Data processing device, data processing method, data processing program, and data processing system

Abstract

To provide a technique capable of performing data analysis quickly and with high accuracy while reducing a workload of an analyst.SOLUTION: A data processing device 100 comprises: a data acquisition portion 105 that acquires detection data indicating signal strength corresponding to a component in a sample detected by a detection device 10; and a computing portion 101 that processes the detection data acquired by the data acquisition portion. The computing portion generates a plurality of pieces of analysis data including a peak of the signal strength on the basis of the detection data, generates a plurality of clusters by grouping the peak included in each of the plurality of pieces of analysis data using hierarchical clustering on the basis of peak information corresponding to the peak, and prohibits grouping a plurality of peaks included in the same analysis data, into the same cluster in the hierarchical clustering.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a data processing device, a data processing method, a data processing program, and a data processing system for processing data. [Background technology]

[0002] When analyzing components such as compounds in a sample, a plurality of analytical data detected from each of a plurality of samples is compared. For example, a technique called chromatography is known in which components contained in a sample to be analyzed are separated using a separation device called a chromatograph to obtain a separation result called a chromatogram. A chromatogram obtained by chromatography has peaks of signal intensity. An analyst can identify components contained in the sample to be analyzed by comparing the peaks contained in each of the plurality of chromatograms between the plurality of chromatograms.

[0003] For example, Patent Document 1 discloses generating a table that lists the peaks contained in each of a plurality of chromatograms obtained by chromatography according to peak information such as peak area or peak height. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2023-004872 A Summary of the Invention [Problem to be solved by the invention]

[0005] When generating a table such as that disclosed in Patent Document 1, for example, an analyst observes multiple chromatograms, estimates similar peaks among the multiple chromatograms, and registers the estimated peaks of each chromatogram in the table as peaks of the same component. The analyst can identify the components of each peak included in each chromatogram by performing an identification process on the peaks of each chromatogram registered in the table as peaks of the same component. However, in order to generate the above-mentioned table, the analyst needs to individually observe the peaks included in each of the multiple chromatograms and compare them with the peaks included in other chromatograms, which requires a huge amount of work time, and the accuracy of the generated table varies depending on the analyst's knowledge or experience.

[0006] The present disclosure has been made to solve such problems, and has as its purpose to provide a technique that enables rapid and highly accurate data analysis while reducing the workload of an analyst. [Means for solving the problem]

[0007] A data processing device according to an aspect of the present disclosure includes a data acquisition unit that acquires detection data indicating signal intensities corresponding to components in a sample detected by a detection device, and a calculation unit that processes the detection data acquired by the data acquisition unit. The calculation unit generates a plurality of analysis data including signal intensity peaks based on the detection data, groups the peaks included in each of the plurality of analysis data by hierarchical clustering based on peak information corresponding to the peaks to generate a plurality of clusters, and prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster.

[0008] A data processing method according to another aspect of the present disclosure includes the steps of acquiring detection data indicating signal intensities corresponding to components in a sample detected by a detection device, and processing the detection data acquired by the acquiring step. The step of processing the detection data includes the steps of generating a plurality of analysis data including signal intensity peaks based on the detection data, and grouping the peaks included in each of the plurality of analysis data by hierarchical clustering based on peak information corresponding to the peaks to generate a plurality of clusters. The step of generating a plurality of clusters prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster.

[0009] A data processing program according to another aspect of the present disclosure causes a computer to execute a step of acquiring detection data indicating signal intensity corresponding to components in a sample detected by a detection device, and a step of processing the detection data acquired by the acquiring step. The step of processing the detection data includes a step of generating a plurality of analysis data including signal intensity peaks based on the detection data, and a step of grouping the peaks included in each of the plurality of analysis data by hierarchical clustering based on peak information corresponding to the peaks to generate a plurality of clusters. The step of generating a plurality of clusters prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster.

[0010] A data processing system according to another aspect of the present disclosure includes a detection device and a data processing device that processes data. The data processing device includes a data acquisition unit that acquires detection data indicating signal intensities corresponding to components in a sample detected by the detection device, and a calculation unit that processes the detection data acquired by the data acquisition unit. The calculation unit generates a plurality of analysis data including peaks of signal intensity based on the detection data, groups the peaks included in each of the plurality of analysis data by hierarchical clustering based on peak information corresponding to the peaks to generate a plurality of clusters, and prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster. Effect of the Invention

[0011] According to the present disclosure, the data processing device can generate multiple clusters by grouping peaks included in each of multiple analytical data based on peak information corresponding to the peaks by hierarchical clustering, so that the analyst does not need to estimate peaks common to multiple analytical data by himself / herself to generate multiple clusters, and can perform data analysis quickly. Furthermore, the data processing device prohibits multiple peaks that satisfy a specific condition based on peak information from being grouped into the same cluster in the hierarchical clustering, so that multiple peaks that satisfy a specific condition based on peak information are not grouped into the same cluster as the same component, and data analysis can be performed with high accuracy using hierarchical clustering. [Brief description of the drawings]

[0012] [Figure 1] 1 is a diagram showing a configuration of a data processing system and a data processing device according to an embodiment; [Diagram 2] FIG. 2 is a diagram showing three-dimensional vector data constituted by a spectrum vector and a chromatogram generated by the data processing device of the embodiment. [Diagram 3]FIG. 1 is a diagram for explaining peak tracking using multiple chromatograms. [Figure 4] FIG. 13 is a diagram for explaining an example of comparison of peaks between a plurality of chromatograms. [Diagram 5] FIG. 13 is a diagram for explaining an example of generating a compound table. [Figure 6] FIG. 2 is a diagram for explaining an example of a graph in which peak vectors corresponding to each peak included in a plurality of chromatograms are represented in a two-dimensional coordinate system. [Figure 7] FIG. 13 is a diagram for explaining an application example of hierarchical clustering according to the centroid method. [Figure 8] 11 is a diagram for explaining an application example of hierarchical clustering according to a centroid method executed by a data processing device according to an embodiment. FIG. [Figure 9] 4 is a flowchart of data processing executed by a data processing device according to an embodiment. [Figure 10] FIG. 13 is a diagram for explaining an example of weighting of peak information in hierarchical clustering. [Figure 11] FIG. 13 is a diagram for explaining an example of a comparison between precision and recall in generating a compound table. [Figure 12] FIG. 13 is a diagram for explaining an example of excluding peaks from a cluster in hierarchical clustering. [Figure 13] FIG. 13 is a diagram for explaining an example of calculation of a representative value in each cluster grouped by hierarchical clustering. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] The present embodiment will now be described in detail with reference to the drawings. In the drawings, the same or corresponding parts are designated by the same reference characters and, in principle, descriptions thereof will not be repeated.

[0014] [Configuration of data processing system and data processing device] The configurations of a data processing system 1 and a data processing device 100 according to an embodiment will be described with reference to Fig. 1. Fig. 1 is a diagram showing the configurations of a data processing system 1 and a data processing device 100 according to an embodiment. As shown in Fig. 1, the data processing system 1 includes a chromatograph 10 and a data processing device 100.

[0015] The chromatograph 10 is an example of a "detection device". The chromatograph 10 includes a container 11, a liquid delivery pump 12, an injector 13, a column 14, and a detector 15. The container 11 contains a mobile phase. The liquid delivery pump 12 draws the mobile phase from the container 11 and delivers it at a constant flow rate. The injector 13 injects a sample to be analyzed into the mobile phase delivered by the liquid delivery pump 12. The column 14 contains a stationary phase and separates various components contained in the sample injected by the injector 13. The detector 15 detects the components eluted from the column 14. As the detector 15, for example, an absorptiometric detector (PDA (Photo Diode Array) detector), a fluorescence detector, a differential refractive index detector, a conductivity detector, or a mass spectrometer is used. Detection data indicating signal intensity corresponding to the components in the sample detected by the detector 15 is output to the data processing device 100.

[0016] Although the chromatograph 10 according to the embodiment is a liquid chromatograph (LC) that uses a liquid as the mobile phase, the chromatograph 10 may be another type of chromatograph, such as a gas chromatograph that uses a gas as the mobile phase.

[0017] In the chromatograph 10, an injector 13 injects a sample to be analyzed into a mobile phase. The injected sample reaches the column 14 along with the flow of the mobile phase fed by the liquid feed pump 12, and passes through the column 14. The various components contained in the sample take different times to pass through the column 14 depending on their affinity with the stationary phase or the mobile phase. For example, among the components contained in the sample, a component that is easily adsorbed to the stationary phase takes a longer time to pass through the column 14 (also referred to as a "retention time") than a component that is not easily adsorbed to the stationary phase. This separates the various components contained in the sample in the time direction by the column 14. An eluate containing the components separated in the column 14 is introduced from the column 14 to the detector 15. The detector 15 outputs detection data indicating a signal intensity corresponding to the concentration (amount) of the component introduced by the column 14. The detection data is processed by the data processing device 100 to generate a chromatogram. The solution that has passed through the detector 15 is discharged as a waste liquid.

[0018] The data processing device 100 may be a general-purpose computer, or may be a computer dedicated to the data processing system 1 for processing detection data from the chromatograph 10. The data processing device 100 includes a calculation device 101, a memory 102, a storage device 103, and an interface 104.

[0019] The arithmetic device 101 is an example of a "arithmetic unit." The arithmetic device 101 is a computing entity (computer) that executes various processes by executing various programs. The arithmetic device 101 is configured with a processor such as a CPU (Central Processing Unit), an MPU (Micro Processing Unit), or a GPU (Graphics Processing Unit). The processor, which is an example of the arithmetic device 101, has a function of executing various processes by executing a program, but some or all of these functions may be implemented using a dedicated hardware circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array). The "processor" is not limited to a processor in the narrow sense that executes processes using a stored program method such as a CPU or an MPU, but may include a hardwired circuit such as an ASIC or an FPGA. Therefore, the "processor," which is an example of the arithmetic device 101, may be read as a processing circuitry in which processing is defined in advance by computer-readable code and / or a hardwired circuit. The arithmetic device 101 may be configured with one chip or multiple chips. Furthermore, the processor and associated processing circuitry may be implemented as multiple computers interconnected by wire or wirelessly, such as via a local area network or wireless network. The processor and associated processing circuitry may be implemented as a cloud computer that performs remote calculations based on input data and outputs the results of the calculations to other devices at remote locations.

[0020] The memory 102 includes a volatile storage area (for example, a working area) that temporarily stores program codes, work memory, and the like when the arithmetic device 101 executes various programs. Examples of the storage unit include volatile memories such as DRAM (Dynamic Random Access Memory) and SRAM (Static Random Access Memory), and non-volatile memories such as ROM (Read Only Memory) and flash memory.

[0021] The storage device 103 stores various programs or various data executed by the arithmetic device 101. The storage device 103 may be one or more non-transitory computer readable media, or one or more computer readable storage media. Examples of the storage device 103 include a hard disk drive (HDD) and a solid state drive (SSD). The storage device 103 according to the embodiment stores a data processing program 130 for executing data processing in which the arithmetic device 101 processes detection data acquired from the chromatograph 10.

[0022] The interface 104 is an example of a "data acquisition unit." The interface 104 transmits and receives data to and from an external device or external equipment via wired or wireless communication. For example, the interface 104 communicates with the chromatograph 10 to acquire detection data output from the chromatograph 10. The interface 104 may also be a communication device that transmits detection data acquired from the chromatograph 10 to the cloud server or transmits execution results of data processing by the arithmetic device 101 to the cloud server by communicating with a cloud server (not shown). Furthermore, the interface 104 may transmit and receive data to and from the display unit 110 or the input unit 120, which are user interfaces, via wired or wireless communication. The data processing device 100 is not limited to one interface 104, and may have multiple interfaces 104 depending on the number of communication targets.

[0023] The display unit 110 is, for example, a display configured with a liquid crystal panel or the like, and displays the results of data processing performed by the data processing device 100. The input unit 120 is, for example, a pointing device such as a keyboard or a mouse, and receives instructions from a user. When a touch panel is used as the user interface, the display unit 110 and the input unit 120 may be integrally formed. Note that the display unit 110 and the input unit 120 may be configured to be included in the data processing device 100.

[0024] In the data processing system 1 configured as described above, the data processing device 100 acquires, in time series, detection data indicating signal intensities corresponding to components in a sample detected by the chromatograph 10 via the interface 104. The data processing device generates a chromatogram indicating the time change in signal intensity based on the acquired detection data. This allows the data processing device 100 to acquire a chromatogram relating to the components contained in the sample to be analyzed.

[0025] [Example of chromatogram] A chromatograph processed by the data processing device 100 according to the embodiment will be described with reference to Fig. 2. Fig. 2 is a diagram showing three-dimensional vector data composed of a spectrum vector and a chromatogram generated by the data processing device 100 according to the embodiment. In the embodiment described below, a "compound" is used as an example of a "component".

[0026] The data processing device 100 can generate a spectrum vector based on the detection data acquired from the chromatograph 10. For example, if the chromatograph 10 is configured to detect the ion intensity of various compounds contained in the sample as a signal intensity for each mass-to-charge ratio, the data processing device 100 can generate a mass spectrum vector indicating the signal intensity for the mass-to-charge ratio at a specific time timing based on the detection data acquired from the chromatograph 10. Alternatively, if the chromatograph 10 is configured to detect the absorbance of various compounds contained in the sample as a signal intensity for each wavelength, the data processing device 100 can generate a wavelength spectrum vector indicating the signal intensity for the wavelength at a specific time timing based on the detection data acquired from the chromatograph 10.

[0027] 2, the data processing device 100 can generate three-dimensional vector data by arranging the above-mentioned mass spectrum vectors or wavelength spectrum vectors in chronological order for each retention time. The data processing device 100 can generate a chromatogram by plotting retention time on the horizontal axis and signal intensity (ion intensity or absorbance) on the vertical axis. In other words, a chromatogram is data showing the change in signal intensity over time.

[0028] [An example of peak tracking] Peak tracking will be described with reference to Figures 3 and 4. Figure 3 is a diagram for explaining peak tracking using multiple chromatograms. For example, assume that the hydrogen ion concentration index in the solution, which is the mobile phase, is set to multiple indexes between pH 2.5 and pH 6.0, and a chromatogram is generated at each index.

[0029] By comparing multiple chromatograms obtained for each separation condition, such as pH, an analyst can determine the optimal separation conditions for separating compounds contained in a sample. In this case, the analyst can determine the optimal separation conditions by tracking how the retention time of each peak in each chromatogram changes depending on the separation conditions. By performing peak tracking in this way, the analyst can identify peaks of the same compound in each chromatogram.

[0030] Here, in performing peak tracking, it is required to extract two chromatograms having similar peak information from among a plurality of chromatograms, and perform peak tracking between the two extracted chromatograms. The peak information includes, for example, at least one of retention time, peak number, peak width, peak area, peak area ratio, peak height, and peak height ratio. The peak number is the order of the peak detected by the chromatograph 10. The peak width is the width between the time when the rising edge of the peak starts and the time when the falling edge of the peak ends. The peak area is the area of ​​the part surrounded by the peak. The peak area ratio is the ratio of the peak area of ​​the target peak to the total peak area of ​​all peaks included in the same chromatogram. The peak height is the maximum value of the signal intensity of the peak. The peak height ratio is the ratio of the peak height of the target peak to the total peak height of all peaks included in the same chromatogram.

[0031] For example, an analyst selects two chromatograms and matches peaks with similar peak information between the two selected chromatograms by peak tracking. For example, in the example of FIG. 3, among multiple chromatograms, the chromatogram at pH 6.0 and the chromatogram at pH 5.5 have the highest degree of similarity. In this case, the analyst performs peak tracking between the chromatogram at pH 6.0 and the chromatogram at pH 5.5. In this way, an analyst can efficiently investigate optimal separation conditions by matching similar peaks between multiple chromatograms with different separation conditions.

[0032] In addition, when comparing peaks between multiple chromatograms, the analyst can compare the peaks based on the spectrum vectors (mass spectrum vector, wavelength spectrum vector) of each peak. For example, Fig. 4 is a diagram for explaining an example of comparing peaks between multiple chromatograms.

[0033] 4, when comparing peak 1 in chromatogram 1 with peak 1 in chromatogram 2, the analyst compares the shapes of the spectral vectors of peak 1 in chromatogram 1 with peak 1 in chromatogram 2. If the shapes of the spectral vectors of peak 1 in chromatogram 1 and peak 1 in chromatogram 2 are similar, the analyst can predict that peak 1 in chromatogram 1 and peak 1 in chromatogram 2 are peaks corresponding to the same compound.

[0034] [Example of a compound table] The compound table will be described with reference to Fig. 5. Fig. 5 is a diagram for explaining an example of generating the compound table.

[0035] The analyst observes a plurality of chromatograms, estimates similar peaks between the plurality of chromatograms, and registers the estimated peaks in each chromatogram in a compound table as peaks of the same compound.

[0036] For example, as shown in FIG. 5, in compound table 1 of chromatogram 1, the analyst registers peak information of peak 1 as a peak corresponding to compound 1, registers peak information of peak 2 as a peak corresponding to compound 2, and registers peak information of peak 3 as a peak corresponding to compound 3. In compound table 2 of chromatogram 2, the analyst registers peak information of peak 1 as a peak corresponding to compound 1, registers peak information of peak 3 as a peak corresponding to compound 2, and registers peak information of peak 2 as a peak corresponding to compound 3. In compound table 1 of chromatogram 3, the analyst registers peak information of peak 1 as a peak corresponding to compound 1, registers peak information of peak 2 as a peak corresponding to compound 2, and registers peak information of peak 3 as a peak corresponding to compound 3.

[0037] That is, the analyst, as a result of comparing the peaks of chromatograms 1 to 3, finds that peak 1 in chromatogram 1, peak 1 in chromatogram 2, and peak 1 in chromatogram 3 are similar, and predicts that these peaks correspond to compound 1. Furthermore, the analyst, as a result of comparing the peaks of chromatograms 1 to 3, finds that peak 2 in chromatogram 1, peak 3 in chromatogram 2, and peak 2 in chromatogram 3 are similar, and predicts that these peaks correspond to compound 2. Furthermore, the analyst, as a result of comparing the peaks of chromatograms 1 to 3, finds that peak 3 in chromatogram 1, peak 2 in chromatogram 2, and peak 3 in chromatogram 3 are similar, and predicts that these peaks correspond to compound 3.

[0038] In this way, the analyst observes a plurality of chromatograms, estimates similar peaks among the plurality of chromatograms, and registers the estimated peaks of each chromatogram in the compound table as peaks of the same compound. However, in order to generate the compound table, the analyst needs to individually observe at least one peak included in each of the plurality of chromatograms and compare it with at least one peak included in the other chromatograms, which requires a huge amount of work time. In addition, as shown in the example of FIG. 5, even if the same compound is present among the plurality of chromatograms, the peak numbers may be interchanged or there may be many peaks with similar spectral vectors, and the accuracy of the generated compound table may vary depending on the analyst's knowledge or experience.

[0039] Therefore, the data processing device according to the first embodiment is configured to perform rapid and accurate chromatogram data analysis while reducing the workload of the analyst by grouping peaks contained in each of a plurality of chromatograms by clustering based on peak information corresponding to the peaks to generate a plurality of clusters.

[0040] [Creating a compound table by clustering] Generation of a compound table by clustering performed by the data processing device 100 according to the embodiment will be described with reference to Figs. 6 to 12. Fig. 6 is a diagram for explaining an example of a graph in which peak vectors corresponding to each peak included in a plurality of chromatograms are represented in a two-dimensional coordinate system. A "peak vector" is data obtained by converting peak information of each peak into a multidimensional vector in a multidimensional coordinate system in which a plurality of pieces of peak information such as retention time or peak area are taken on different dimensional axes. Hereinafter, a point of a peak vector converted into a multidimensional vector in a multidimensional coordinate system is also referred to as a "peak point".

[0041] 6 shows a graph in which the peak information of each peak contained in a plurality of chromatograms is converted into a two-dimensional vector in a two-dimensional coordinate system in which the horizontal axis represents the peak area and the vertical axis represents the retention time. In this graph, the peak point of the peak corresponding to compound 1 is shown by a circle, the peak point of the peak corresponding to compound 2 is shown by a triangle, and the peak point of the peak corresponding to compound 3 is shown by a square.

[0042] As shown in Figure 6, the peak points corresponding to each compound tend to cluster in similar locations on the graph, because multiple peaks corresponding to the same compound have similar peak information.

[0043] The data processing device 100 can group the peaks by compound type by grouping the peak points of the peaks included in a plurality of chromatograms having the tendency as shown in FIG. 6 by clustering.

[0044] Here, types of clustering algorithms include non-hierarchical clustering and hierarchical clustering.

[0045] Non-hierarchical clustering is a method of grouping the most similar data and classifying them into a specified number of clusters. Non-hierarchical clustering includes the k-means method. For example, the k-means method is a method in which the number of clusters is determined in advance, each data is randomly assigned to one of the clusters, the distance between each data and the center of gravity of each cluster is calculated, and based on the calculated distance, each data is reassigned to the cluster with the smallest distance. Using the k-means method, multiple data that are close to each other can be grouped into the same cluster, but the number of clusters must be determined in advance. For this reason, the analyst must know in advance the number of compounds (i.e., the number of clusters) contained in the sample that is the target of chromatogram detection. If the number of compounds (the number of clusters) contained in the sample is not known in advance, the k-means method cannot be used.

[0046] Hierarchical clustering is a method that generates one cluster by grouping two data sets with the smallest inter-data distance, and also generates one cluster by grouping two clusters with the smallest inter-cluster distance among multiple clusters. Hierarchical clustering is classified into the shortest distance method, the longest distance method, and the center of gravity method depending on how the distance between clusters is defined.

[0047] An example of the application of hierarchical clustering according to the centroid method to grouping of peak points in a chromatogram will be described with reference to Fig. 7. Fig. 7 is a diagram for explaining an application example of hierarchical clustering according to the centroid method. In the example shown in Fig. 7, peak points A to E corresponding to a plurality of peaks A to E are shown in a two-dimensional coordinate system in which the horizontal axis represents the peak area and the vertical axis represents the retention time.

[0048] According to the center of gravity method, as shown in Fig. 7(A), among peak points A to E, peak points A and B, which have the smallest distance between the two points, are grouped together to generate one cluster (A, B). Similarly, among peak points A to E, peak points C and D, which have the smallest distance, are grouped together to generate one cluster (C, D). As a result, peak points A to E are divided into a cluster (A, B) consisting of peak points A and B, a cluster (C, D) consisting of peak points C and D, and peak point E.

[0049] 7B, the cluster (A, B) and the cluster (C, D) with the smallest inter-point distance among the center of gravity of the cluster (A, B) (for example, the center position between peak point A and peak point B), the center of gravity of the cluster (C, D) (for example, the center position between peak point C and peak point D), and peak point E are combined to generate one cluster (A, B, C, D). As a result, peak points A to D are divided into a cluster (A, B, C, D) consisting of peak point A, peak point B, peak point C, and peak point D, and peak point E.

[0050] Next, as shown in FIG. 7C, the cluster (A, B, C, D) and the peak point E are combined to generate one cluster (A, B, C, D, E).

[0051] In this way, if hierarchical clustering according to the centroid method is applied directly to grouping peaks in a chromatogram, the peaks contained in multiple chromatograms will end up being grouped into one cluster. In other words, this result means that the peaks contained in multiple chromatograms correspond to the same compound, which is an impossible result for a chromatogram of a sample containing multiple compounds. This is also true when using hierarchical clustering methods other than the centroid method.

[0052] Therefore, as shown in FIG. 8, the data processing device 100 according to the first embodiment is configured to prohibit multiple peaks that satisfy a specific condition based on peak information from being grouped in the same cluster when applying the hierarchical clustering to grouping of peaks in a chromatogram. The specific condition includes at least one of the following: multiple peaks to be grouped are included in the same chromatogram; the spectra of the multiple peaks to be grouped are different; and the difference in peak information of the multiple peaks to be grouped exceeds a predetermined range. More specifically, the data processing device 100 prohibits multiple peaks included in the same chromatogram from being grouped in the same cluster in the hierarchical clustering. Furthermore, the data processing device 100 prohibits multiple peaks having different mass spectra (MS spectra) or wavelength spectra (PDA spectra) from being grouped in the same cluster in the hierarchical clustering. Furthermore, the data processing device 100 prohibits multiple peaks having a difference in peak information (such as peak area or peak height) that exceeds a predetermined range from being grouped in the same cluster in the hierarchical clustering. The specified range can be set by the analyst, and when the difference in peak information is within the specified range, it is more likely that the multiple peaks correspond to the same compound than when the difference in peak information exceeds the specified range.

[0053] Fig. 8 is a diagram for explaining an application example of hierarchical clustering according to the centroid method executed by the data processing device 100 according to the embodiment. Fig. 8 shows an example in which the data processing device 100 performs grouping by hierarchical clustering on peak points 1 to 3 corresponding to three peaks 1 to 3 in chromatograms 1 to 3, respectively, to generate a plurality of clusters.

[0054] 8, the data processing device 100 generates one cluster 1A by grouping peak point 1 of chromatogram 2 and peak point 2 of chromatogram 3. Furthermore, the data processing device 100 generates one cluster 1 by grouping peak point 1 of chromatogram 1 and cluster 1A.

[0055] The data processing device 100 groups the peak point 2 of the chromatogram 1 and the peak point 3 of the chromatogram 2 to generate one cluster 2A. Furthermore, the data processing device 100 groups the peak point 1 of the chromatogram 3 and the cluster 2A to generate one cluster 1.

[0056] The data processing device 100 groups the peak point 2 of the chromatogram 2 and the peak point 3 of the chromatogram 3 to generate one cluster 3A. Furthermore, the data processing device 100 groups the peak point 3 of the chromatogram 1 and the cluster 3A to generate one cluster 3.

[0057] Cluster 1 is grouped by selecting only one peak point from each of chromatograms 1 to 3, such as peak point 1 in chromatogram 1, peak point 1 in chromatogram 2, and peak point 2 in chromatogram 3. In other words, peak 1 in chromatogram 1, peak 1 in chromatogram 2, and peak 2 in chromatogram 3 are highly likely to be peaks corresponding to the same compound.

[0058] Cluster 2 is grouped by selecting only one peak point from each of chromatograms 1 to 3, such as peak point 2 in chromatogram 1, peak point 3 in chromatogram 2, and peak point 1 in chromatogram 3. In other words, peak 2 in chromatogram 1, peak 3 in chromatogram 2, and peak 1 in chromatogram 3 are highly likely to be peaks corresponding to the same compound.

[0059] Cluster 3 is grouped by selecting only one peak point from each of chromatograms 1 to 3, such as peak point 3 in chromatogram 1, peak point 2 in chromatogram 2, and peak point 3 in chromatogram 3. In other words, peak 3 in chromatogram 1, peak 2 in chromatogram 2, and peak 3 in chromatogram 3 are highly likely to be peaks corresponding to the same compound.

[0060] The data processing device 100 can generate a compound table as shown in Fig. 5 using the plurality of clusters thus generated. For example, the data processing device 100 sets a compound corresponding to each peak included in cluster 1 as compound 1, inputs peak information of peak 1 in chromatogram 1 into the compound 1 column of compound table 1, inputs peak information of peak 1 in chromatogram 2 into the compound 1 column of compound table 2, and inputs peak information of peak 2 in chromatogram 3 into the compound 1 column of compound table 3. The data processing device 100 sets a compound corresponding to each peak included in cluster 2 as compound 2, inputs peak information of peak 2 in chromatogram 1 into the compound 2 column of compound table 1, inputs peak information of peak 3 in chromatogram 2 into the compound 2 column of compound table 2, and inputs peak information of peak 1 in chromatogram 3 into the compound 2 column of compound table 3. The data processing device 100 regards a compound corresponding to each peak included in cluster 3 as compound 3, inputs peak information of peak 3 in chromatogram 1 into the compound 3 column of compound table 1, inputs peak information of peak 2 in chromatogram 2 into the compound 3 column of compound table 2, and inputs peak information of peak 3 in chromatogram 3 into the compound 3 column of compound table 3. In this way, the data processing device 100 can quickly and accurately generate compound tables corresponding to each of chromatograms 1 to 3 while reducing the workload of the analyst.

[0061] The data processing device 100 may generate one compound table by collecting multiple chromatograms, not limited to the compound table for each chromatogram as shown in FIG. 5. For example, the data processing device 100 may calculate a representative value for each of the clusters 1 to 3 and generate a table that collects the representative values ​​for each of the multiple clusters. Specifically, the data processing device 100 calculates the representative value for the cluster 1 based on the peak information of each of the peaks 1 in the chromatogram 1, 1 in the chromatogram 2, and 2 in the chromatogram 3, which are included in the cluster 1. The data processing device 100 calculates the representative value for the cluster 2 based on the peak information of each of the peaks 2 in the chromatogram 1, 3 in the chromatogram 2, and 1 in the chromatogram 3, which are included in the cluster 2. The data processing device 100 calculates the representative value for the cluster 3 based on the peak information of each of the peaks 3 in the chromatogram 1, 2 in the chromatogram 2, and 3 in the chromatogram 3, which are included in the cluster 3. The data processing device 100 can generate one compound table by entering a representative value of cluster 1 into the compound 1 column of the compound table, a representative value of cluster 2 into the compound 2 column of the compound table, and a representative value of cluster 3 into the compound 3 column of the compound table.

[0062] The representative value of each cluster may be the average value or the median value of the peak information of the peaks included in each cluster. For example, the representative value may be the average value or median value of the retention time, the average value or median value of the peak width, the average value or median value of the peak area, the average value or median value of the peak area ratio, the average value or median value of the peak height, the average value or median value of the peak height, etc., of the peaks included in each cluster.

[0063] In this way, the data processing device 100 generates clusters 1 to 3 by grouping peaks 1 to 3 included in each of the chromatograms 1 to 3 by hierarchical clustering based on peak information corresponding to each of the peaks 1 to 3. At this time, the data processing device 100 generates clusters 1 to 3 while prohibiting grouping multiple peaks included in the same chromatogram into the same cluster. In other words, the data processing device 100 prohibits grouping each of the clusters 1 to 3 with other clusters. The data processing device 100 generates a compound table based on peak information of peaks included in each of the multiple clusters generated in this way. As a result, the data processing device 100 can generate a compound table by performing chromatogram data analysis quickly and accurately while reducing the workload of the analyst.

[0064] [Data processing flow of data processing device 100] The flow of data processing executed by the data processing device 100 will be described with reference to Fig. 9. Fig. 9 is a flowchart of data processing executed by the data processing device 100 according to an embodiment. The processing steps (hereinafter abbreviated as "S") shown in Fig. 9 are realized by the calculation device 101 executing the data processing program 130.

[0065] 9, the data processing device 100 acquires (S1) detection data indicating signal intensities corresponding to components in a sample detected by the chromatograph 10. The data processing device 100 generates a chromatogram indicating the change in signal intensity over time based on the detection data acquired from the chromatograph 10 (S2).

[0066] The data processing device 100 performs hierarchical clustering on the multiple chromatograms generated in the process of S2 to generate multiple clusters (S3). At this time, the data processing device 100 prohibits multiple peaks included in the same chromatogram from being grouped into the same cluster.

[0067] The data processing device 100 generates a compound table based on peak information of at least one peak included in each of the multiple clusters generated in the process of S3 (S4). After that, the data processing device 100 ends this process.

[0068] As described above, the data processing device 100 generates a plurality of chromatograms based on the detection data acquired from the chromatograph 10, groups the peaks included in each of the plurality of chromatograms based on the peak information corresponding to the peaks by hierarchical clustering to generate a plurality of clusters, and generates a compound table using the generated plurality of clusters. Furthermore, the data processing device 100 prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster. Specifically, in the hierarchical clustering, the data processing device 100 prohibits, in the same clustering, grouping a plurality of peaks included in the same analysis data, a plurality of peaks with different spectra (MS spectrum, PDA spectrum), or a plurality of peaks whose difference in peak information (such as peak area or peak height) exceeds a predetermined range. This allows the data processing device 100 to quickly and accurately perform data analysis of the chromatograms while reducing the workload of the analyst and generate a compound table.

[0069] [Variations] The present disclosure is not limited to the above-described embodiment, and various modifications and applications are possible. Modifications that can be applied to the present disclosure will be described below.

[0070] (Variation 1) 6, the data processing device 100 converts the peak information of each peak included in a plurality of chromatograms into a two-dimensional vector in a two-dimensional coordinate system in which the horizontal axis represents the peak area and the vertical axis represents the retention time, but the coordinate system may be three or more dimensional instead of two-dimensional. The data processing device 100 may generate a multidimensional coordinate system using other peak information such as a peak number, a peak width, a peak area, a peak area ratio, a peak height, or a peak height ratio, in addition to the peak area and retention time.

[0071] (Variation 2) When an analyst manually creates a compound table, the analyst not only observes the peak to be observed, but also the peak that appears before the peak in the chromatogram (the peak with the previous peak number) or the peak that appears before the peak (the peak with the next peak number) to comprehensively estimate the compound that corresponds to the peak.

[0072] Therefore, when generating a multidimensional coordinate system, the data processing device 100 may generate the multidimensional coordinate system by taking, as the coordinate axes, peak information of a peak that appears before the peak (the peak that is the target of hierarchical clustering) corresponding to the peak point plotted in the multidimensional coordinate system, or peak information of a peak that appears after the peak. This allows the data processing device 100 to generate multiple clusters by hierarchical clustering, taking into account the context of multiple peaks that appear in the chromatogram.

[0073] (Variation 3) When generating a multidimensional coordinate system, the data processing device 100 may weight at least one piece of peak information. For example, Fig. 10 is a diagram for explaining an example of weighting of peak information in hierarchical clustering.

[0074] As shown in FIG. 10, the data processing device 100 may weight the area more than the retention time among the peak information, placing emphasis on the area rather than the retention time. For example, the data processing device 100 may weight the area more than the retention time by multiplying the scale of the area on the horizontal axis by a constant while not multiplying the scale of the retention time on the vertical axis by a constant. In this way, in a two-dimensional coordinate system generated based on the weighted area and the unweighted retention time, the horizontal axis corresponding to the area is more dominant in the calculation of the distance between points than the vertical axis corresponding to the retention time. In the weighted two-dimensional coordinate system, the data processing device 100 converts the peak information of each peak included in a plurality of chromatograms into a two-dimensional vector and performs hierarchical clustering, thereby performing hierarchical clustering so that the peak information to be emphasized is more influenced.

[0075] (Variation 4) When calculating the distance between multiple peak points plotted in a multidimensional coordinate system, the data processing device 100 may calculate the difference in real numbers (scalars) of peak information of each peak point, such as retention time, peak area, peak width, etc., and group two peak points with the smallest calculated difference. For example, the data processing device 100 may calculate the difference in real numbers (scalars) of peak information between peak point a and peak point B by calculating the following formula (1).

number

[0076] The data processing device 100 may group the peak points a and b when the calculation result calculated by the formula (1) is the smallest.

[0077] (Variation 5) The peak points plotted in the multidimensional coordinate system are points of the peak vectors represented in the multidimensional coordinate system. Therefore, when calculating the distance between the points of a plurality of peak points plotted in the multidimensional coordinate system, the data processing device 100 may calculate the difference between the peak vectors corresponding to each peak point, and group two peak points with the smallest calculated difference. For example, the data processing device 100 may calculate the difference between the peak vector corresponding to peak point a and the peak vector corresponding to peak point b by calculating the following formula (2).

number

[0078] In addition, in formula (2), θ is the angle between the peak vector corresponding to peak point a and the peak vector corresponding to peak point b. When the calculation result calculated by formula (2) is minimum, the data processing device 100 may group the peak point a and the peak point b.

[0079] (Variation 6) Fig. 11 is a diagram for explaining an example of a comparison of precision and recall in generating a compound table. Fig. 11 shows a correspondence relationship between a correct answer and a prediction when predicting which of compounds A to C each peak corresponds to for 30 peaks contained in at least one chromatogram. In the example shown in Fig. 11, 10 peaks each correspond to compounds A to C among the 30 peaks.

[0080] In the example shown in Fig. 11(A), of the 30 peaks, 10 peaks corresponding to compound A are grouped in cluster 1, 10 peaks corresponding to compound B are grouped in cluster 2, and 10 peaks corresponding to compound C are grouped in cluster 3. In this way, in the example of Fig. 11(A), the 10 peaks corresponding to each of compounds A to C are grouped evenly into clusters 1 to 3. Such a compound table can be said to have both high precision and recall.

[0081] In the example shown in FIG. 11(B), of the 30 peaks, 6 peaks corresponding to compound A are grouped in cluster 1, 4 peaks corresponding to compound A are grouped in cluster 2, 10 peaks corresponding to compound B are grouped in cluster 3, and 10 peaks corresponding to compound C are grouped in cluster 4. Thus, in the example of FIG. 11(B), the 10 peaks corresponding to compounds B and C are grouped evenly into clusters 3 and 4, while the 10 peaks corresponding to compound A are grouped not into one cluster but into two clusters 1 and 2. It can be said that such a compound table has a high precision rate but a low recall rate.

[0082] In the example shown in Fig. 11(C), of the 30 peaks, 10 peaks corresponding to compound A are grouped in cluster 1, and 10 peaks corresponding to compound B and 10 peaks corresponding to compound C are grouped in cluster 2. Thus, in the example of Fig. 11(C), the 10 peaks corresponding to compound A are grouped in cluster 1, while the 10 peaks corresponding to each of compounds B and C are grouped in one cluster 2 without being divided into two clusters. Such a compound table can be said to have a low precision rate but a high recall rate.

[0083] In the example shown in FIG. 11(D), of the 30 peaks, three peaks corresponding to compound A, four peaks corresponding to compound B, and three peaks corresponding to compound C are grouped in cluster 1, three peaks corresponding to compound A, three peaks corresponding to compound B, and four peaks corresponding to compound C are grouped in cluster 2, and five peaks corresponding to compound A, two peaks corresponding to compound B, and three peaks corresponding to compound C are grouped in cluster 3. Thus, in the example shown in FIG. 11(D), the ten peaks corresponding to each of compounds A to C are grouped so as to be included in any of clusters 1 to 3. It can be said that such a compound table has low precision and recall.

[0084] Here, when comparing a compound table with a high precision rate but a low recall rate as shown in FIG. 11(B) with a compound table with a low precision rate but a high recall rate as shown in FIG. 11(C), the compound table in FIG. 11(B) is more preferable than the compound table in FIG. 11(c) in the analysis of chromatogram data. The reason for this is that, in a cluster generated by hierarchical clustering, when peaks corresponding to a plurality of different types of compounds are mixed as shown in FIG. 11(C), the analyst cannot identify a compound based on the cluster, which causes inconvenience in practical use. On the other hand, in a cluster generated by hierarchical clustering, when the number of data is small but only peaks corresponding to one type of compound are mixed as shown in FIG. 11(B), the analyst can identify a compound based on the cluster.

[0085] Therefore, the precision rate is more important than the recall rate in generating the compound table by the data processing device 100. Therefore, in order to improve the precision rate, the data processing device 100 is configured to execute several processes as described below.

[0086] As a measure to improve the matching rate, the data processing device 100 may prohibit the generation of a compound table using one of the clusters according to the number of peaks included in the cluster. For example, when the data processing device 100 obtains a clustering result as shown in FIG. 11(B), the data processing device 100 may prohibit the generation of a compound table using cluster 2, which has a smaller number of data, among clusters 1 and 2, which include peaks corresponding to the same compound A. Alternatively, the data processing device 100 may prohibit the generation of a compound table using one of the clusters when the number of peaks included in one of the clusters is less than a predetermined number. For example, when the data processing device 100 obtains a clustering result as shown in FIG. 11(B), the data processing device 100 may prohibit the generation of a compound table using cluster 2, which has less than five data. The predetermined number, which is a threshold value, may be set by the analyst.

[0087] As a measure to improve the matching rate, the data processing device 100 may exclude from one cluster a peak close to a peak included in another cluster. For example, FIG. 12 is a diagram for explaining an example of excluding peaks from a cluster in hierarchical clustering. As shown in FIG. 12, a peak point surrounded by a dashed line is a predetermined distance or more away from other peak points corresponding to the same type of compound. Such a peak point may have been generated by an error in waveform processing or a false detection when generating a chromatogram, and if hierarchical clustering is performed using such a peak point, the matching rate may decrease. For this reason, it is preferable that the data processing device 100 performs clustering excluding peak points that are a predetermined distance or more away from other peak points corresponding to the same type of compound, such as the peak points surrounded by the dashed line. In the example of FIG. 12, the data processing device 100 may exclude a peak point indicated by a circle surrounded by a dashed line from a cluster corresponding to compound 1, exclude a peak point indicated by a triangle surrounded by a dashed line from a cluster corresponding to compound 2, and exclude a peak point indicated by a square surrounded by a dashed line from a cluster corresponding to compound 3.

[0088] As a measure to improve the matching rate, the data processing device 100 may extract peaks for generating a compound table by calculating the following formula (3).

number

[0089] In the above formula (3), the first term indicates the sum of the reciprocals of the distances between the target peak point included in one cluster and the other peak points included in the other clusters, and the second term indicates the sum of the reciprocals of the distances between the target peak point included in one cluster and the other peaks included in the one cluster. When the value pi calculated by formula (3) is negative, it can be said that the target peak point is close to the other peak points in its own grouped cluster and far from the other peak points in the other clusters. On the other hand, when the value pi calculated by formula (3) is positive, it can be said that the target peak point is far from the other peak points in its own grouped cluster and close to the other peak points in the other clusters. For this reason, the data processing device 100 may extract only the peak points for which the value Pi calculated by formula (3) is negative, and generate a compound table using only the extracted peak points.

[0090] The above formula (3) may be rewritten as the following formula (4).

number

[0091] In the above formula (4), the first term indicates the sum of values ​​obtained by transforming the distance between a target peak point included in one cluster and other peak points included in other clusters using a function f that monotonically decreases with respect to the distance, and the second term indicates the sum of values ​​obtained by transforming the distance between a target peak point included in one cluster and other peak points included in the same cluster using a function f that monotonically decreases with respect to the distance.

[0092] Moreover, the above formulas (3) and (4) may be rewritten as the following formula (5).

number

[0093] In the above formula (5), the first term indicates the sum of the distances between a target peak point included in one cluster and other peak points included in other clusters, transformed using a function f that monotonically decreases with respect to the distances, and the second term indicates the sum of the values ​​of the distances between a target peak point included in one cluster and other peak points included in the same cluster, transformed using a function f that monotonically decreases with respect to the distances.

[0094] (Variation 7) When calculating the representative value of a cluster, the data processing device 100 may identify a representative peak from among the peaks included in the cluster, and calculate the representative value of the cluster based on peak information of the representative peak. Fig. 13 is a diagram for explaining an example of calculation of the representative value in each cluster grouped by hierarchical clustering.

[0095] For example, as shown in FIG. 13, the data processing device 100 may extract a medoid point of each cluster based on peak information of a peak point included in each cluster corresponding to compounds 1 to 3 as a representative value. The data processing device 100 may calculate the peak information of the extracted medoid point as a representative value. Note that a medoid point is a peak point included in one cluster that has a minimum sum of distances between the peak point and other peaks included in the same cluster (for example, the calculation result of the second term in formula (5) when f(x)=x).

[0096] Here, for a medoid point, the distance difference between the medoid point and other peak points in the cluster to which it belongs is taken into consideration, but the distance difference between the medoid point and other peak points in other clusters (for example, the calculation result of the first term in formulas (4) and (5)) is not taken into consideration. For this reason, there may be cases where the medoid point is close to the cluster corresponding to compound 1, such as the medoid point of the cluster corresponding to compound 2 in FIG. 13.

[0097] Therefore, it is preferable that the data processing device 100 determines the peak point at which the value Pi calculated by the above-mentioned formulas (3) to (5) is the smallest as the representative value of each cluster, and calculates the representative value of the cluster based on the peak information of the representative peak point. In this way, the data processing device 100 extracts the peak point at which the value Pi calculated by the formulas (3) to (5) is the smallest in each cluster, and generates a compound table by collecting the representative values ​​of the extracted representative peak points, thereby generating a compound table with high accuracy.

[0098] (Variation 8) In the above-described embodiment, a chromatogram is used as the analytical data, but the data processing device 100 is also applicable to analytical data having a plurality of peaks other than a chromatogram.

[0099] The above-described embodiment and modifications can be appropriately combined and applied to one data processing system 1 and data processing device 100.

[0100] [Aspects] It will be appreciated by those skilled in the art that the exemplary embodiments described above are examples of the following aspects.

[0101] (Item 1) A data processing device according to one embodiment includes a data acquisition unit that acquires detection data indicating signal intensity corresponding to a component in a sample detected by a detection device, and a calculation unit that processes the detection data acquired by the data acquisition unit. The calculation unit generates a plurality of analysis data including signal intensity peaks based on the detection data, groups the peaks included in each of the plurality of analysis data based on peak information corresponding to the peaks by hierarchical clustering to generate a plurality of clusters, and prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster.

[0102] According to the data processing device 100 described in paragraph 1, peaks included in each of a plurality of analytical data can be grouped by hierarchical clustering based on peak information corresponding to the peaks to generate a plurality of clusters, so that the analyst does not need to estimate peaks common to a plurality of analytical data by himself / herself to generate a plurality of clusters, and can perform data analysis quickly. Furthermore, the data processing device 100 prohibits grouping a plurality of peaks that satisfy a specific condition based on peak information into the same cluster in the hierarchical clustering, so that a plurality of peaks that satisfy a specific condition based on peak information are not grouped into the same cluster as the same component, and data analysis can be performed with high accuracy using hierarchical clustering.

[0103] (2) In the data processing device according to 1, the detection device is a chromatograph. Each of the plurality of analytical data is a chromatogram showing a peak of signal intensity versus retention time.

[0104] According to the data processing device described in paragraph 2, the analyst does not need to estimate peaks common to multiple chromatograms and generate multiple clusters, and can perform data analysis quickly and accurately.

[0105] (Item 3) In the data processing device according to item 1 or 2, the calculation unit generates a table based on peak information corresponding to peaks included in each of the plurality of clusters.

[0106] According to the data processing device described in paragraph 3, the analyst does not need to generate a compound table by estimating peaks common to a plurality of chromatograms, and can generate a compound table quickly and accurately.

[0107] (4) In the data processing device described in 3, the calculation unit calculates a representative value for each of the plurality of clusters, and generates a table based on the representative value for each of the plurality of clusters.

[0108] According to the data processing device described in the fourth aspect, the analyst can generate a compound table based on the representative values ​​of each of a plurality of clusters.

[0109] (Clause 5) In the data processing device described in clause 3 or 4, the calculation unit prohibits generating a table using one of the multiple clusters depending on the number of peaks contained in the one cluster.

[0110] According to the data processing device described in paragraph 5, for example, an analyst can prohibit generating a compound table using cluster 2 which contains a small number of peaks in one cluster, thereby improving the accuracy rate in generating a compound table.

[0111] (Clause 6) In the data processing device described in any one of clauses 3 to 5, the calculation unit generates a table using peaks extracted based on the difference between a peak included in one cluster and another peak included in the one cluster, and the difference between the peak included in the one cluster and another peak included in the other cluster.

[0112] According to the data processing device described in paragraph 6, an analyst can generate a compound table using peaks extracted based on the difference between a peak included in one cluster and another peak included in the cluster, and the difference between a peak included in the one cluster and another peak included in the cluster, thereby improving the accuracy rate in generating the compound table.

[0113] (Clause 7) In the data processing device described in clause 4, the calculation unit calculates a representative value using peaks extracted based on the difference between a peak included in a cluster and another peak included in the cluster, and the difference between the peak included in the cluster and a peak included in the other cluster.

[0114] According to the data processing device described in paragraph 7, an analyst can generate a compound table based on a representative value calculated using peaks extracted based on a difference between a peak included in one cluster and another peak included in the one cluster, and a difference between a peak included in the one cluster and another peak included in the one cluster.

[0115] (Clause 8) In the data processing device described in any one of clauses 1 to 7, the calculation unit weights at least one of the peak information more heavily than the other peak information, and performs hierarchical clustering based on the weighted peak information.

[0116] According to the data processing device described in Section 8, the analyst can perform hierarchical clustering so that peak information that the analyst desires to emphasize is more influenced.

[0117] (Item 9) In the data processing device according to any one of items 1 to 8, the calculation unit excludes from one cluster a peak that is close to a peak included in another cluster.

[0118] According to the data processing device described in paragraph 9, an analyst can generate a compound table by excluding from a cluster a peak that is close to a peak included in another cluster, thereby improving the accuracy rate in generating the compound table.

[0119] (Item 10) In the data processing device according to any one of items 1 to 9, the peak information includes at least one of the retention time, number, width, area, area ratio, height, and height ratio of the peak.

[0120] According to the data processing device described in paragraph 10, an analyst can generate a compound table based on at least one of the retention time, number, width, area, area ratio, height, and height ratio of peaks contained in a chromatogram.

[0121] (Clause 11) In a data processing device described in any one of clauses 1 to 10, the peak information includes at least one of the retention time, number, width, area, area ratio, height, and height ratio of a peak before or after the peak to be subjected to hierarchical clustering.

[0122] According to the data processing device described in paragraph 11, an analyst can generate a compound table by generating a plurality of clusters through hierarchical clustering, taking into consideration the context of a plurality of peaks appearing in a chromatogram.

[0123] (Item 12) In the data processing device according to any one of items 1 to 11, the hierarchical clustering is clustering according to at least one of a shortest distance method, a longest distance method, and a centroid method.

[0124] According to the data processing device described in paragraph 12, an analyst can generate a compound table using a plurality of clusters generated by hierarchical clustering according to at least one of the shortest distance method, the longest distance method, and the center of gravity method.

[0125] (Clause 13) In the data processing device described in any one of clauses 1 to 12, the specific conditions include at least one of: multiple peaks are included in the same analysis data; the spectra of the multiple peaks are different; and the difference in peak information among the multiple peaks exceeds a predetermined range.

[0126] According to the data processing device 100 described in paragraph 13, in hierarchical clustering, multiple peaks contained in the same analytical data, multiple peaks with different spectra, or multiple peaks whose peak information difference exceeds a predetermined range are not grouped in the same cluster as the same component, and data analysis can be performed with high accuracy using hierarchical clustering.

[0127] (Clause 14) A data processing method according to one embodiment includes, as processing executed by a computer (a calculation unit), a step of acquiring detection data indicating signal intensity corresponding to a component in a sample detected by a detection device, and a step of processing the detection data acquired by the acquiring step. The step of processing the detection data includes a step of generating a plurality of analysis data including signal intensity peaks based on the detection data, and a step of grouping peaks included in each of the plurality of analysis data by hierarchical clustering based on peak information corresponding to the peaks to generate a plurality of clusters. The step of generating a plurality of clusters prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster.

[0128] According to the data processing method described in paragraph 14, the computer can generate multiple clusters by grouping peaks included in each of the multiple analytical data based on peak information corresponding to the peaks by hierarchical clustering, so that the analyst does not need to estimate peaks common to the multiple analytical data and generate multiple clusters, and can perform data analysis quickly. Furthermore, the computer prohibits multiple peaks that satisfy a specific condition based on peak information from being grouped into the same cluster in the hierarchical clustering, so that multiple peaks that satisfy a specific condition based on peak information are not grouped into the same cluster as the same component, and data analysis can be performed with high accuracy using hierarchical clustering.

[0129] (Clause 15) A data processing program according to one embodiment causes a computer (operation unit) to execute the steps of acquiring detection data indicating signal intensity corresponding to components in a sample detected by a detection device, and processing the detection data acquired by the acquiring step. The step of processing the detection data includes the steps of generating a plurality of analysis data including signal intensity peaks based on the detection data, and grouping peaks included in each of the plurality of analysis data by hierarchical clustering based on peak information corresponding to the peaks to generate a plurality of clusters. The step of generating a plurality of clusters prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster.

[0130] According to the data processing program described in paragraph 15, the computer can generate multiple clusters by grouping peaks included in each of the multiple analytical data based on peak information corresponding to the peaks by hierarchical clustering, so that the analyst does not need to estimate peaks common to the multiple analytical data and generate multiple clusters, and can perform data analysis quickly. Furthermore, the computer prohibits grouping multiple peaks that satisfy a specific condition based on peak information into the same cluster in the hierarchical clustering, so that multiple peaks that satisfy a specific condition based on peak information are not grouped into the same cluster as the same component, and data analysis can be performed with high accuracy using hierarchical clustering.

[0131] (Clause 16) A data processing system according to one embodiment includes a detection device and a data processing device that processes data. The data processing device includes a data acquisition unit that acquires detection data indicating signal intensity corresponding to a component in a sample detected by the detection device, and a calculation unit that processes the detection data acquired by the data acquisition unit. The calculation unit generates a plurality of analysis data including peaks of signal intensity based on the detection data, groups the peaks included in each of the plurality of analysis data based on peak information corresponding to the peaks by hierarchical clustering to generate a plurality of clusters, and prohibits, in the hierarchical clustering, grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster.

[0132] According to the data processing system described in paragraph 16, peaks included in each of a plurality of analytical data can be grouped by hierarchical clustering based on peak information corresponding to the peaks to generate a plurality of clusters, so that the analyst does not need to estimate peaks common to a plurality of analytical data by himself / herself to generate a plurality of clusters, and can perform data analysis quickly. Furthermore, the data processing device 100 prohibits grouping a plurality of peaks that satisfy a specific condition based on peak information into the same cluster in the hierarchical clustering, so that a plurality of peaks that satisfy a specific condition based on peak information are not grouped into the same cluster as the same component, and data analysis can be performed with high accuracy using hierarchical clustering. [Explanation of symbols]

[0133] 1 data processing system, 10 chromatograph, 11 container, 12 liquid delivery pump, 13 injector, 14 column, 15 detector, 100 data processing device, 101 computing device, 102 memory, 103 storage device, 104 interface, 110 display unit, 120 input unit, 130 data processing program.

Claims

1. A data processing device for processing data, a data acquisition unit that acquires detection data indicating a signal intensity corresponding to a component in a sample detected by the detection device; a calculation unit that processes the detection data acquired by the data acquisition unit, The calculation unit is generating a plurality of analysis data including the peaks of the signal intensity based on the detection data; generating a plurality of clusters by grouping the peaks included in each of the plurality of analytical data by hierarchical clustering based on peak information corresponding to the peaks; The data processing device prohibits grouping, in the hierarchical clustering, a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster.

2. the detection device is a chromatograph; 2. The data processing apparatus according to claim 1, wherein each of the plurality of analytical data is a chromatogram showing a peak of the signal intensity versus retention time.

3. The data processing apparatus according to claim 1 , wherein the calculation unit generates a table based on the peak information corresponding to peaks included in each of the plurality of clusters.

4. The calculation unit is Calculating a representative value for each of the plurality of clusters; The data processing apparatus according to claim 3 , wherein the table is generated based on the representative value of each of the plurality of clusters.

5. 5. The data processing apparatus according to claim 3, wherein the calculation unit prohibits one of the plurality of clusters from using the cluster to generate the table in accordance with the number of peaks contained in the one cluster.

6. 5. The data processing device according to claim 3, wherein the calculation unit generates the table using peaks extracted based on a difference between a peak included in a cluster and another peak included in the cluster, and a difference between a peak included in the cluster and a peak included in the other cluster.

7. 5. The data processing device according to claim 4, wherein the calculation unit calculates the representative value using peaks extracted based on a difference between a peak included in a cluster and another peak included in the cluster, and a difference between a peak included in the cluster and a peak included in the other cluster.

8. 3 . The data processing device according to claim 1 , wherein the calculation unit weights at least one of the pieces of peak information more heavily than other pieces of peak information, and performs the hierarchical clustering based on the weighted peak information. 4 .

9. 3 . The data processing device according to claim 1 , wherein the calculation unit excludes, from a cluster, a peak that is close to a peak that is included in another cluster. 4 .

10. The data processing apparatus according to claim 2 , wherein the peak information includes at least one of the retention time, number, width, area, area ratio, height, and height ratio of a peak.

11. The data processing apparatus according to claim 2 , wherein the peak information includes at least one of the retention time, number, width, area, area ratio, height, and height ratio of a peak preceding or succeeding the peak to be subjected to the hierarchical clustering.

12. 3. The data processing apparatus according to claim 1, wherein the hierarchical clustering is clustering according to at least one of a shortest distance method, a longest distance method, and a centroid method.

13. 3. The data processing device according to claim 1, wherein the specific condition includes at least one of the following: the plurality of peaks are included in the same analysis data; the spectra of the plurality of peaks are different; and a difference in the peak information among the plurality of peaks exceeds a predetermined range.

14. A data processing method for a computer to process data, comprising the steps of: The process executed by the computer is obtaining detection data indicative of signal intensities corresponding to components in the sample detected by the detection device; and processing the detection data acquired by the acquiring step. The step of processing the detection data includes: generating a plurality of analysis data including the peaks of the signal intensity based on the detection data; and generating a plurality of clusters by grouping peaks included in each of the plurality of analytical data by hierarchical clustering based on peak information corresponding to the peaks, A data processing method, wherein the step of generating a plurality of clusters prohibits grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster in the hierarchical clustering.

15. A data processing program for processing data, On the computer, obtaining detection data indicative of signal intensities corresponding to components in the sample detected by the detection device; and processing the detection data acquired by the acquiring step. The step of processing the detection data includes: generating a plurality of analysis data including the peaks of the signal intensity based on the detection data; and generating a plurality of clusters by grouping peaks included in each of the plurality of analytical data by hierarchical clustering based on peak information corresponding to the peaks, The step of generating a plurality of clusters prohibits grouping a plurality of peaks that satisfy a specific condition based on the peak information into the same cluster in the hierarchical clustering.

16. A data processing system for processing data, A detection device; a data processing device for processing data, The data processing device includes: a data acquisition unit that acquires detection data indicating a signal intensity corresponding to a component in a sample detected by the detection device; a calculation unit that processes the detection data acquired by the data acquisition unit, The calculation unit is generating a plurality of analysis data including the peaks of the signal intensity based on the detection data; generating a plurality of clusters by grouping the peaks included in each of the plurality of analytical data by hierarchical clustering based on peak information corresponding to the peaks; In the hierarchical clustering, a plurality of peaks that satisfy a specific condition based on the peak information are prohibited from being grouped in the same cluster.

Images