Liquid chromatogram detection result processing method based on big data analysis

Abstract

The invention relates to the technical field of big data result processing, in particular to a big data analysis-based liquid chromatogram detection result processing method, which comprises the following steps: extracting peak shape characteristics to construct a structural table, establishing a peak mapping relation to generate a response matrix, constructing a co-occurrence graph connection set to identify a closed path and mark an offset label, according to the method, through peak shape index screening and cross-sample dynamic grouping, the accuracy of data integration and mapping is improved, map structure establishment is combined with peak co-occurrence and index distribution difference, high-strength correlation recognition is achieved by means of edge weight sorting, and the method is high in practicability and high in practicability. And through closed path analysis, unstable connection of a structure is eliminated, the stability of a result is enhanced, a connected peak group is divided according to generality attributes, accurate judgment of substance attribution is realized, structure extraction, anomaly elimination and generality clustering are considered in the whole process, and the automatic processing capacity, attribution precision and analysis credibility under complex chromatographic data are improved.

Description

Technical Field

[0001] This invention relates to the field of big data result processing technology, and in particular to a method for processing liquid chromatography detection results based on big data analysis. Background Technology

[0002] The field of big data results processing technology involves methods for collecting, preprocessing, extracting features, statistically analyzing, and visualizing large-scale data from diverse sources. Core aspects include parallel processing of structured, semi-structured, and unstructured data; data mining and model building; and dynamic updating and correlation analysis methods based on high-performance computing architectures. This overall technological field is oriented towards the analytical needs of specific scenarios, comprehensively utilizing distributed computing frameworks, batch and stream processing mechanisms, and data semantic understanding techniques to achieve full-process processing and feature reveal of target datasets. Traditional liquid chromatography (LC) detection results processing methods involve manually reading chromatogram data collected by the chromatograph or using analytical tools with fixed parameters to extract chromatographic peak characteristic parameters such as retention time, peak area, and peak height. Qualitative or quantitative analysis of the analyte is then performed according to predefined criteria. This typically involves peak identification based on template matching, noise removal conditions using fixed thresholds, and peak standards set using empirical models to complete data processing and preliminary result judgment.

[0003] In existing technologies, the processing of chromatographic detection results often relies on manual interpretation or pre-defined rule-based analysis tools. Peak parameters are extracted from each individual chromatogram, and qualitative and quantitative judgments are made accordingly. The overall process lacks a unified organization across sample structures. Manual interpretation is prone to subjective bias when noise fluctuations are large or peak shapes are complex. Fixed parameter rules are difficult to adjust in a timely manner when faced with baseline drift, overall retention time shifts, and peak shape variations caused by matrix effects. Weak response peaks are often masked by thresholds, while occasional interference peaks are easily misidentified as target signals. Traditional operating modes often analyze single or small samples, paying less attention to the systematic relationships between peak positions and responses across different samples. In high-throughput experimental scenarios, a large number of chromatograms can only be processed by batch repetition of the same template, making it difficult to identify stable common features and unstable abnormal combinations from a holistic perspective. Template matching relies on existing standard spectra and experienced peak ranges. When faced with new substances, new formulations, or changes in instrument status, mapping relationships often become disordered, manifesting as inconsistent attribution of the same substance in different batches, or different substances being assigned to the same template, affecting the consistency of trend analysis and quality evaluation results. Insufficient structuring also limits the exploration of co-occurrence relationships between peaks. Multi-peak combinations in chromatograms are often examined only from a single-peak perspective, making it impossible to identify peak groups that stably and synergistically change in multiple samples. This significantly compresses the depth of insight into potential common components, synergistic impurities, and formulation differences, and also increases the workload and uncertainty when tracing the causes and optimizing the process later. Summary of the Invention

[0004] To address the technical problems existing in the prior art, embodiments of the present invention provide a method for processing liquid chromatography detection results based on big data analysis. S1: Extract the time series signal of the sample, identify the peak region, calculate the retention time, peak area, peak width-to-height ratio and symmetry, set a threshold based on the peak area to filter peak records that meet the response criteria as the participants, organize them into structured data according to the sample number, and generate a peak response structure table. S2: Using the retention time information in the peak response structure table, set the error tolerance, divide the peaks with similar retention times in the difference samples into the same mapping group, establish the mapping relationship between peaks and sample numbers, construct the response state matrix, and generate a cross-sample peak mapping structure set. S3: Construct graph nodes based on the cross-sample peak mapping structure set, analyze the co-occurrence and response distribution of peaks among samples, calculate the weight of connection edges, reorder connection paths, and establish a multi-peak co-occurrence graph connection set; S4: Identify three-node closed paths in the multi-peak co-occurrence spectrum connection set, extract edge weights and calculate the difference ratio, mark edges that exceed the tolerance as structurally unstable combinations, and generate peak group structure offset labels. S5: Based on the peak group structure offset label, filter peak groups that maintain connectivity and mapping consistency, divide them into potential common material units, and generate material attribution structure results between samples.

[0005] As a further embodiment of the present invention, the peak response structure table includes a sample index set, a peak feature parameter set, a peak response evaluation set, and a peak record entry set; the cross-sample peak mapping structure set includes a retention time mapping set, a sample peak correspondence set, and a cross-sample response state matrix set; the multi-peak co-occurrence spectrum connection set includes a spectrum node structure set, a connection edge weight parameter set, a multi-peak co-occurrence path sequence set, and a complete spectrum topology set; the peak group structure offset label includes a closed path identification label, a connection edge difference ratio label, and a structurally unstable combination label; and the sample material attribution structure result includes a potential common material attribution set, a stable peak group connectivity set, and a sample attribution relationship mapping set.

[0006] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Obtain the time series signal of each sample in the chromatogram dataset, identify peak regions with continuous fluctuation characteristics based on the change of response value over time, mark the start and end time points, number them in chronological order, extract the time distribution information of all peak regions, and generate peak region interval distribution. S102: Based on the distribution of the peak region intervals, extract the retention time, peak area, peak width-to-height ratio and symmetry parameters of each peak region, classify and organize the parameters according to the sample number, and generate a peak parameter set; S103: Based on the set of peak parameters, the peak area and response threshold are judged, the peak shape regions that meet the requirements are selected, and the corresponding parameter information is extracted, and the data is summarized and organized into structured data according to the sample number to generate a peak response structure table.

[0007] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Based on the retention time information in the peak response structure table, set an error tolerance range, judge the retention time difference of all peaks, divide the peaks in the same error range into the same mapping group, and label the corresponding group number to generate a retention time mapping label sequence. S202: Based on the retention time mapping label sequence, extract the mapping group number and the sample number of each peak, group them according to the mapping group, classify the sample numbers in the same group, establish the grouping relationship between peaks and sample numbers, and generate the peak group sample number correspondence relationship. S203: Based on the correspondence between the peak group sample numbers, extract the retention time and peak area information of the corresponding peak, construct a matrix-form response state structure according to the mapping group and sample number, fill in the position identifier of the missing information, and generate a cross-sample peak mapping structure set.

[0008] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Call the mapping group number in the cross-sample peak mapping structure set, extract the sample number and peak area information corresponding to the peak in the same group, pair the sample numbers according to the co-occurrence of the same group, construct the connection structure of the nodes in the spectrum, and generate the sample node connection relationship value. S302: Based on the sample node connection relationship value, extract the peak area information of the node pair, calculate the weight parameters of the connection edge according to the response difference and the common peak distribution state, and establish the node edge connection weight coefficient. S303: Based on the node edge connection weight coefficients, arrange the connection relationships between nodes in order of weight values, construct the graph connection path according to the sorting result, integrate the nodes and paths into a unified structure, and generate a multi-peak co-occurrence graph connection set.

[0009] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Identify closed paths composed of three nodes in the multi-peak co-occurrence map connection set, extract the node numbers and connection edge information of the structure, filter the path combinations that satisfy the interconnection relationship and constitute a closed structure, and establish a set of three-node closed paths. S402: Based on the set of closed paths of the three nodes, extract the edge weight parameters of the connecting edges in each path, calculate the edge weight difference ratio, and judge the ratio result according to the structural stability tolerance value to generate the edge weight difference ratio distribution value. S403: Based on the edge weight difference ratio distribution value, mark the connecting edges whose difference ratio exceeds the structural stability tolerance value, extract the node number and edge status information in the corresponding path, and establish peak group structure offset label.

[0010] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Based on the edge connection information marked in the peak group structure offset label, remove the corresponding connection edges from the multi-peak co-occurrence graph connection set, identify the structural paths that maintain the integrity of the graph topology in the remaining connection relationships, filter the node combinations with continuous and unbroken edge connection relationships in the path, and establish a connected structural path set. S502: Based on the connected structural path set, determine the co-occurrence attribute and mapping correspondence between node groups in the path, extract combined units that have the same mapping number and adjacent node position sequence, filter structural paths with consistent mapping performance, and generate a common mapping belonging unit set. S503: Based on the common mapping attribution unit set, divide the node combinations with connection continuity and mapping consistency, extract the node number and path index of the combination, mark the attribution correspondence between each combination, and establish the material attribution structure result between samples.

[0011] As a further aspect of the present invention, the peak response structure table is a structured table that records the retention time, peak area, peak width-to-height ratio, and peak shape symmetry data items of the corresponding peak in each sample, and is filtered and organized according to the set response criteria.

[0012] As a further aspect of the present invention, the cross-sample peak mapping structure set refers to the set of mapping relationships between peaks across multiple samples, and by comparing the retention time characteristics between samples, the same or similar peaks are found. The term "spectral node" refers to a structure in the co-occurrence spectrum where each peak group is a node, and connections are established through the co-occurrence relationships between nodes. The spectral structure records the relationships between peaks in the sample. The multi-peak co-occurrence spectrum connection set is a graphical structure constructed based on the co-occurrence relationship between peaks in multiple sample data. The spectrum records the connection strength and mutual relationship pattern between peak nodes.

[0013] As a further aspect of the present invention, the peak group structure offset label refers to marking unstable or offset peak groups in a multi-peak co-occurrence spectrum by calculating the difference in connection strength in the path, thereby identifying potential structural problems or inconsistent peak responses. The inter-sample material attribution structure result refers to classifying substances in different samples by analyzing the commonalities between spectral structures and peak groups, and deriving a set of peak groups with similar attributes.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the accuracy of data integration and mapping is improved by screening peak shape indicators and dynamically grouping across samples. The spectral structure is established by combining peak co-occurrence and indicator distribution differences, and high-strength association identification is achieved by using edge weight sorting. The stability of the results is enhanced by eliminating unstable structural connections through closed path analysis. Connected peak groups are divided according to common attributes to achieve accurate determination of substance attribution. The overall process takes into account structure extraction, anomaly removal and common clustering, improving the automated processing capability, attribution accuracy and analytical reliability of complex chromatographic data. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0019] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] Please see Figure 1This invention provides a method for processing liquid chromatography detection results based on big data analysis, including the following steps: S1: Obtain the time series signal corresponding to each sample in the chromatogram dataset, identify the peak region and extract the retention time, peak area, peak width-to-height ratio and symmetry index, make a threshold judgment based on the peak area index, record the peaks that meet the response criteria as the participating objects, and organize them into structured data according to the sample number and extracted parameters to generate a peak response structure table. A chromatographic dataset refers to a collection of chromatographic data from multiple samples obtained by liquid chromatography, including characteristic parameters such as retention time, peak area, peak shape, and peak height for each sample, which serve as the basis for data processing. The peak response structure table is a structured table that records the retention time, peak area, peak width-to-height ratio, and peak shape symmetry data items of the corresponding peak in each sample. It is filtered and organized according to the set response criteria for analysis and comparison. S2: Based on the retention time information in the peak response structure table, set the error tolerance interval, divide the peaks with the same retention time in the difference samples into the same mapping group, construct the mapping relationship between each group of peaks and sample numbers, organize the response state into a structure matrix, and obtain the cross-sample peak mapping structure set. Cross-sample peak mapping structure set refers to the set of mapping relationships between peaks across multiple samples. By comparing the retention time characteristics between samples, the same or similar peaks can be identified, providing a data basis for inferring the material attribution between samples. S3: Call the mapping group in the cross-sample peak mapping structure set to construct the graph node structure, analyze the co-occurrence situation between samples and the distribution differences of response indicators, calculate the weight parameters between the connection edges based on the information, reorder the connection paths according to the weight parameters, establish a complete graph structure, and generate a multi-peak co-occurrence graph connection set; The spectral node structure refers to the structure in a co-occurrence spectrum where each peak group is a node, and connections are established through the co-occurrence relationships between nodes. The spectral structure records the relationships between peaks in a sample; Multi-peak co-occurrence spectral connection set is a graphical structure constructed based on the co-occurrence relationship between peaks in multiple sample data. The spectrum records the connection strength and mutual relationship pattern between peak nodes, which is used for spectrum optimization and structural analysis. S4: Identify closed paths consisting of three nodes in the multi-peak co-occurrence spectrum connection set, extract weight parameters from the connection edges in the path and calculate the difference ratio, mark the connection edges whose difference ratio exceeds the tolerance, identify them as structurally unstable combinations, and generate peak group structure offset labels. Peak group structure shift labeling refers to marking structurally unstable or shifted peak groups in a multi-peak co-occurrence spectrum by calculating the difference in connection strength in the path, thereby identifying potential structural problems or inconsistent peak responses; S5: Based on the edge connection information marked in the peak group structure offset label, unstable connection relationships are removed from the multi-peak co-occurrence spectrum connection set, peak groups that still maintain connectivity and have common mapping attributes are screened in the spectrum and divided into potential common material attribution units to generate material attribution structure results between samples. The inter-sample material attribution structure result refers to classifying substances in different samples by analyzing the commonalities between spectral structures and peak groups, and deriving a set of peak groups with similar attributes for inference of material attribution.

[0023] The peak response structure table includes a sample index set, a peak feature parameter set, a peak response evaluation set, and a peak record entry set. The cross-sample peak mapping structure set includes a retention time mapping set, a peak correspondence relationship set between samples, and a cross-sample response state matrix set. The multi-peak co-occurrence graph connection set includes a graph node structure set, a connection edge weight parameter set, a multi-peak co-occurrence path sequence set, and a complete graph topology set. The peak group structure offset label includes a closed path identification label, a connection edge difference ratio label, and a structurally unstable combination label. The inter-sample material attribution structure results include a potential common material attribution set, a stable peak group connectivity set, and an inter-sample attribution relationship mapping set.

[0024] Please see Figure 2 The specific steps of S1 are as follows: S101: Obtain the time series signal of each sample in the chromatogram dataset, identify peak regions with continuous fluctuation characteristics based on the change of response value over time, mark the start and end time points, number them in chronological order, extract the time distribution information of all peak regions, and generate peak region interval distribution. Time-series signals for each sample are acquired using a chromatograph, reflecting the changes in the concentration of each component over time. Peak regions in the chromatographic data represent these concentration changes, and peak regions with continuous fluctuations can be identified by analyzing the fluctuations in these signals. Peak detection algorithms can automatically identify the start and end times of these peaks, thus extracting the temporal distribution of the peak regions. For example, when analyzing a drug in a sample, the drug concentration may rise sharply over a certain period, forming a peak. Identifying these peaks relies on the signal fluctuation pattern and can be achieved using a threshold detection method based on response value changes. A response threshold is set; when the signal exceeds this threshold, the region is considered a valid peak region. After determining the start and end times of the peak, its temporal distribution can be generated. Each peak region is numbered chronologically, and its duration, width, and other relevant information are extracted for subsequent analysis.

[0025] S102: Based on the distribution of peak regions, extract the retention time, peak area, peak width-to-height ratio and symmetry parameters of each peak region, classify and organize the parameters according to the sample number, and generate a set of peak parameters; After extracting the relevant parameters for each peak region, the retention time is calculated first. Retention time represents the time difference between the start and end times of the peak. Peak area represents the area beneath the peak, which can be obtained by multiplying the signal value by the time interval, reflecting the peak's intensity. The peak width-to-height ratio (WHM) indicates the relationship between the peak's width and its maximum response time, typically calculated as the difference between the maximum response point and the peak's start or end time. Symmetry measures whether the peak is symmetrical, usually assessed by comparing the shape changes of the first and second halves of the peak. Calculating these parameters provides a comprehensive understanding of each peak's characteristics, thus supporting subsequent screening and analysis. For example, assuming a peak region in a sample has a start time of 10 seconds, an end time of 20 seconds, and a maximum response value at 15 seconds, its retention time is 10 seconds, and the peak area and width are calculated using numerical integration. Symmetry can be determined by the rate of change of the response value before and after the peak; if the change is relatively uniform between the first and second halves, the peak is considered symmetrical.

[0026] S103: Based on the peak parameter set, the peak area and response threshold are judged, the peak shape regions that meet the requirements are selected, and the corresponding parameter information is extracted. The data is then summarized and organized into structured data according to the sample number to generate a peak response structure table. After obtaining the parameters for each peak region, valid peak regions need to be screened by comparing peak area and response threshold. The response threshold is set based on the background noise of the sample, typically by calculating the average and standard deviation of the background noise and setting a threshold higher than the background noise. For example, with an average response value of 10 and a standard deviation of 2, the response threshold can be set to 14, which is the noise average plus twice the standard deviation, used to determine whether the peak area reaches a valid response level. If the peak area and response value of a peak exceed this threshold, the peak is considered valid; otherwise, it is discarded. This ensures that the selected peaks represent significant changes in the components of the sample, rather than noise interference. During the screening process, if the response value of a peak region is lower than the set threshold, it will be excluded without affecting subsequent data analysis. Finally, the peak regions that meet the threshold requirements are organized by sample number to generate a structured data table. The table includes the retention time, peak area, width, symmetry, and a flag indicating whether it meets the response threshold for each peak, facilitating subsequent data analysis.

[0027] Please see Figure 3 The specific steps of S2 are as follows: S201: Based on the retention time information in the peak response structure table, set the error tolerance range, judge the retention time difference of all peaks, divide the peaks in the same error range into the same mapping group, and label the corresponding group number to generate a retention time mapping label sequence. The setting of the error tolerance interval needs to consider the fluctuation range of the actual data. For example, setting a reasonable tolerance interval of ±0.5 seconds means that peaks with retention times differing by no more than 0.5 seconds are considered to belong to the same mapping group. For the retention time of each peak, calculate its time difference with other peaks. If the time difference is within the set error tolerance interval, these peaks are classified into the same group. For each group of peaks, assign a unique group number and generate a retention time mapping label sequence. For example, if there are two peaks in the sample with retention times of 5.2 seconds and 5.4 seconds respectively, and the set error tolerance interval is 0.5 seconds, then these two peaks will be classified into the same mapping group and assigned the same group number. In this way, peaks in all samples can be reasonably classified according to their retention time differences, ensuring that the association between peaks and groups is clear, which facilitates subsequent grouping and analysis.

[0028] S202: Based on the retention time mapping label sequence, extract the mapping group number and the sample number to which each peak belongs, group them according to the mapping group, classify the sample numbers in the same group, establish the grouping relationship between peaks and sample numbers, and generate the peak group sample number correspondence relationship. Each peak is associated with its corresponding sample and grouped according to the mapping group number. First, the group number of each peak is found based on the previous retention time mapping label sequence. Then, these peak group numbers are matched one-to-one with the numbers of their respective samples, grouping them according to the mapping group. During grouping, the sample numbers within each group need to be categorized and recorded to ensure that sample numbers within the same group clearly correspond to the same peak. For example, if a group contains samples A, B, and C, the peaks of these samples all belong to the same mapping group and have the same retention time. This method establishes the relationship between peaks and sample numbers, forming a classification of sample numbers within each mapping group, ultimately generating a peak group sample number correspondence, which is crucial for subsequent response state analysis.

[0029] S203: Based on the correspondence between peak group sample numbers, extract the retention time and peak area information of the corresponding peak, construct a matrix-form response state structure according to the mapping group and sample number, fill in the position identifier of the missing information, and generate a cross-sample peak mapping structure set. First, based on the correspondence between peaks and sample numbers, the specific retention time and peak area information for each peak are determined. Then, a matrix-like response state structure is constructed according to the mapping group and sample number, where each cell represents the peak response data of a specific sample within a group. This matrix structure clearly demonstrates the relationships between different mapping groups and different samples. For missing data, appropriate labels can be used to fill in the missing data, ensuring the integrity of the response state structure. The location for filling in missing data can be determined according to certain rules, such as using the average value of surrounding sample data to fill in the missing parts, or inferring a reasonable range for the missing parts based on historical data. Finally, this method generates a cross-sample peak mapping structure set, which comprehensively presents the peak data between different samples and mapping groups, providing a basis for further data analysis. For example, if the peak area of ​​a sample is missing, it can be filled in based on the response data of other samples in the same group, ensuring the integrity and consistency of the dataset.

[0030] Please see Figure 4 The specific steps of S3 are as follows: S301: Call the mapping group number in the cross-sample peak mapping structure set, extract the sample number and peak area information corresponding to the peak in the same group, pair the sample numbers according to the co-occurrence of the same group, construct the connection structure of the nodes in the spectrum, and generate the sample node connection relationship value. First, for each mapping group, the sample ID and peak area information of all peaks in that group are retrieved. For example, if a group contains samples A, B, and C, and their peak areas are 50, 55, and 60 respectively, then these sample IDs and corresponding peak areas are extracted. Next, based on the co-occurrence of samples, sample IDs within the same group are paired. For instance, if samples A and B, or samples A and C, appear in the same group, connections between these sample nodes are established in the graph. Then, based on these paired sample IDs, the connection structure of nodes in the graph is constructed, and sample node connection values ​​are generated. If sample A is connected to sample B, the connection value can be weighted by factors such as peak area difference and retention time between samples, ultimately forming a connection matrix where each cell represents the degree of connection between samples, thus generating sample node connection values.

[0031] S302: Based on the sample node connection relationship value, extract the peak area information of node pairs, calculate the weight parameters of the connection edges according to the response difference and the common peak distribution state, and establish the node edge connection weight coefficient. In this embodiment, the weight parameters of the connecting edges are quantitatively calculated based on the peak area difference and co-occurrence distribution of sample pairs. For ease of understanding, the weights can be determined as follows: ; in, and Samples and samples Peak area in this mapping group Indicates the difference in response; This represents the number of times the two samples co-occur in this mapping group; and As a weighting coefficient, this formula realizes the negative correlation quantification of "response difference" and the positive correlation quantification of "common peak distribution state", so that sample pairs with smaller differences and more co-occurrence have higher connection weights; First, for each pair of connected sample nodes, their peak area information is extracted. For example, if the peak areas of sample A and sample B are 50 and 55 respectively, the difference is 5, which can be used as a basis for calculating the weight of the connection edge. Furthermore, the co-occurrence of the two samples needs to be considered. If sample A and sample B share 3 peaks, the weight of the connection edge between them will be higher; otherwise, it will be lower. In this way, the weight coefficient between each pair of samples can be calculated based on the differences and co-occurrence of the samples, thus providing support for subsequent map path construction.

[0032] S303: Based on the node edge connection weight coefficients, arrange the connection relationships between nodes in order of weight values, construct the graph connection path according to the sorting results, integrate the nodes and paths into a unified structure, and generate a multi-peak co-occurrence graph connection set. After sorting, node pairs with higher connection weights will be ranked first, indicating a closer relationship and stronger correlation. For example, if the connection weight between node A and node B is 0.8, and the connection weight between node B and node C is 0.6, then the connection between node A and node B will be prioritized in the sorting. Next, the weighted connections will be connected step-by-step to form the connection paths in the graph. Each connection path illustrates the association between nodes, and these paths are combined to construct a complete graph structure. Finally, all nodes and connection paths are integrated into a unified multi-modal co-occurrence graph connection set, helping to reveal the similarities and differences between different samples and providing support for further data analysis.

[0033] Please see Figure 5 The specific steps of S4 are as follows: S401: Identify closed paths composed of three nodes in the multi-peak co-occurrence map connection set, extract the node numbers and connection edge information of the structure, filter the path combinations that satisfy the interconnection relationship and constitute a closed structure, and establish a set of three-node closed paths. First, by analyzing the nodes and their connections in the graph, we find and identify closed paths consisting of three nodes. For example, suppose there are connections between nodes A, B, and C in the graph, and these connections form a closed structure. This means that from node A to node B, then to node C, and finally back to node A, a complete closed loop is formed. For each three-node path, we extract the node numbers and the connection edge information between the nodes. A connection edge refers to the relationship between nodes, such as the connection edge between node A and node B. Then, we filter out path combinations that satisfy the interconnection relationships and form a closed structure, ensuring that each path is a valid closed path. These path combinations that meet the criteria are organized into a set of three-node closed paths.

[0034] S402: Based on the set of three-node closed paths, extract the edge weight parameters of the connecting edges in each path, calculate the edge weight difference ratio, and judge the ratio result according to the structural stability tolerance value to generate the edge weight difference ratio distribution value. In this embodiment, for each closed path consisting of three nodes, the edge weights of the three connecting edges are denoted as follows: From this, the maximum edge weight is obtained. with minimum edge weight Simultaneously calculate the average of the weights of the three edges. and with The edge weight difference ratio of the closed path is used as the reference. When the difference ratio is greater than the preset structural stability tolerance value, it is considered that the peak group structure corresponding to the closed path has obvious offset or instability, and the connecting edges in the closed path are marked as structurally unstable combinations. First, the connection edge information for each path is extracted from the set of three-node closed paths, including the weight parameters of each connection edge. For example, suppose the connection edges in the path connect nodes A and B, B and C, and C and A, respectively; the weights of these connection edges might be 0.7, 0.6, and 0.75. Next, the weight difference of these connection edges is calculated to obtain the edge weight difference ratio. If the weight difference of the connection edges in a path is large, it indicates that the stability of this path is low. For example, if the weight of the connection edge between nodes A and B is 0.7, and the weight between nodes B and C is 0.6, the difference ratio between them is 14.2%. Then, these difference ratios are compared with a preset structural stability tolerance value. If the edge weight difference ratio in a path exceeds the tolerance value, for example, the set tolerance value is 20%, then the path is considered to have significant structural instability. Finally, through the calculation of these difference ratios, an edge weight difference ratio distribution value is generated to show the stability distribution of all three-node paths in the graph.

[0035] S403: Based on the distribution value of the edge weight difference ratio, mark the connecting edges whose difference ratio exceeds the structural stability tolerance value, extract the node number and edge status information in the corresponding path, and establish peak group structure offset label; Based on the calculated edge weight difference ratio, connecting edges whose difference ratio exceeds the structural stability tolerance value are marked. For example, if the edge weight difference ratio between node B and node C in a certain path is 25%, exceeding the set tolerance value of 20%, then this edge will be marked as an unstable connecting edge. Next, the node numbers and edge state information of the paths containing these unstable edges are extracted. Assuming the connecting edges in the path are node A and node B, node B and node C, and node C and node A, if the connecting edge between node B and node C is marked as an unstable edge, then the state information of node B and node C in the path is extracted. Finally, based on this information, peak group structure offset labels are established to mark structurally unstable parts in the graph. These labels can help analyze which nodes or paths have potential structural problems, providing a basis for subsequent optimization and adjustment.

[0036] Please see Figure 6 The specific steps of S5 are as follows: S501: Based on the edge connection information marked in the peak group structure offset label, remove the corresponding connection edges from the multi-peak co-occurrence graph connection set, identify the structural paths that maintain the integrity of the graph topology in the remaining connection relationships, filter the node combinations with continuous and unbroken edge connection relationships in the path, and establish a connected structural path set. First, edge connection information is extracted from the generated peak group structure offset labels. Unstable edges are removed from the multi-peak co-occurrence graph. Taking nodes A, B, C, and D in the graph as an example, suppose the connection between node A and node B is marked as an unstable edge, and therefore needs to be removed from the graph. After removing these unstable edges, the remaining connections in the graph are analyzed. At this point, the focus is on identifying which paths still maintain the topological integrity of the graph, i.e., paths without breaks. For example, if the connection between node C and node D still exists, and there is also a connection between node C and node A, then A, C, and D can still form a valid path, maintaining the connectivity of the graph. Next, node combinations with continuous and unbroken edge connections in these paths are selected. For example, if the connection between node B and node C is removed, the original path connecting A, B, and C will be broken, but A, C, and D will still remain connected. Through this selection, a set of connected structural paths is finally obtained, which contains all valid and stable node connection paths in the graph.

[0037] S502: Based on the connected structural path set, determine the co-occurrence attribute and mapping correspondence between node groups in the path, extract combined units that have the same mapping number and adjacent node position sequence, filter structural paths with consistent mapping performance, and generate a set of common mapping belonging units. First, based on the set of connected structural paths, the co-occurrence attributes and mapping relationships between node groups in each path are analyzed. Co-occurrence attributes may refer to the frequency of relationships between nodes, while mapping relationships refer to the correspondence of the same nodes between different paths in the graph. Taking paths ABC and BCD as examples, nodes B and C appear in both paths, so nodes B and C have a certain co-occurrence attribute. Next, it is necessary to extract combined units that simultaneously possess the same mapping number and adjacent node position sequence. For example, assuming that the adjacent order of nodes B and C is completely consistent between path ABC with mapping number 1 and path BCD with mapping number 1, such combined units can be extracted. Then, structural paths with consistent mapping performance are selected. Consistency performance here refers to the consistency of the node order and co-occurrence attributes of the path, that is, the relative positions of nodes in the path and the frequently occurring node patterns. For example, if the order of nodes B, C, and D is consistent in multiple paths, these paths can be classified as structural paths with consistent mapping performance. Finally, through these selections, a set of common mapping affiliation units is generated, which contains all path combinations that meet the consistent mapping requirements.

[0038] S503: Based on the common mapping attribution unit set, divide the node combinations with connectivity continuity and mapping consistency, extract the node number and path index of the combination, mark the attribution correspondence between each combination, and establish the material attribution structure results between samples. First, based on the path information in the common mapping unit set, node combinations with continuity and mapping consistency are identified and divided. For example, if nodes A, B, and C maintain a certain connection and have consistent mapping numbers and adjacent position sequences in multiple paths, these nodes can be grouped together. For instance, if nodes B and C in paths ABC and BCD appear in the same order in multiple paths, then the combinations of A, B, C and B, C, and D possess continuity and mapping consistency. Next, the node numbers and path index information in these combination units are extracted. For example, if the node numbers of a path are A, B, and C, and the path index is 1, then this set of node numbers and path indexes will be extracted and included as part of the combination unit. Then, the attribution correspondence between each combination is labeled. At this point, an attribution label is assigned to each combination, indicating that the combination belongs to a specific structural unit, and its position and role in the entire graph are labeled. Finally, based on these attribution labels, a material attribution structure is established among the samples to demonstrate the attribution relationships of different path and node combinations, facilitating further analysis and research on the structural stability and material properties of the map.

[0039] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for processing liquid chromatography detection results based on big data analysis, characterized in that, Includes the following steps: S1: Extract the time series signal of the sample, identify the peak region, calculate the retention time, peak area, peak width-to-height ratio and symmetry, set a threshold based on the peak area to filter peak records that meet the response criteria as the participants, organize them into structured data according to the sample number, and generate a peak response structure table. S2: Using the retention time information in the peak response structure table, set the error tolerance, divide the peaks with similar retention times in the difference samples into the same mapping group, establish the mapping relationship between peaks and sample numbers, construct the response state matrix, and generate a cross-sample peak mapping structure set. S3: Construct graph nodes based on the cross-sample peak mapping structure set, analyze the co-occurrence and response distribution of peaks among samples, calculate the weight of connection edges, reorder connection paths, and establish a multi-peak co-occurrence graph connection set; S4: Identify three-node closed paths in the multi-peak co-occurrence spectrum connection set, extract edge weights and calculate the difference ratio, mark edges that exceed the tolerance as structurally unstable combinations, and generate peak group structure offset labels. S5: Based on the peak group structure offset label, filter peak groups that maintain connectivity and mapping consistency, divide them into potential common material units, and generate material attribution structure results between samples.

2. The method for processing liquid chromatography detection results based on big data analysis according to claim 1, characterized in that, The peak response structure table includes a sample index set, a peak feature parameter set, a peak response evaluation set, and a peak record entry set. The cross-sample peak mapping structure set includes a retention time mapping set, a sample peak correspondence set, and a cross-sample response state matrix set. The multi-peak co-occurrence spectral connectivity set includes a spectral node structure set, a connectivity edge weight parameter set, a multi-peak co-occurrence path sequence set, and a complete spectral topology set. The peak group structure offset label includes a closed path identification label, a connectivity edge difference ratio label, and a structurally unstable combination label. The sample material attribution structure results include a potential common material attribution set, a stable peak group connectivity set, and a sample attribution relationship mapping set.

3. The method for processing liquid chromatography detection results based on big data analysis according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain the time series signal of each sample in the chromatogram dataset, identify peak regions with continuous fluctuation characteristics based on the change of response value over time, mark the start and end time points, number them in chronological order, extract the time distribution information of all peak regions, and generate peak region interval distribution. S102: Based on the distribution of the peak region intervals, extract the retention time, peak area, peak width-to-height ratio and symmetry parameters of each peak region, classify and organize the parameters according to the sample number, and generate a peak parameter set; S103: Based on the set of peak parameters, the peak area and response threshold are judged, the peak shape regions that meet the requirements are selected, and the corresponding parameter information is extracted, and the data is summarized and organized into structured data according to the sample number to generate a peak response structure table. The peak region that meets the requirements refers to the peak region whose peak area is greater than the preset response threshold. The response threshold is determined based on the background noise level and is an effective response judgment value that is higher than the noise upper limit.

4. The method for processing liquid chromatography detection results based on big data analysis according to claim 3, characterized in that, The specific steps of S2 are as follows: S201: Based on the retention time information in the peak response structure table, set an error tolerance range, judge the retention time difference of all peaks, divide the peaks in the same error range into the same mapping group, and label the corresponding group number to generate a retention time mapping label sequence. S202: Based on the retention time mapping label sequence, extract the mapping group number and the sample number of each peak, group them according to the mapping group, classify the sample numbers in the same group, establish the grouping relationship between peaks and sample numbers, and generate the peak group sample number correspondence relationship. S203: Based on the correspondence between the peak group sample numbers, extract the retention time and peak area information of the corresponding peak, construct a matrix-form response state structure according to the mapping group and sample number, fill in the position identifier of the missing information, and generate a cross-sample peak mapping structure set.

5. The method for processing liquid chromatography detection results based on big data analysis according to claim 4, characterized in that, The specific steps for S3 are as follows: S301: Call the mapping group number in the cross-sample peak mapping structure set, extract the sample number and peak area information corresponding to the peak in the same group, pair the sample numbers according to the co-occurrence of the same group, construct the connection structure of the nodes in the spectrum, and generate the sample node connection relationship value. S302: Based on the sample node connection relationship value, extract the peak area information of the node pair, calculate the weight parameters of the connection edge according to the response difference and the common peak distribution state, and establish the node edge connection weight coefficient. S303: Based on the node edge connection weight coefficients, arrange the connection relationships between nodes in order of weight values, construct the graph connection path according to the sorting result, integrate the nodes and paths into a unified structure, and generate a multi-peak co-occurrence graph connection set.

6. The method for processing liquid chromatography detection results based on big data analysis according to claim 5, characterized in that, The specific steps of S4 are as follows: S401: Identify closed paths composed of three nodes in the multi-peak co-occurrence map connection set, extract the node numbers and connection edge information of the structure, filter the path combinations that satisfy the interconnection relationship and constitute a closed structure, and establish a set of three-node closed paths. S402: Based on the set of closed paths of the three nodes, extract the edge weight parameters of the connecting edges in each path, calculate the edge weight difference ratio, and judge the ratio result according to the structural stability tolerance value to generate the edge weight difference ratio distribution value. The calculated edge weight difference ratio refers to the ratio obtained by determining the maximum and minimum values ​​of the edge weights of the three connecting edges, and dividing the difference between the maximum and minimum edge weights by the average of the three edge weights. S403: Based on the edge weight difference ratio distribution value, mark the connecting edges whose difference ratio exceeds the structural stability tolerance value, extract the node number and edge status information in the corresponding path, and establish peak group structure offset label.

7. The method for processing liquid chromatography detection results based on big data analysis according to claim 6, characterized in that, The specific steps of S5 are as follows: S501: Based on the edge connection information marked in the peak group structure offset label, remove the corresponding connection edges from the multi-peak co-occurrence graph connection set, identify the structural paths that maintain the integrity of the graph topology in the remaining connection relationships, filter the node combinations with continuous and unbroken edge connection relationships in the path, and establish a connected structural path set. S502: Based on the connected structural path set, determine the co-occurrence attribute and mapping correspondence between node groups in the path, extract combined units that have the same mapping number and adjacent node position sequence, filter structural paths with consistent mapping performance, and generate a common mapping belonging unit set. S503: Based on the common mapping attribution unit set, divide the node combinations with connection continuity and mapping consistency, extract the node number and path index of the combination, mark the attribution correspondence between each combination, and establish the material attribution structure result between samples.

8. The method for processing liquid chromatography detection results based on big data analysis according to claim 1, wherein the peak response structure table is a structured table that records the retention time, peak area, peak width-to-height ratio, and peak shape symmetry data items of the corresponding peak in each sample, and is screened and organized according to the set response criteria.

9. The method for processing liquid chromatography detection results based on big data analysis according to claim 1, characterized in that, The cross-sample peak mapping structure set refers to the set of mapping relationships between peaks across multiple samples. By comparing the retention time characteristics between samples, peaks of the same or similar types can be identified. The term "spectral node" refers to a structure in the co-occurrence spectrum where each peak group is a node, and connections are established through the co-occurrence relationships between nodes. The spectral structure records the relationships between peaks in the sample. The multi-peak co-occurrence spectrum connection set is a graphical structure constructed based on the co-occurrence relationship between peaks in multiple sample data. The spectrum records the connection strength and mutual relationship pattern between peak nodes.

10. The method for processing liquid chromatography detection results based on big data analysis according to claim 1, characterized in that, The peak group structure offset label refers to marking unstable or offset peak groups in a multi-peak co-occurrence spectrum by calculating the difference in connection strength in the path, thereby identifying potential structural problems or inconsistent peak responses. The inter-sample material attribution structure result refers to classifying substances in different samples by analyzing the commonalities between spectral structures and peak groups, and deriving a set of peak groups with similar attributes.

Images