] In terms of mass spectrometry data analysis, the existing technology focuses on the mass-to-charge ratio matching of fragment ions or the sequence comparison of limited secondary mass spectrometry. This method regards each detected fragment as an isolated point, and lacks systematic modeling of the fragmentation logic and derivation relationship between fragment ions. When facing unknown compounds or incomplete standard spectrum library information, only relying on the matching of fragment list cannot effectively verify the rationality of the identification result, and false positives or the integrity of the fragmentation path cannot be confirmed, and the discrimination ability for isomers or compounds with similar fragmentation rules is weak. SUMMARY

[0004] The purpose of the present application is to solve the problems existing in the prior art, and to provide a traditional Chinese medicine compound component rapid identification analysis method and system.

[0005] In order to achieve the above purpose, the technical scheme adopted by the present application is as follows: a traditional Chinese medicine compound component rapid identification analysis method, comprising:

[0006] Performing a composite scan on the sample to be tested to collect the original spectral response and mass spectrometry fragment sequence of the sample to be tested;

[0007] Expanding the original spectral response step by step to generate a spectral hierarchical structure containing multiple decomposition levels, each decomposition level corresponding to a spectral feature mode;

[0008] Trajectory tracking of the mass spectrometry fragment sequence to establish a mass spectrometry fragmentation network from the initial ion to the terminal fragment;

[0009] Establishing an interactive verification channel between the spectral hierarchical structure and the mass spectrometry fragmentation network, and transmitting the matching degree parameters between the spectral feature modes and the fragmentation nodes through the interactive verification channel;

[0010] According to the matching degree parameter, a candidate compound list is screened out from the standard compound spectrum;

[0011] For each member in the candidate compound list, its feature contribution degree in the spectrum hierarchy and its path completeness degree in the mass spectrum fragmentation network are calculated;

[0012] The feature contribution degree and the path completeness degree are merged to generate a confidence index of each candidate member, and a component identification report of the sample to be tested is completed according to the confidence indexes of all candidate members.

[0013] Preferably, the original spectrum response is unfolded step by step to generate a spectrum hierarchy containing multiple decomposition levels, including:

[0014] A set of initial spectrum waveforms is extracted from the original spectrum response;

[0015] A modal decoupling operation is applied to each initial spectrum waveform to separate each initial spectrum waveform into a trend component, a detail component and a noise component;

[0016] The trend components from different initial spectrum waveforms are aggregated to form a first decomposition level of the spectrum hierarchy, which reflects the overall drift characteristics of the spectrum;

[0017] The detail components from different initial spectrum waveforms are classified, and the detail components with similar frequency and amplitude characteristics are classified into the same group to form a second decomposition level of the spectrum hierarchy, which reflects the vibration characteristics of specific functional groups;

[0018] The noise components are statistically analyzed to generate a noise base model, and the noise base model is attached to the spectrum hierarchy as a background reference level of the spectrum hierarchy.

[0019] Preferably, the mass spectrum fragment sequence is tracked to establish a mass spectrum fragmentation network from the initial ion to the terminal fragment, including:

[0020] The mass spectrum fragment sequence is analyzed to identify all mass spectrum peak points with mass-to-charge ratio and intensity information;

[0021] The mass-to-charge ratio of the mass spectrum peak point is taken as a node, and the possible fragmentation relationship inferred from the difference between the mass-to-charge ratios is taken as a directed edge to construct an initial fragmentation atlas;

[0022] Each directed edge in the initial fragmentation atlas is assigned a fragmentation possibility weight, which is calculated based on fragmentation chemical rules and intensity ratios of adjacent mass spectrum peak points;

[0023] In the initial fragmentation graph, starting from the molecular ion peak node, a depth-first search is performed to all possible sub-fragment nodes, and all fragmentation paths reaching the terminal fragment nodes are recorded;

[0024] All fragmentation paths and the nodes and directed edges on the paths are integrated to form a mass spectrum fragmentation network containing multiple fragmentation levels and branches, and each path in the mass spectrum fragmentation network represents a possible molecular cleavage process.

[0025] Preferably, the interactive verification channel between the spectral hierarchy and the mass spectrum fragmentation network includes:

[0026] From the second decomposition level of the spectral hierarchy, the feature frequency and feature intensity data of a group of detail components are extracted;

[0027] From the mass spectrum fragmentation network, the mass-to-charge ratio data of a group of key fragmentation nodes and the fragmentation possibility weight of the directed edges connecting the key fragmentation nodes are extracted;

[0028] A mapping relationship table between the feature frequency and the feature mass-to-charge ratio is established, which defines the typical molecular fragment mass interval corresponding to a specific functional group vibration frequency;

[0029] According to the mapping relationship table, each detail component in the spectral hierarchy and each key fragmentation node in the mass spectrum fragmentation network are traversed, and a spectral mass spectrum matching score of each detail component and the mass spectrum fragmentation network is calculated;

[0030] The spectral mass spectrum matching score and the corresponding fragmentation possibility weight are weighted and fused to generate the matching degree parameter, and the matching degree parameter is transmitted back to the spectral hierarchy and the mass spectrum fragmentation network to mark the spectral feature mode and the fragmentation node with high matching degree.

[0031] Preferably, the candidate compound list is screened from the standard compound spectrum according to the matching degree parameter, including:

[0032] Receive multiple matching degree parameters from the interactive verification channel;

[0033] Retrieve all known compounds from the standard spectrum library that have an intersection with the matching degree parameter in the feature frequency and mass-to-charge ratio interval to form a preliminary candidate set;

[0034] For each known compound in the preliminary candidate set, its standard spectrum is compared layer by layer with the spectral hierarchy to calculate the spectral compliance degree;

[0035] For each known compound in the preliminary candidate set, its standard mass spectrum is compared with the mass spectrum fragmentation network to calculate the mass spectrum compliance degree;

[0036] Triply operating the spectrum coincidence, the mass spectrum coincidence and the corresponding matching degree parameter to generate a comprehensive screening score of each known compound;

[0037] According to a preset screening score threshold, selecting known compounds with a comprehensive screening score exceeding the threshold from the preliminary candidate set to form the candidate compound list.

[0038] Preferably, for each member in the candidate compound list, a characteristic contribution degree thereof in the spectrum hierarchical structure is calculated, including:

[0039] Obtaining a reference spectrum feature pattern of the candidate compound from a standard spectrum library;

[0040] Projecting the reference spectrum feature pattern onto a second decomposition level of the spectrum hierarchical structure of the sample to be tested to identify a detail component explained by the candidate compound;

[0041] Accumulating intensity values of all explained detail components to obtain a total explained signal intensity of the candidate compound in the sample to be tested;

[0042] Calculating a proportion of the total explained signal intensity in total intensities of all detail components of the sample to be tested, and defining the proportion as the characteristic contribution degree of the candidate compound in the spectrum hierarchical structure.

[0043] Preferably, for each member in the candidate compound list, a path completeness degree thereof in the mass spectrum fragmentation network is calculated, including:

[0044] Obtaining a reference mass spectrum fragmentation path of the candidate compound from a standard spectrum library;

[0045] Finding a matching path in the mass spectrum fragmentation network of the sample to be tested that is most similar to the reference mass spectrum fragmentation path in node sequence and fragmentation order;

[0046] Comparing the matching path and the reference mass spectrum fragmentation path to count coincidence degrees of shared fragmentation nodes and similarity degrees of fragmentation possibility weights of the matching path and the reference mass spectrum fragmentation path;

[0047] Combining the coincidence degrees of shared fragmentation nodes and the similarity degrees of fragmentation possibility weights to calculate a path similarity coefficient;

[0048] Multiplying the path similarity coefficient and a coverage rate of the matching path in the mass spectrum fragmentation network to obtain the path completeness degree of the candidate compound in the mass spectrum fragmentation network.

[0049] Preferably, the characteristic contribution degree and the path completeness degree are combined to generate a confidence index of each candidate member, including:

[0050] The characteristic contribution degree of the candidate compound is normalized so that its value range is mapped to a preset uniform interval;

[0051] The path integrity of the candidate compound is normalized so that its value range is mapped to the same preset uniform interval;

[0052] A spectral weight factor is assigned to the normalized characteristic contribution degree, and a mass spectrum weight factor is assigned to the normalized path integrity, and the sum of the spectral weight factor and the mass spectrum weight factor is a fixed value;

[0053] The weighted normalized characteristic contribution degree and the weighted normalized path integrity are added to obtain a preliminary confidence value of the candidate compound;

[0054] A matching degree parameter related to the candidate compound from an interactive verification channel is introduced to correct the preliminary confidence value to generate a final confidence index.

[0055] Preferably, after generating the component identification report of the sample to be tested, the method further comprises:

[0056] Verification data is collected from a new batch of traditional Chinese medicine compound samples;

[0057] The compound information predicted to exist in the component identification report is checked one by one with the compound information confirmed by the control sample in the verification data;

[0058] The number of correctly predicted compounds and the number of incorrectly predicted compounds are counted to calculate the accuracy index of the identification process;

[0059] Based on the accuracy index, at least one of the following parameters is adjusted in reverse: the depth of the spectral hierarchical structure expansion, the calculation rule of the fragmentation possibility weight in the mass spectrum fragmentation network, and the generation mode of the matching degree parameter;

[0060] The next batch of traditional Chinese medicine compound samples is subjected to component identification analysis using the adjusted parameters.

[0061] Preferably, the present application further comprises a traditional Chinese medicine compound component rapid identification analysis system, which comprises a processor and a memory, the memory and the processor are connected, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to realize the traditional Chinese medicine compound component rapid identification analysis method as described above.

[0062] Compared with the prior art, the present application has the following advantages and positive effects:

[0063] The original spectral signal is unfolded step by step and a multi-level structure is generated, realizing separation and structural analysis of different scale feature patterns in the mixed spectrum. The technology can effectively peel off the background baseline, wide peak signal and sharp feature peak and place them in independent decomposition levels. This processing reduces the covering of noise and strong signal components to trace amounts of characteristic information, so that the spectral matching is based on the signal level with more chemical specificity, thereby improving the recognition ability of low-abundance components and structurally similar compounds in complex matrix.

[0064] The trajectory of the mass spectrum fragment sequence is tracked and a fragmentation network is constructed, connecting discrete fragment ions into a path graph with logical relationships according to the fragmentation rules. The network integrates information such as mass difference and ion intensity, and systematically represents the potential fragmentation process of the compound. The path completeness based on network calculation provides a dynamic verification basis for identification results beyond fragment list matching, which can effectively identify and exclude false positive candidates with unreasonable or unbuildable fragmentation paths, and enhance the reliability of the results and the inference ability of unknown fragmentation behavior.

[0065] An interactive verification channel is established between the spectral hierarchical structure and the mass spectrum fragmentation network, and the matching degree parameters are transmitted, realizing the deep fusion and cross verification of the two types of heterogeneous information. The channel forms a two-way constraint between the spectral feature pattern and the mass spectrum fragmentation node, and the candidate components need to have characteristic contribution at a specific level in the spectrum and reasonable fragmentation path support in the mass spectrum dimension. This collaborative verification mechanism builds a judgment system with mutual verification of multi-dimensional evidence, improving the robustness of the overall identification process and the reliability of analyzing complex samples. BRIEF DESCRIPTION OF DRAWINGS

[0066] Figure 1 A flowchart of the rapid identification and analysis method of the traditional Chinese medicine compound described in the present application;

[0067] Figure 2 A flowchart for the mass spectrum fragmentation network;

[0068] Figure 3 A flowchart for the interactive verification channel;

[0069] Figure 4 A composition analysis diagram for the candidate compound confidence index;

[0070] Figure 5 A distribution diagram of the spectral and mass spectrum coincidence and matching degree of the candidate compound. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application is further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the present application.

[0072] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0073] See Figure 1 The overall implementation of a rapid identification and analysis method for components of traditional Chinese medicine compound formulas includes a series of sequential and interactive data processing stages. First, a composite scan is performed on the sample to acquire the raw spectral response and mass spectrometry fragment sequences. Then, the raw spectral response is progressively expanded to generate a spectral hierarchy structure containing multiple decomposition levels, each corresponding to a spectral feature mode. Simultaneously, the mass spectrometry fragment sequences are tracked to establish a mass spectrometry fragmentation network from initial ions to terminal fragments. An interactive verification channel is established between the spectral hierarchy structure and the mass spectrometry fragmentation network to transmit the matching degree parameter between the spectral feature modes and fragmentation nodes. Based on the matching degree parameter, a candidate compound list is selected from a standard compound library. For each member in the candidate compound list, its feature contribution in the spectral hierarchy structure and its path integrity in the mass spectrometry fragmentation network are calculated. The feature contribution and path integrity are combined to generate a confidence index for each candidate member, and a component identification report for the sample is completed based on the confidence indices of all candidate members. This scheme improves the reliability and efficiency of identification through multi-dimensional data fusion and verification.

[0074] In one embodiment of the present invention, see [reference] Figure 2In practice, after acquiring the original spectral response of the sample to be tested, the original spectral response is expanded stepwise to generate a spectral hierarchy structure. A set of initial spectral waveforms is extracted from the original spectral response, and modal decoupling is applied to each initial spectral waveform to separate it into trend components, detail components, and noise components. In practice, the trend components from different initial spectral waveforms are aggregated to form the first decomposition level of the spectral hierarchy structure, which reflects the overall drift characteristics of the spectrum. The detail components from different initial spectral waveforms are categorized, and detail components with similar frequency and amplitude characteristics are grouped together to form the second decomposition level of the spectral hierarchy structure, which reflects the vibrational characteristics of specific functional groups. In practice, the noise components are statistically analyzed to generate a noise basis model, which is then attached to the spectral hierarchy structure as a background reference level. In some embodiments, the modal decoupling operation employs an empirical mode decomposition algorithm, which decomposes the initial spectral waveform into intrinsic mode functions (EMFs) arranged from high to low frequencies through an iterative screening process. The trend component corresponds to the lowest frequency EMF, the detail component corresponds to the intermediate frequency EMF, and the noise component corresponds to the highest frequency EMF. Optionally, the classification of detail components uses a hierarchical clustering method based on Euclidean distance, using frequency and amplitude features as clustering dimensions. Detail components with similarity exceeding a preset threshold are merged into the same feature group, where each feature group represents a potential functional group vibration mode. It can be understood that the statistical analysis of the noise basis model includes calculating the root mean square error and probability distribution of the noise components to quantify the fluctuation range and statistical characteristics of the background noise. The noise basis model serves as a reference baseline in the spectral hierarchy to distinguish between real signals and random interference.

[0075] In specific implementation, trajectory tracking is performed on the mass spectrometry fragment sequences to establish a mass spectrometry fragmentation network. The mass spectrometry fragment sequences are analyzed to identify all mass spectrometry peaks with mass-to-charge ratio and intensity information. Using the mass-to-charge ratio of the mass spectrometry peaks as nodes and the possible fragmentation relationships inferred from the differences in mass-to-charge ratios as directed edges, an initial fragmentation map is constructed. In specific implementation, each directed edge in the initial fragmentation map is assigned a fragmentation probability weight, calculated based on the fragmentation chemistry rules and the intensity ratio of adjacent mass spectrometry peaks. The fragmentation chemistry rules include the mass defect rules of common chemical bond breakage and the fragment stability principle. In specific implementation, starting from the molecular ion peak node in the initial fragmentation map, a depth-first search is performed to all possible sub-fragment nodes, recording all fragmentation paths that can reach the terminal fragment node. All fragmentation paths and the nodes on those paths are integrated with the directed edges to form a mass spectrometry fragmentation network containing multiple fragmentation levels and branches. Each path in the mass spectrometry fragmentation network represents a possible molecular fragmentation process. In some embodiments, the calculation of the fragmentation probability weight uses a weighted summation formula:

[0076]

[0077] in: This represents the weight of the probability of breakage from mass spectrum peak a to mass spectrum peak b. and Let represent the intensity values ​​of mass spectrum peak a and mass spectrum peak b, respectively. This represents a score based on the rules of chemical breakage. The score is assigned according to the chemical bond type and breakage probability corresponding to the mass-to-charge ratio difference. and Pre-defined adjustment coefficients are used to balance the influence of intensity ratio and chemical rules. Optionally, the depth-first search employs a recursive algorithm, starting from the molecular ion peak node and sequentially exploring each possible sub-fragment node until a terminal fragment node with a mass-to-charge ratio below a preset minimum value or a node where further fragmentation is impossible is reached. During the search, the node sequence and directed edge sequence on each path are recorded. It can be understood that the integration of the mass spectrometry fragmentation network involves merging all recorded fragmentation paths into a directed acyclic graph structure. In the graph, nodes represent mass spectrometry peaks, directed edges represent fragmentation relationships, fragmentation probability weights are stored as edge attributes, and branch points in the network correspond to multiple possible fragmentation directions, thus comprehensively describing the possibility space of molecular fragmentation.

[0078] In one embodiment of the present invention, see [reference] Figure 3 In the specific implementation, when establishing an interactive verification channel between the spectral hierarchy structure and the mass spectrometry fracture network, a set of characteristic frequency and characteristic intensity data of detailed components are extracted from the second decomposition level of the spectral hierarchy structure. The characteristic frequencies are expressed as wavenumbers, and the characteristic intensities are expressed as relative absorbance. In the specific implementation, a set of mass-to-charge ratio data of key fracture nodes and the fracture probability weights of the directed edges connecting the key fracture nodes are extracted from the mass spectrometry fracture network. Key fracture nodes are defined as nodes corresponding to molecular ion peaks, major fragment peaks, and nodes located at the intersection of multiple fracture paths in the mass spectrometry fracture network. In the specific implementation, a mapping table between characteristic frequencies and characteristic mass-to-charge ratios is established. The mapping table defines the typical molecular fragment mass range corresponding to the vibrational frequencies of specific functional groups. The mapping table is pre-constructed based on an organic compound structural spectroscopy database, where the stretching vibration frequency range of carboxyl groups corresponds to the fragment mass range of fragments without carboxyl groups, and the skeletal vibration frequency range of aromatic rings corresponds to the fragment mass range containing aromatic rings.

[0079] In some embodiments, when traversing each detail component in the spectral hierarchy and each key break node in the mass spectrometry break network according to the mapping table and calculating the spectral-mass spectrometry matching score of each detail component and the mass spectrometry break network, the calculation process is as follows: For a given detail component, it has a characteristic frequency With characteristic strength For a given critical fracture node, it has a mass-to-charge ratio. and the set of breakage probability weights for several directed edges connected to it. In practice, the characteristic frequencies are found by querying the mapping table. The corresponding one or more mass-to-charge ratio intervals [R1, R2, ...]. If the mass-to-charge ratio of the critical fracture node... Falling into any of the characteristic frequencies Within the associated mass-to-charge ratio range, a potential match is considered to exist. For each potential match pair, a spectral-mass spectrometry match score is calculated using the following formula:

[0080]

[0081] Where: characters This represents the calculated spectral-mass spectrometry matching score; character Indicates the feature intensity of the current detail component; character This represents the maximum feature intensity of all detail components extracted from the second decomposition level of the spectral hierarchy, used to evaluate the feature intensity. Normalize; characters Indicates the mass-to-charge ratio of the current critical fracture node; character Indicates the center value of the current mass-to-charge ratio interval R; character Represents the absolute difference between the current mass-to-charge ratio and the center value of the interval; character It is a preset scaling parameter used to control the decay rate of the impact of mass-to-charge ratio deviation on the score; character This represents the arithmetic mean of the break probability weights of all directed edges connected to the current critical break node. The average break probability weight reflects the connectivity importance of this node in the mass spectrometry break network. It can be understood that the exponential term... This term is used to quantify the tightness of the mass-to-charge ratio match. It reaches its maximum value when the mass-to-charge ratio is exactly equal to the center value of the interval, and decreases as the deviation increases.

[0082] Optionally, when weighting and fusing the spectral mass spectrometry matching score with the corresponding fracture probability weight to generate the matching degree parameter, for each detail component and key fracture node pair with a potential match, the corresponding fracture probability weight is selected as the fracture probability weight value of the directed edge connecting the key fracture node and its direct predecessor node on the matching path. In practice, the weighted fusion operation uses a linear combination method, with the matching degree parameter as follows:

[0083]

[0084] in: This indicates the generated matching degree parameter. and These are preset positive coefficients used to adjust the contribution ratio of the spectral mass spectrometry matching score and the breakage probability weight in the matching degree parameter. The coefficients satisfy... The relationship is as follows. In some embodiments, when the matching degree parameter is passed back to the spectral hierarchy structure and the mass spectrometry fracture network, the specific operation is to add an attribute field to the detail component in the second decomposition level of the spectral hierarchy structure in the data structure to store the highest matching degree parameter value associated with it; at the same time, the highest matching degree parameter value associated with it is recorded in the attribute field of the key fracture node in the mass spectrometry fracture network. Optionally, high-matching spectral feature patterns and fracture nodes are marked by setting a matching degree threshold. This is to achieve the following: when the matching degree parameter value associated with a detail component or a key break node is greater than or equal to the matching degree threshold. At that time, the detailed component is marked as a highly matched spectral feature mode, and the key break point is marked as a highly matched break point. It can be understood that the marking operation enables subsequent processing steps to quickly locate and focus on highly correlated parts of the spectral hierarchy and mass spectrometry break network.

[0085] In one embodiment of the present invention, when selecting a list of candidate compounds from standard compound spectra based on matching degree parameters, multiple matching degree parameters are received from the interactive verification channel. Each matching degree parameter is associated with a set of spectral characteristic frequencies, a mass-to-charge ratio range, and a matching degree value characterizing the correlation strength between the spectrum and the mass spectrum. In this embodiment, all known compounds that intersect with the matching degree parameters in terms of characteristic frequencies and mass-to-charge ratio ranges are retrieved from the standard spectral library to form a preliminary candidate set. The retrieval process involves comparing the characteristic absorption peak frequency range and mass spectrometry fragment mass range registered in the standard spectral library for each known compound with the spectral characteristic frequencies and mass-to-charge ratio ranges covered by the received matching degree parameters one by one. If any overlap exists, the known compound is included in the preliminary candidate set. In some embodiments, when calculating the spectral consistency for each known compound in the preliminary candidate set, a standard spectrum of the known compound is obtained. The standard spectrum contains a series of standard characteristic peaks and their standard intensity information. The standard spectrum is compared layer by layer with the spectral hierarchy of the sample to be tested. The comparison process focuses on the detail component groups in the second decomposition level of the spectral hierarchy. The spectral consistency is calculated by traversing all standard characteristic peaks in the standard spectrum, finding the detail component with the closest frequency in the spectral hierarchy, and comprehensively evaluating the frequency shift and intensity match of the matching peak pairs. In a specific implementation, when calculating the mass spectrometry consistency for each known compound in the preliminary candidate set, a standard mass spectrum of the known compound is obtained. The standard mass spectrum contains the standard mass-to-charge ratio and standard relative abundance information of the main fragment ions. The standard mass spectrum is compared with the mass spectrometry fracture network. The path comparison starts from the molecular ion peak and checks whether the main fragment ion sequence in the standard mass spectrum can be found as a corresponding node sequence in the mass spectrometry fracture network. The calculation of the mass spectrometry consistency involves counting the number of matching nodes, comparing the relative abundance ratio between matching nodes, and evaluating the logical rationality of the fracture path. Optionally, when performing a ternary operation on the spectral fit, mass spectrometry fit, and corresponding matching parameters to generate a comprehensive screening score for each known compound, the operation uses a weighted geometric mean. The comprehensive screening score is as follows:

[0086]

[0087] Where: characters Indicates the generated composite screening score; character Indicates the calculated spectral compliance; character Indicates the calculated mass spectrometry agreement; character This represents the average of all relevant matching parameters for the known compound; character , and These are weight indices pre-assigned to the parameters of spectral matching, mass spectrometry matching, and average matching, and all are positive numbers; This is used to normalize the weighted product to a reasonable scale range. It is understood that the weighted geometric mean requires high values ​​for all three indicators: spectral conformity, mass spectrometry conformity, and average matching parameter. A low value for any one indicator will significantly lower the overall screening score. In some embodiments, known compounds with overall screening scores exceeding a preset screening score threshold are selected from the preliminary candidate set to form a candidate compound list. The screening score threshold is a pre-set value based on the data quality of the standard spectral library and the rigor of the identification task. Optionally, the preset screening score threshold is an empirical value determined by analyzing historical validation data, used to balance the completeness and accuracy of the candidate list. In specific implementations, only known compounds with an overall screening score greater than or equal to this threshold are included in the final candidate compound list. It is understood that the entire screening process, by integrating spectral conformity, mass spectrometry conformity, and the matching parameter generated from previous interactive validation, achieves a comprehensive evaluation of multi-dimensional evidence, thereby improving the reliability of the candidate compound list.

[0088] In one embodiment of the present invention, when calculating the feature contribution of each member in the spectral hierarchy of the candidate compound list, a reference spectral feature mode of the candidate compound is obtained from a standard spectral library. The reference spectral feature mode is stored in list form, containing the known characteristic absorption peak frequency values ​​of the compound and their corresponding reference intensities. In this embodiment, the reference spectral feature mode is projected onto the second decomposition level of the spectral hierarchy of the sample to be tested to identify the detail components explained by the candidate compound. The projection process involves searching among all detail components in the second decomposition level for detail components whose difference between their characteristic frequencies and a certain characteristic absorption peak frequency value in the reference spectral feature mode is within a preset tolerance range. These searched detail components are then marked as detail components explained by the candidate compound. In some embodiments, the intensity values ​​of all explained detail components are accumulated to obtain the total explained signal intensity of the candidate compound in the sample to be tested. The intensity value uses the characteristic intensity data recorded by the detail components in the spectral hierarchy. In this embodiment, the proportion of the total explained signal intensity to the total intensity of all detail components in the sample to be tested is calculated, and this proportion is defined as the feature contribution of the candidate compound in the spectral hierarchy. The formula for calculating the feature contribution is:

[0089]

[0090] Where: characters Represents the calculated feature contribution; character This represents summing the feature intensities of all interpreted detail components. The number of detail components being interpreted; characters This represents the summation of the feature intensities of all detail components in the second decomposition level of the spectral hierarchy, where N is the total number of detail components. It can be understood that the feature contribution quantifies the overall interpretability of the candidate compound's reference spectral feature mode for the measured spectral signal; the closer the feature contribution is to 1, the more of the spectral signal the candidate compound can interpret. For a clear illustration of the intensity accumulation process, refer to Table 1, which presents a simplified example.

[0091] Table 1: Example Data Table for Feature Contribution Calculation

[0092] Detail component number Feature frequency (cm"1) Feature intensity Whether explained by compound A 1 1725 0.85 Yes 2 1610 0.92 No 3 1510 0.78 Yes 4 1450 0.65 No 5 1270 0.58 Yes

[0093] In practice, when calculating the path integrity of each member in the mass spectrometry fragmentation network for each member of the candidate compound list, a reference mass spectrometry fragmentation path for the candidate compound is obtained from a standard spectral library. This reference path is represented by an ordered sequence of nodes, such as [M]⁺→[M-H₂O]⁺→[M-H₂O-CO]⁺, and includes the reference mass-to-charge ratio for each node. In the mass spectrometry fragmentation network of the sample to be tested, a matching path is sought that is most similar to the reference path in terms of node sequence and fragmentation order. This search process begins with the molecular ion peak node in the mass spectrometry fragmentation network and follows the fragmentation order of the reference path, searching for subsequent nodes in the network whose mass-to-charge ratio matches that of the reference node, thus forming a matching path. Optionally, the matching path is compared with the reference mass spectrometry fracture path, and the degree of conformity between the matching path and the reference mass spectrometry fracture path at the common fracture nodes and the similarity of the fracture probability weights are statistically analyzed. The degree of conformity is determined by calculating the relative error between the mass-to-charge ratio of the matching node and the mass-to-charge ratio of the reference node, and the similarity of the fracture probability weights is determined by calculating the correlation coefficient between the fracture probability weight of the corresponding directed edge on the matching path and the reference standard weight.

[0094] In some embodiments, the degree of similarity between shared fracture nodes and the degree of similarity in fracture probability weights are combined to calculate the path similarity coefficient. The calculation integrates node matching accuracy with the consistency of fracture logic. In specific implementation, the path similarity coefficient is multiplied by the coverage of the matched path in the mass spectrometry fracture network to obtain the path integrity of the candidate compound in the mass spectrometry fracture network. The coverage of the matched path is defined as the proportion of the number of nodes contained in the matched path to the total number of all key fracture nodes in the entire mass spectrometry fracture network. It can be understood that the path integrity reflects both the degree of fit between the fracture path of the candidate compound and the measured network and the representativeness of the path in the network. The value range of path integrity is between 0 and 1.

[0095] See Figure 4This is a confidence index composition analysis chart for candidate compounds, showing the confidence index composition of five candidate compounds (baicalin, puerarin, glycyrrhizic acid, rutin, and ferulic acid). Orange represents the "weighted feature contribution" (spectral dimension), and purple represents the "weighted path integrity" (mass spectrometry dimension). The final confidence indices for baicalin, puerarin, and glycyrrhizic acid are all approximately 0.6, indicating that the evidence for their presence in traditional Chinese medicine compound formulas is the strongest among these three compounds. However, their compositions differ significantly. Rutin has the lowest weighted feature contribution and path integrity, indicating weaker matching evidence in both spectral and mass spectrometry dimensions, suggesting it may be a compound with low content or significant interference. The confidence indices of different compounds rely on different dimensions of support: glycyrrhizic acid relies on spectral feature contribution, and puerarin relies on mass spectrometry path integrity. This reflects the complementarity of the "spectral-mass spectrometry" dual-dimensional verification, consistent with the core logic of component identification in traditional Chinese medicine compound formulas.

[0096] In one embodiment of the present invention, when merging feature contribution and path integrity to generate a confidence index for each candidate member, the feature contribution of the candidate compound is normalized to map its value range to a preset unified interval. The normalization process employs a linear transformation method to scale the original feature contribution values ​​proportionally to a target interval, such as [0,1]. The specific transformation is calculated based on the maximum and minimum values ​​of the feature contributions of all candidate compounds. In another embodiment, the path integrity of the candidate compound is normalized to map its value range to the same preset unified interval. The normalization of path integrity uses the same linear transformation method and target interval as the feature contribution to ensure that the two metrics are on comparable orders of magnitude. In some embodiments, a spectral weighting factor is assigned to the normalized feature contribution, and a mass spectrometry weighting factor is assigned to the normalized path integrity. The sum of the spectral weighting factor and the mass spectrometry weighting factor is a fixed value, and the specific values ​​of the spectral weighting factor and the mass spectrometry weighting factor are preset based on the relative quality or importance of the spectral data and the mass spectrometry data. In practice, the weighted normalized feature contribution and the weighted normalized path completeness are added together to obtain the preliminary confidence value of the candidate compound. The addition operation is a simple linear summation. Optionally, a matching degree parameter related to the candidate compound from the interactive validation channel is introduced to correct the preliminary confidence value and generate the final confidence index. The correction process incorporates the matching degree parameter as an adjustment factor into the calculation. The formula for calculating the final confidence index is:

[0097]

[0098] Where: characters Indicates the final generated confidence index; character Indicates the pre-assigned spectral weighting factor; character Indicates the normalized feature contribution; character Indicates pre-assigned mass spectrometry weighting factors; character Indicates the normalized path completeness; character Represents the average of all matching parameters related to the candidate compound from the interactive validation channel; weighted sum. This is the initial confidence value; logarithmic terms Used to introduce a matching degree parameter for correction, when the average matching degree parameter A higher value corresponds to a larger correction factor, thus improving the final confidence index. This formula combines the contribution of spectral features, the integrity of the mass spectrometry path, and the results of previous spectral-mass spectrometry cross-validation to jointly determine the final confidence index.

[0099] In specific implementation, after generating the component identification report for the sample to be tested, verification data is collected from a new batch of traditional Chinese medicine compound samples. The verification data includes information on compounds confirmed by reference standards and their corresponding spectral and mass spectrometric responses. In specific implementation, the compound information predicted in the component identification report is compared one by one with the compound information confirmed by reference standards in the verification data. The comparison process involves precisely matching compound identifiers and recording whether the prediction results match the verification results. In some embodiments, the number of correctly predicted compounds and the number of incorrectly predicted compounds are counted to calculate the accuracy index of this identification process. The accuracy index is the proportion of correctly predicted compounds to the total number of predicted compounds. Optionally, the depth of the spectral hierarchy structure expansion is adjusted inversely based on the accuracy index. When the accuracy index is lower than the expected threshold, the number of iterations of the modal decoupling operation is increased to obtain a deeper spectral hierarchy structure. In specific implementation, the calculation rules for the fracture probability weights in the mass spectrometry fracture network are adjusted inversely based on the accuracy index. The adjustment operation involves modifying the weighting coefficients of the intensity ratio and the fracture chemical rule score in the formula for calculating the fracture probability weights. It is understandable that adjusting the matching degree parameters based on the accuracy index involves modifying the weight allocation in the weighted fusion strategy of spectral-mass spectrometry matching score and breakage probability weight. The adjusted parameters are then used to perform component identification analysis on the next batch of traditional Chinese medicine compound samples, forming a closed-loop process of analysis-validation-optimization.

[0100] See Figure 5This is a distribution chart of the spectral and mass spectrometric consistency and matching degree of candidate compounds, showing the matching degree (blue bars), spectral consistency (orange broken line), and mass spectrometric consistency (green broken line) of six candidate compounds (AF), used to assess the strength of evidence for compounds in the identification of traditional Chinese medicine compound prescriptions. The mass spectrometric consistency of all compounds is higher than that of the spectral consistency, and the trends of the three are highly consistent, reflecting that the matching stability of the mass spectrometry fracture network in this identification system is better than that of the spectral hierarchy structure, and also demonstrating the synergistic effect of the "spectral-mass spectrometry" dual-dimensional verification. The matching degree, as a weighted fusion index of spectral and mass spectrometric consistency, consistently falls between the two and has a stronger correlation with the mass spectrometric consistency, indicating that this system places greater emphasis on the evidence weight of the mass spectrometry dimension when calculating the matching degree.

[0101] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for rapid identification and analysis of components of traditional Chinese medicine compounds, characterized in that, The method comprises the following processing stages: Performing a composite scan on the sample to be tested to obtain the original spectral response and mass spectrum fragment sequence of the sample to be tested; Performing step-by-step unfolding on the original spectral response to generate a spectral hierarchy structure comprising a plurality of decomposition levels, each decomposition level corresponding to a spectral feature mode; Performing trajectory tracking on the mass spectrum fragment sequence to establish a mass spectrum fragmentation network from the initial ion to the terminal fragment; Establishing an interactive verification channel between the spectral hierarchy structure and the mass spectrum fragmentation network, and transmitting a matching degree parameter between the spectral feature mode and the fragmentation node through the interactive verification channel; According to the matching degree parameter, screening a candidate compound list from the standard compound spectrum; For each member in the candidate compound list, calculating its feature contribution degree in the spectral hierarchy structure and path completeness degree in the mass spectrum fragmentation network; Merging the feature contribution degree and the path completeness degree to generate a confidence index of each candidate member, and completing the component identification report of the sample to be tested according to the confidence indexes of all candidate members.

2. The method for rapid identification and analysis of traditional Chinese medicine compound ingredients according to claim 1, characterized in that, The step of performing step-by-step unfolding on the original spectral response to generate a spectral hierarchy structure comprising a plurality of decomposition levels comprises: Extracting a group of initial spectral waveforms from the original spectral response; Applying a modal decoupling operation to each initial spectral waveform to separate each initial spectral waveform into a trend component, a detail component and a noise component; Aggregating the trend components from different initial spectral waveforms to form a first decomposition level of the spectral hierarchy structure, which reflects the overall drift characteristics of the spectrum; Classifying the detail components from different initial spectral waveforms, and grouping the detail components with similar frequency and amplitude characteristics into the same group to form a second decomposition level of the spectral hierarchy structure, which reflects the vibration characteristics of specific functional groups; Performing statistical analysis on the noise components to generate a noise base model, and attaching the noise base model to the spectral hierarchy structure as a background reference level of the spectral hierarchy structure.

3. The method of claim 2, wherein the method is characterized by, The step of performing trajectory tracking on the mass spectrum fragment sequence to establish a mass spectrum fragmentation network from the initial ion to the terminal fragment comprises: Analyzing the mass spectrum fragment sequence to identify all mass spectrum peak points with mass-to-charge ratio and intensity information; Constructing an initial fragmentation spectrum by taking the mass-to-charge ratio of the mass spectrum peak points as nodes and the possible fragmentation relationship inferred from the difference between the mass-to-charge ratios as directed edges; Assigning a fragmentation possibility weight to each directed edge in the initial fragmentation spectrum, which is calculated based on the fragmentation chemical rules and the intensity ratio of adjacent mass spectrum peak points; In the initial fragmentation spectrum, starting from the molecular ion peak node, performing a depth traversal search to all possible sub-fragment nodes, and recording all fragmentation paths that can reach the terminal fragment node; Integrating all fragmentation paths and the nodes and directed edges on the paths to form a mass spectrum fragmentation network comprising a plurality of fragmentation levels and branches, each path in the mass spectrum fragmentation network representing a possible molecular cleavage process.

4. The method of claim 3, wherein the method is characterized by, The step of establishing an interactive verification channel between the spectral hierarchy structure and the mass spectrum fragmentation network comprises: extracting, from the second decomposition level of the spectral hierarchy, feature frequency and intensity data of a set of detail components; extracting, from the mass spectral fragmentation network, mass-to-charge ratio data of a set of key fragmentation nodes and fragmentation likelihood weight of directed edges connecting the key fragmentation nodes; establishing a mapping table between feature frequency and feature mass-to-charge ratio, which defines a typical molecular fragment mass interval corresponding to a specific functional group vibration frequency; according to the mapping table, traversing each detail component in the spectral hierarchy and each key fragmentation node in the mass spectral fragmentation network, calculating a spectral-mass spectral matching score of each detail component and the mass spectral fragmentation network; weighting and fusing the spectral-mass spectral matching score and the corresponding fragmentation likelihood weight to generate the matching degree parameter, and passing the matching degree parameter back to the spectral hierarchy and the mass spectral fragmentation network to mark the spectral feature patterns and fragmentation nodes with high matching degree.

5. The method for rapid identification and analysis of traditional Chinese medicine compound ingredients according to claim 4, characterized in that, the candidate compound list is screened from the standard compound spectrum according to the matching degree parameter, including: receiving a plurality of matching degree parameters from the interactive verification channel; retrieving all known compounds from the standard spectrum library that have an intersection with the matching degree parameter in the feature frequency and mass-to-charge ratio interval to form a preliminary candidate set; for each known compound in the preliminary candidate set, comparing its standard spectrum with the spectral hierarchy layer by layer to calculate the spectral compliance; for each known compound in the preliminary candidate set, comparing its standard mass spectrum with the mass spectral fragmentation network to calculate the mass spectral compliance; performing ternary operation on the spectral compliance, the mass spectral compliance, and the corresponding matching degree parameter to generate a comprehensive screening score for each known compound; according to a preset screening score threshold, selecting known compounds with a comprehensive screening score exceeding the threshold from the preliminary candidate set to form the candidate compound list.

6. The method for rapid identification and analysis of traditional Chinese medicine compound ingredients according to claim 5, characterized in that, calculating the feature contribution degree of each member in the candidate compound list in the spectral hierarchy, including: obtaining the reference spectral feature pattern of the candidate compound from the standard spectrum library; projecting the reference spectral feature pattern onto the second decomposition level of the spectral hierarchy of the sample to be tested to identify the detail components explained by the candidate compound; accumulating the intensity values of all explained detail components to obtain the total explained signal intensity of the candidate compound in the sample to be tested; calculating the proportion of the total explained signal intensity in the total intensity of all detail components in the sample to be tested, and defining this proportion as the feature contribution degree of the candidate compound in the spectral hierarchy.

7. The method according to claim 6, characterized in that, calculating the path completeness degree of each member in the candidate compound list in the mass spectral fragmentation network, including: obtaining the reference mass spectral fragmentation path of the candidate compound from the standard spectrum library; finding a matching path in the mass spectral fragmentation network of the sample to be tested that is most similar to the reference mass spectral fragmentation path in node sequence and fragmentation order; comparing the matching path and the reference mass spectral fragmentation path to count the compliance degree of the matching path and the reference mass spectral fragmentation path at common fragmentation nodes, and the similarity degree of fragmentation likelihood weight; The path similarity coefficient is calculated by combining the similarity of the shared breaking node with the similarity of the breaking possibility weight; The path similarity coefficient is multiplied by the coverage of the matching path in the mass spectrometry fragmentation network to obtain the path integrity of the candidate compound in the mass spectrometry fragmentation network.

8. The method according to claim 7, characterized in that, The combined feature contribution degree and path integrity generate a confidence index of each candidate member, including: The feature contribution degree of the candidate compound is normalized to map its value range to a preset uniform interval; The path integrity of the candidate compound is normalized to map its value range to the same preset uniform interval; A spectral weight factor is assigned to the normalized feature contribution degree, and a mass spectrometry weight factor is assigned to the normalized path integrity, and the sum of the spectral weight factor and the mass spectrometry weight factor is a fixed value; The weighted normalized feature contribution degree and the weighted normalized path integrity are added to obtain a preliminary confidence value of the candidate compound; A matching degree parameter related to the candidate compound from an interactive verification channel is introduced to correct the preliminary confidence value to generate a final confidence index.

9. The method of claim 1, wherein the method is characterized by, After generating the component identification report of the sample to be tested, it further includes: Collecting verification data from a new batch of traditional Chinese medicine compound samples; The compound information predicted to exist in the component identification report is compared with the compound information confirmed by the control sample in the verification data one by one; The number of correctly predicted compounds and the number of incorrectly predicted compounds are counted to calculate the accuracy index of the identification process; Based on the accuracy index, at least one of the following parameters is adjusted in reverse: the depth of the spectral hierarchical structure expansion, the calculation rule of the breaking possibility weight in the mass spectrometry fragmentation network, and the generation method of the matching degree parameter; Using the adjusted parameters, the next batch of traditional Chinese medicine compound samples is subjected to component identification analysis.

10. A rapid identification and analysis system for Chinese medicine compound ingredients, characterized in that, A processor and a memory are included, the memory and the processor are connected, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to realize the traditional Chinese medicine compound component rapid identification analysis method of any one of claims 1 to 9.

Images