Computer program, information processing device, and information processing method

Abstract

The present invention provides a computer program or the like, capable of improving the utilization degree of a detection signal detected by using an optical sensor.　A computer program according to one aspect of the present disclosure causes a computer to execute processing for: acquiring a first profile indicating a change in interaction between substances detected by using an optical sensor; Fourier-transforming the acquired first profile to acquire a spectrum for each frequency; inverse-Fourier-transforming the spectrum of a specific frequency among the acquired spectra for the respective frequencies to generate a second profile; and acquiring a classification of a group to which the second profile belongs when the second profile has been classified into a plurality of groups.

Description

Computer program, information processing device, and information processing method

[0001] The present invention relates to a computer program, an information processing device, and an information processing method.

[0002] Optical sensors are used to analyze interactions between substances and measure the concentrations of target substances, and techniques have been proposed for analyzing signals detected by the sensors.

[0003] Patent Literature 1 discloses a method for analyzing surface plasmon resonance (SPR) signals, which includes a step of detecting a change in the refractive index of a conductive layer after acquiring a first signal indicating the light intensity obtained from the conductive layer in which SPR occurs and before acquiring a second signal indicating the light intensity obtained from the conductive layer in which SPR occurs, and which can provide an improved signal-to-noise ratio.

[0004] Special Publication No. 2015-509597

[0005] However, the technique described in Patent Document 1 has a problem in that the detection signal detected using the optical sensor is not fully utilized.

[0006] An object of the present disclosure is to provide a computer program or the like that can improve the utilization of a detection signal detected using an optical sensor.

[0007] A computer program according to one aspect of the present disclosure causes a computer to perform a process of acquiring a first profile indicating a change in an interaction between substances detected using an optical sensor, performing a Fourier transform on the acquired first profile to acquire a spectrum for each frequency, performing an inverse Fourier transform on a spectrum of a specific frequency among the acquired spectrum for each frequency to generate a second profile, and acquiring a classification of a group to which the second profile belongs when the second profile is classified into a plurality of groups.

[0008] A computer program according to one aspect of the present disclosure causes a computer to perform the following processes: acquire a first profile indicating changes in interactions between substances detected using an optical sensor; perform spectral analysis on the acquired first profile to acquire a spectrum for each frequency; perform an inverse transformation of the spectral analysis on a spectrum of a specific frequency among the acquired spectra for each frequency to generate a second profile; and acquire the classification of the group to which the second profile belongs when the second profile is classified into multiple groups.

[0009] An information processing device according to one aspect of the present disclosure includes a control unit that executes a process of acquiring a first profile indicating a change in an interaction between substances detected using an optical sensor, performing a Fourier transform on the acquired first profile to acquire a spectrum for each frequency, performing an inverse Fourier transform on a spectrum of a specific frequency among the acquired spectrum for each frequency to generate a second profile, and acquiring a classification of a group to which the second profile belongs when the second profile is classified into a plurality of groups.

[0010] An information processing method according to one aspect of the present disclosure includes a computer executing a process of acquiring a first profile indicating a change in an interaction between substances detected using an optical sensor, performing a Fourier transform on the acquired first profile to acquire a spectrum for each frequency, performing an inverse Fourier transform on a spectrum of a specific frequency among the acquired spectrum for each frequency to generate a second profile, and acquiring a classification of a group to which the second profile belongs when the second profile is classified into a plurality of groups.

[0011] According to the present disclosure, it is possible to improve the utilization of a detection signal detected using an optical sensor.

[0012] FIG. 1 is a block diagram showing an example of the configuration of an information processing device of this embodiment; FIG. 2 is an explanatory diagram showing an example of the contents of information stored in an analysis DB; FIG. 3 is an explanatory diagram explaining an analysis process for an SPR profile; FIG. 4 is a schematic diagram showing an example of a screen displayed on a display unit; FIG. 5 is a flowchart showing an example of a processing procedure executed by an information processing device; FIG. 6 is a block diagram showing an example of the configuration of an information processing device of a second embodiment; FIG. 7 is an explanatory diagram showing an overview of a learning model; FIG. 8 is a flowchart showing an example of a processing procedure for generating a learning model; and FIG. 9 is a flowchart showing an example of a processing procedure for estimating minor classification.

[0013] The present disclosure will be specifically described with reference to the drawings showing embodiments thereof.

[0014] First Embodiment Fig. 1 is a block diagram showing an example of the configuration of an information processing device 1 according to this embodiment. The information processing device 1 is a device capable of various types of information processing and transmitting and receiving information, such as a server computer, a personal computer, or a quantum computer. The information processing device 1 executes analysis processing based on measurement data received from an optical sensor 2 that measures interactions between substances. In this embodiment, the optical sensor 2 is an SPR device that uses surface plasmon resonance (SPR), and an SPR profile indicating a change in an SPR signal is detected as an example.

[0015] The information processing device 1 includes a control unit 11, a storage unit 12, a communication unit 13, a display unit 14, an operation unit 15, and an input / output unit 16. The information processing device 1 may be a multi-computer consisting of multiple computers, or may be a virtual machine virtually constructed by software.

[0016] The control unit 11 includes one or more arithmetic processing devices such as a central processing unit (CPU), a micro-processing unit (MPU), a graphics processing unit (GPU), etc. The control unit 11 controls each component and executes processing using built-in memories such as a read-only memory (ROM) or a random access memory (RAM), a clock, a counter, etc. Each functional unit of the control unit 11 may be realized by software, by hardware (for example, an FPGA or an ASIC), or by a combination thereof.

[0017] The storage unit 12 includes a nonvolatile memory such as a hard disk, a flash memory, or an SSD (Solid State Drive). The storage unit 12 may be an external storage device connected to the information processing device 1. The storage unit 12 stores various computer programs and data referenced by the control unit 11. The storage unit 12 of this embodiment stores a program 1P for causing a computer to execute processing related to analysis of SPR profiles, and an analysis DB (Data Base) 121.

[0018] A computer program (computer program product) including program 1P may be provided by a non-transitory recording medium 1A on which the computer program is readably recorded. The storage unit 12 stores the computer program read from the recording medium 1A by a reading device (not shown). The recording medium 1A may be, for example, a magnetic disk, an optical disk, or a semiconductor memory. Alternatively, the computer program may be downloaded from an external server connected to a communications network and stored in the storage unit 12. Program 1P may be a single computer program or may be composed of multiple computer programs, and may be executed on a single computer or on multiple computers interconnected by a communications network.

[0019] The communication unit 13 includes a communication module that communicates with an external device via a network such as the Internet. The control unit 11 can transmit and receive various information to and from the external device via the communication unit 13. The communication unit 13 may be omitted.

[0020] The display unit 14 includes a display device such as a liquid crystal display, an organic EL (Electro Luminescence) display, etc. The display unit 14 displays information to be notified to the user in accordance with instructions from the control unit 11.

[0021] The operation unit 15 is an interface that accepts user operations. The operation unit 15 includes, for example, a keyboard, a mouse, a touch panel device with a built-in display, a speaker, a microphone, etc. The operation unit 15 accepts operation input from the user and sends a control signal corresponding to the operation content to the control unit 11.

[0022] The input / output unit 16 includes an input / output interface for connecting to an external device via a wired or wireless connection. The input / output unit 16 is connected to the optical sensor 2.

[0023] As described above, the optical sensor 2 measures interactions between substances using surface plasmon resonance (SPR). The optical sensor 2 acquires a profile showing changes in the interactions over time. In this embodiment, as an example, the optical sensor 2 detects interactions between target RNA and a compound, but the target substance may be any of DNA, RNA, protein, etc.

[0024] In SPR measurements, a compound (analyte) to be detected for binding is applied to target RNA, which serves as a target substance immobilized on the metal film surface of a sensor chip, and the binding of the analyte to the target RNA is observed by changes in the surface plasmon resonance (SPR) signal. Specifically, light is irradiated onto the prism portion at the bottom of the sensor chip while varying the angle of incidence within a range equal to or greater than the critical angle, and the intensity of the reflected light is measured, thereby observing the point at which the intensity of the reflected light is minimized. The angle at which the intensity of the reflected light is minimized varies depending on the mass on the sensor chip. Therefore, the angle at which the intensity of the reflected light is minimized changes with changes such as binding or dissociation between the ligand immobilized on the metal film and the added target RNA. By detecting the angle at which the intensity of the reflected light is minimized over time using the optical sensor 2, interactions between substances can be monitored in real time.

[0025] The optical sensor 2 of this embodiment acquires an SPR profile that indicates changes over time in response units (RU) converted from changes in the angle of the SPR signal. The SPR profile represents changes over time in the interaction between the target RNA and the compound. The optical sensor 2 outputs the SPR profile obtained by detection to the information processing device 1.

[0026] The optical sensor 2 is not limited to an SPR device, and may be one that detects a profile indicating an interaction between substances using other methods. For example, the optical sensor 2 may be a detection device that uses a bio-layer interferometry (BLI) method.

[0027] The information processing device 1 is not limited to a configuration including the optical sensor 2 and acquiring the SPR profile directly from the optical sensor 2. The information processing device 1 may acquire the SPR profile stored in a predetermined storage area via the optical sensor 2, or may receive the SPR profile transmitted from the optical sensor 2 or an external device via communication.

[0028] 2 is an explanatory diagram showing an example of the contents of information stored in the analysis DB 121. The analysis DB 121 is a database that stores information related to the analysis of SPR profiles. The analysis DB 121 stores records in which, for example, compound identification information that identifies a compound is used as a key to link information such as original SPR profile information, reconstructed SPR profile information, major category names, and minor category names.

[0029] The compound identification information includes, for example, a compound ID, a compound name, etc. The original SPR profile information represents the SPR profile of the compound detected by the optical sensor 2. The reconstructed SPR profile information represents an SPR profile reconstructed based on the SPR profile detected by the optical sensor 2. The major category name represents the name of the major group to which the compound belongs. The major category name includes a binding group and a non-binding group. The minor category name represents the name of the minor group to which the compound belongs. The minor category name includes multiple patterns from pattern 1 to pattern N. Details of the above-mentioned two types of SPR profiles and the classification contents will be described later.

[0030] Generally, in interaction analysis, the binding ability between a target RNA and a compound is evaluated by the dissociation constant obtained from the SPR profile. The dissociation constant is calculated by fitting the SPR profiles obtained in the binding phase and dissociation phase to a theoretical formula to calculate the binding rate constant and the dissociation rate constant, and is then determined from the ratio of the obtained binding rate constant to the dissociation rate constant. The binding phase is a phase in which a solution containing an analyte is applied to the target RNA to bind the analyte to the target RNA. The dissociation phase is a phase in which a solution not containing an analyte is applied to the target RNA to dissociate the analyte bound to the target RNA. Alternatively, in the case of qualitative evaluation, binding evaluation is performed by visually judging the magnitude of the RU value (response intensity), the shape of the SPR profile (e.g., the magnitude of the slope), etc.

[0031] The above-described analysis has the problem that it is difficult to calculate the dissociation rate constant, especially when the RU value in the SPR profile is small, making binding evaluation difficult. Furthermore, visual evaluation requires specialized knowledge. Furthermore, while it is believed that the SPR profile contains information other than the dissociation constant, the information contained in the SPR profile has not been fully utilized.

[0032] In this embodiment, a new SPR profile is reconstructed by extracting only the spectrum of a specific frequency based on the SPR profile detected by the optical sensor 2. This makes it possible to obtain more information than can be obtained from the original SPR profile alone.

[0033] Fig. 3 is an explanatory diagram illustrating the analysis process for an SPR profile. The analysis process executed by the information processing device 1 will be described in detail with reference to Fig. 3. The information processing device 1 acquires an SPR profile (hereinafter also referred to as an original SPR profile) detected by the optical sensor 2. As shown in the upper part of Fig. 3, the original SPR profile is time-series data indicating the interaction between the target RNA and the compound.

[0034] The information processing device 1 acquires a plurality of raw SPR profiles obtained by measuring a plurality of compounds. The raw SPR profile may be generated for each compound, or may be acquired as data summarizing raw SPR profiles for a plurality of compounds. The raw SPR profile is associated with identification information (e.g., compound name, compound ID, etc.) of the compound to be measured. The information processing device 1 may acquire the raw SPR profiles through a plurality of optical sensors 2.

[0035] As shown in the center of Figure 3, the information processing device 1 performs a Fourier transform on the original SPR profile to obtain a spectrum for each frequency. The information processing device 1 extracts a spectrum of a specific frequency from the obtained spectrum for each frequency and performs an inverse Fourier transform on the extracted spectrum of the specific frequency to generate a new SPR profile (hereinafter also referred to as a reconstructed SPR profile). The spectrum extracted to generate the reconstructed SPR profile may be a spectrum of a frequency equal to or lower than a predetermined frequency threshold. As shown in Figure 3, it is expected that the spectrum obtained by Fourier transforming the original SPR profile can be roughly divided into a group of high-frequency spectra and a group of low-frequency spectra sufficiently separated from the group of high-frequency spectra. Therefore, the spectrum included in the group of low-frequency spectra may be the target of extraction.

[0036] The information processing device 1 performs the above-described processing on each of the raw SPR profiles for a plurality of compounds to generate a reconstructed SPR profile corresponding to the raw SPR profile of each compound.

[0037] As shown in the lower part of FIG. 3 , the information processing device 1 classifies the generated reconstituted SPR profiles into multiple small groups. Specifically, based on the multiple reconstituted SPR profiles, two or more small groups are created, each containing one or more reconstituted SPR profiles. The information processing device 1 classifies the reconstituted SPR profiles using, for example, k-means, a clustering method. k-means automatically classifies each reconstituted SPR profile into a predetermined number of clusters. k-means calculates the distance (e.g., Euclidean distance, Mahalanobis distance, etc.) between the reconstituted SPR profile and the center of each cluster, and classifies the reconstituted SPR profile into the cluster with the smallest calculated distance. Based on the clustering results, the information processing device 1 identifies the small group to which the reconstituted SPR profile of each compound belongs.

[0038] In the above description, the reconstituted SPR profiles are classified using k-means, but the classification method for the reconstituted SPR profiles is not particularly limited. The information processing device 1 may identify the group to which the reconstituted SPR profile belongs using a method such as hierarchical clustering, k-nearest neighbors, an autoregressive model, a neural network, or an SVM (Support Vector Machine).

[0039] The information processing device 1 further obtains a classification of a large group that combines multiple small groups. The large group includes a binding group to which the reconstructed SPR profiles of compounds that bind to the target RNA belong, and a non-binding group to which the reconstructed SPR profiles of compounds that do not bind to the target RNA belong. Each small group is classified into either a binding group or a non-binding group. In other words, each reconstructed SPR profile is classified into one of the small groups corresponding to multiple patterns, and further classified into either a binding group or a non-binding group. Note that the number of small groups included in the binding group or the non-binding group may be one.

[0040] The information processing device 1 may identify the classification of the large group to which each small group belongs based on the distance between the small groups. For example, the information processing device 1 accepts, from among a plurality of small groups, designation of one small group belonging to the combined group (hereinafter also referred to as a first specific group) and one small group belonging to the non-combined group (hereinafter also referred to as a second specific group). The information processing device 1 compares the distance between the small group to be classified and each of the first specific group and the second specific group, and identifies the large group corresponding to the specific group that is closest to the small group to be classified as the large group to which the small group to be classified belongs.

[0041] The classification of the large groups may be performed manually. When the large groups are classified manually, the information processing device 1 may generate a diagram visually showing the distribution of each reconstituted SPR profile and the classification results of the small groups, and display the generated diagram on the display unit 14. The diagram may be obtained by reducing the dimensions of the reconstituted SPR profiles to two or three dimensions using a dimension reduction method such as principal component analysis (PCA), uniform manifold approximation and projection (UMAP), or t-distributed stochastic neighbor embedding (t-SNE). The user checks the diagram displayed on the display unit 14 and assigns a large group classification to each reconstituted SPR profile. The information processing device 1 obtains the large group classification of each reconstituted SPR profile based on the user's operation on the operation unit 15.

[0042] The information processing device 1 stores in the analysis DB 121 compound identification information, original SPR profile, reconstructed SPR profile, large group classification, small group classification, etc. in association with each other for each of a plurality of compounds.

[0043] The information processing device 1 associates the reconstructed SPR profile with the classification result of the reconstructed SPR profile and displays them on the display unit 14.

[0044] Fig. 4 is a schematic diagram showing an example of a screen 30 displayed on the display unit 14. In the example shown in Fig. 4, the screen 30 includes a first display section 31 that displays the original SPR profile, a second display section 32 that displays the reconstructed SPR profile, and a third display section 33 that displays the classification results.

[0045] The information processing device 1 displays the original SPR profile of each compound on the first display unit 31 based on the information stored in the analysis DB 121. The information processing device 1 also displays the reconstructed SPR profile generated based on the original SPR profile displayed on the first display unit 31 on the second display unit 32.

[0046] 4 , the first display unit 31 and the second display unit 32 display SPR profiles for all compounds that have been subjected to clustering in a superimposed manner. The information processing device 1 may display profiles for the same compound in a common display mode, such as a common color or line type, on the first display unit 31 and the second display unit 32, thereby visually displaying the correspondence between the original SPR profile and the reconstituted SPR profile for each compound. This configuration is advantageous because it allows the correspondence between the original SPR profile and the reconstituted SPR profile to be clearly recognized.

[0047] The information processing device 1 displays the classification results obtained by clustering on the third display section 33. The third display section 33 may display the compound IDs and major category names corresponding to the SPR profiles, along with the minor category names.

[0048] The third display section 33 may be displayed in a pop-up format. The user can use the operation section 15 to specify the SPR profile of any compound displayed on the first display section 31 or the second display section 32, thereby displaying the third display section 33 on the screen 30, which shows the classification result for the specified compound.

[0049] 4, the screen 30 may further include a receiving unit 34 that receives selection of display targets for the first display unit 31 and the second display unit 32. The receiving unit 34 has selection fields such as subcategory names to be displayed and check boxes corresponding to each subcategory name. The user can switch between displaying and hiding each small group by appropriately changing the selection of the check boxes using the operation unit 15, and can selectively display only the SPR profiles of any small group on the first display unit 31 and the second display unit 32.

[0050] According to the above-described analysis method, in addition to the broad classification of whether or not a compound binds based on the reconstituted SPR profile, a subclassification indicating a more detailed group can be obtained. This allows for a more detailed classification that reflects the information in the profile itself, such as, for example, compounds that bind to the target RNA, particularly compounds that bind quickly and dissociate slowly, or compounds that bind slowly but dissociate quickly. Obtaining detailed classification information leads to the screening of compounds that are more suited to the search objective.

[0051] 5 is a flowchart showing an example of a processing procedure executed by the information processing device 1. The processing in each of the following flowcharts is executed by the control unit 11 in accordance with a program 1P stored in the storage unit 12 of the information processing device 1.

[0052] The control unit 11 of the information processing device 1 acquires raw SPR profiles for a plurality of compounds (step S11). The raw SPR profiles are associated with compound identification information. The control unit 11 performs a Fourier transform on the acquired raw SPR profiles (step S12) to acquire spectra for each frequency in the raw SPR profiles (step S13).

[0053] The control unit 11 extracts a spectrum of a specific frequency from the acquired spectrum for each frequency (step S14). The control unit 11 performs an inverse Fourier transform on the extracted spectrum of the specific frequency (step S15) and generates a reconstructed SPR profile based on the spectrum of the specific frequency (step S16). The control unit 11 performs the processes of steps S12 to S16 for each of the multiple original SPR profiles acquired in step S11, and generates a reconstructed SPR profile corresponding to each original SPR profile.

[0054] The control unit 11 classifies the generated reconstituted SPR profiles into a plurality of small groups using, for example, a clustering method, and thereby obtains the small group classification to which each reconstituted SPR profile belongs (step S17).

[0055] The control unit 11 obtains a classification of a large group that combines multiple small groups (step S18). The control unit 11 associates the compound identification information, the original SPR profile, the reconstructed SPR profile, the large group classification, and the small group classification, and stores them in the analysis DB 121 (step S19).

[0056] The control unit 11 generates a screen showing the reconstituted SPR profile and the classification of the reconstituted SPR profile (step S20). The control unit 11 displays the generated screen on the display unit 14 (step S21), and ends the series of processes.

[0057] In the above-described processing, the output destination of the analysis results such as the reconstituted SPR profile and the classification of the reconstituted SPR profile is not limited to the display unit 14, and the control unit 11 may output the analysis results to an external device other than the information processing device 1. The control unit 11 may store a plurality of raw SPR profiles collected during a specific period in the analysis DB 121, and perform the above-described processing on the stored plurality of raw SPR profiles at any timing.

[0058] According to this embodiment, more useful information can be obtained from the SPR profile, thereby increasing the utility of the SPR profile. Compounds can be classified more accurately by classifying them based on a reconstructed SPR profile generated from specific frequency components in the original SPR profile. By classifying compounds into small groups by clustering the reconstructed SPR profile and then classifying them into large groups that ultimately indicate whether or not they bind, the presence or absence of binding ability can be more appropriately determined. For example, even if a compound is determined to have no binding ability when visually judged based on the original SPR profile, it can be determined to have binding ability by taking into account the clustering of the reconstructed SPR profile, leading to the screening of useful compounds.

[0059] By generating a reconstructed SPR profile, the characteristics of the SPR sensorgram itself can be reflected in the classification. By classifying compounds into small groups based on the characteristics of the SPR profile in addition to the presence or absence of binding, compounds can be evaluated in more detail.

[0060] In this embodiment, an example has been described in which an SPR profile and a reconstructed SPR profile are generated using Fourier transform and inverse Fourier transform as frequency analysis. The frequency analysis method is not limited to this, and any method can be used as long as it can remove noise from an SPR sensorgram and generate a reconstructed SPR profile suitable for clustering. For example, appropriate frequency transform and inverse transform (noise removal) methods such as wavelet transform and inverse wavelet transform can be used.

[0061] Second Embodiment In the second embodiment, a learning model is constructed to estimate the classification of compounds according to their SPR profiles. In the following embodiment, differences from the first embodiment will be mainly described, and the same reference numerals will be used to designate components common to the first embodiment, and detailed descriptions thereof will be omitted.

[0062] 6 is a block diagram showing an example of the configuration of an information processing device 1 according to the second embodiment. The information processing device 1 according to the second embodiment stores a learning model 122 in the storage unit 12. The learning model 122 is a machine learning model that has learned predetermined training data. The learning model 122 is expected to be used as a program module that constitutes part of artificial intelligence software.

[0063] FIG. 7 is an explanatory diagram showing an overview of the learning model 122. The learning model 122 receives the raw SPR profile of a compound as input and outputs information indicating the classification of small groups for the raw SPR profile of the compound. The learning model 122 is, for example, a convolutional neural network (CNN), which is a type of neural network. The learning model 122 includes an input layer to which the raw SPR profile is input, an output layer that outputs the classification of small groups (small group names), and an intermediate layer (hidden layer). The intermediate layer may include a convolutional layer, a pooling layer, a fully connected layer, etc. The intermediate layer has multiple nodes that extract features of the raw SPR profile and passes the features extracted using various parameters to the output layer. When the raw SPR profile is input to the input layer, calculations are performed in the intermediate layer using the learned parameters, and output information indicating the classification results of the small groups is output from the output layer.

[0064] The input data input to the learning model 122 may include a reconstructed SPR profile instead of or in addition to the original SPR profile.

[0065] The output layer of the learning model 122 has a plurality of nodes each corresponding to a predetermined small group name. Each node corresponding to a small group name outputs the accuracy of the small group name. The output information with the highest accuracy output from the output layer can be used as the output of the learning model 122. Note that the configuration of the output layer is not particularly limited as long as information indicating the classification of the small group to be predicted can be obtained. The learning model 122 may be configured to output the classification of the large group in addition to the classification of the small group.

[0066] The learning model 122 can be generated by preparing training data in which labels indicating subcategory names are associated with raw SPR profiles, and using the training data to perform machine learning on an untrained neural network. The information processing device 1 uses information stored in the analysis DB 121 as training data. As described in the first embodiment, the analysis DB 121 stores raw SPR profiles, reconstructed SPR profiles, and subcategory names for a plurality of compounds.

[0067] The information processing device 1 inputs a plurality of original SPR profiles contained in the training data into the input layer of the neural network model before learning, and obtains subcategory names output from the output layer after arithmetic processing in the intermediate layer. The information processing device 1 compares the subcategory names output from the output layer with the subcategory names contained in the training data, and optimizes parameters such as the weights between neurons using, for example, backpropagation so that the subcategory names output from the output layer approach the correct values. The learning model 122 may be constructed by an external device and deployed to the information processing device 1.

[0068] The configuration of the learning model 122 is not limited to the above example, and may be any model that can identify group classification for the original SPR profile and / or the reconstructed SPR profile. The learning model 122 may be a model constructed using other learning algorithms, such as a recurrent neural network (RNN), a graph neural network (GNN), a transformer, a support vector machine (SVM), logistic regression, eXtreme Gradient Boosting (XGBoost), or a random forest.

[0069] FIG. 8 is a flowchart showing an example of a processing procedure for generating the learning model 122.

[0070] The control unit 11 of the information processing device 1 acquires training data in which the raw SPR profile is associated with the classification of the small group based on the information stored in the analysis DB 121 (step S31).

[0071] The control unit 11 uses the acquired training data to generate a learning model 122 that, when an original SPR profile is input, outputs a classification of small groups for the original SPR profile (step S32). Specifically, the control unit 11 inputs the original SPR profile included in the training data to the learning model 122 and optimizes various parameters so that the output from the learning model 122 approximates a correct value. For example, when learning is completed because the number of learning times meets a predetermined criterion, the control unit 11 stores definition information related to the trained learning model 122 in the storage unit 12 as the trained learning model 122.

[0072] The information processing device 1 acquires classification of new small groups of compounds using the obtained learning model 122. Fig. 9 is a flowchart showing an example of a small classification estimation process procedure.

[0073] The control unit 11 of the information processing device 1 acquires an original SPR profile for a new compound to be estimated (step S41). The control unit 11 inputs the acquired original SPR profile to the learning model 122 (step S42). The control unit 11 may execute the same processes as steps S12 to S16 of the first embodiment to generate a reconstructed SPR profile corresponding to the original SPR profile and input the generated reconstructed SPR profile to the learning model 122.

[0074] The control unit 11 acquires the classification of the small groups output from the learning model 122 (step S43). The control unit 11 outputs information indicating the acquired classification of the small groups via the display unit 14 (step S44), and ends the series of processes.

[0075] The control unit 11 may obtain classification of small groups for raw SPR profiles of multiple compounds using the learning model 122, and extract only compounds classified into specific small groups. The control unit 11 may output, for example, a list showing the extracted compounds via the display unit 14.

[0076] According to this embodiment, it is possible to construct a learning model 122 capable of estimating small group classifications from SPR profiles by utilizing big data accumulated in the analysis DB 121. By using the obtained learning model 122, it is possible to more easily obtain small group classifications, thereby increasing the utility of SPR profiles and improving convenience.

[0077] Based on the estimation results of the learning model 122, only compounds that exhibit a desired SPR profile pattern can be efficiently extracted, thereby improving the accuracy of screening.

[0078] Table 1 below shows the classification accuracy of a learning model trained using training data labeled with the classification results obtained by the method of this embodiment. In evaluating the learning model, training data in which a predetermined number of compounds were labeled as bound or not bound was used to construct a learning model using two types of learning algorithms, Random Forest and XGBoost. The training data used included a dataset labeled with the results of classification into bound or not based on the reconstructed SPR profiles generated by this method, and a dataset labeled with the results of classification into bound or not without using this method. The number of compounds recognized as hits (compounds recognized as bound) without using this method was 104, while the number of compounds recognized as hits with this method was 117. By using this method, the number of compounds recognized as hits was increased.

[0079] Using each of the obtained learning models, the binding status of 400 new compounds was classified. The accuracy of the models was evaluated based on the classification results. The evaluation results are shown in Table 1 below. In Table 1, the average value represents the average value of 100 classification results, and the maximum value represents the maximum value of 100 classification results. The indicators shown are F-value, Precision (accuracy), and Recall (reproducibility). As shown in Table 1, in both Random Forest and XGBoost, it was found that the indicator values ​​were improved when this method was used compared to when this method was not used. For example, in the case of the Random Forest model, the average Precision value increased from 0.13 to 0.17, and the maximum value increased from 0.17 to 0.20. In the case of the XGBoost model, the average Precision value increased from 0.18 to 0.22, and the maximum value increased from 0.14 to 0.19. From Table 1, it was confirmed that the hit rate can be improved by learning using the training data obtained by this method.

[0080]

[0081] The embodiments disclosed herein are illustrative in all respects and should not be considered limiting. The technical features described in each embodiment can be combined with each other, and the scope of the present invention is intended to include all modifications within the scope of the claims and equivalents thereto. The sequences shown in each embodiment are not limited, and within the scope of no contradiction, each processing step may be executed in a different order, or multiple processes may be executed in parallel. The entity that performs each process is not limited, and within the scope of no contradiction, the process of each device may be executed by another device.

[0082] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. A multiple claim (multi-multi claim) that references at least one other multiple claim may also be used.

[0083] REFERENCE SIGNS LIST 1 Information processing device 11 Control unit 12 Storage unit 13 Communication unit 14 Display unit 15 Operation unit 16 Input / output unit 121 Analysis DB 122 Learning model 1P Program 1A Recording medium 2 Optical sensor

Claims

1. A first profile showing the change in the interaction between substances detected using an optical sensor was obtained. The acquired first profile is Fourier transformed to obtain the spectrum for each frequency, A second profile is generated by performing an inverse Fourier transform on the spectrum of a specific frequency from the acquired spectrum for each frequency. When the second profile is classified into multiple groups, obtain the classification of the group to which the second profile belongs. A computer program that instructs a computer to perform a process.

2. Multiple second profiles are generated based on multiple first profiles, The group to which each second profile belongs is identified by classifying the generated second profiles using a clustering method. The computer program according to claim 1.

3. The second profile is generated based on a spectrum below a predetermined frequency. The computer program according to claim 1 or claim 2.

4. The first profile shows the time evolution of the interaction between the compound and the target substance. The computer program according to claim 1 or claim 2.

5. The first profile shows the time evolution of the interaction between the compound and the target substance. The group includes a group having the ability to bind to the target substance and a group not having the ability to bind to the target substance. The computer program according to claim 1.

6. The group having the ability to bind to the target substance or the group not having the ability to bind to the target substance is subdivided into multiple groups. The computer program according to claim 5.

7. The first profile is detected using the optical sensor that utilizes surface plasmon resonance or biolayer interferometry. The computer program according to claim 1 or claim 2.

8. Multiple first profiles and the group to which each second profile generated based on each first profile belongs are obtained in association. Based on the acquired multiple first profiles and the groups corresponding to each first profile, a learning model is generated that outputs the group corresponding to the first profile when a first profile is input. The computer program according to claim 1 or claim 2.

9. A first profile showing the change in the interaction between substances detected using an optical sensor was obtained. The acquired first profile is subjected to spectral analysis to obtain the spectrum for each frequency. A second profile is generated by performing the inverse transform of the spectral analysis on the spectrum of a specific frequency among the acquired spectra for each frequency. When the second profile is classified into multiple groups, obtain the classification of the group to which the second profile belongs. A computer program that instructs a computer to perform a process.

10. A first profile showing the change in the interaction between substances detected using an optical sensor was obtained. The acquired first profile is Fourier transformed to obtain the spectrum for each frequency, A second profile is generated by performing an inverse Fourier transform on the spectrum of a specific frequency from the acquired spectrum for each frequency. When the second profile is classified into multiple groups, obtain the classification of the group to which the second profile belongs. It includes a control unit that performs processing. Information processing device.

11. A first profile showing the change in the interaction between substances detected using an optical sensor was obtained. The acquired first profile is Fourier transformed to obtain the spectrum for each frequency, A second profile is generated by performing an inverse Fourier transform on the spectrum of a specific frequency from the acquired spectrum for each frequency. When the second profile is classified into multiple groups, obtain the classification of the group to which the second profile belongs. An information processing method in which a computer performs the processing.