Antibody selection method, computer program, and information processing device

Abstract

Provided are: an antibody selection method, a computer program, and an information processing device.　An antibody that exhibits specificity for a prescribed antigen is selected by: acquiring sequence data for a plurality of antibodies that includes an antibody for which the antigen specificity is unknown; clustering the acquired sequence data for the plurality of antibodies from the standpoint of similarity of a paratope pseudosequence; discriminating a cluster that exhibits specificity for the prescribed antigen by using a subset of the antibodies in each cluster; and estimating that, within a discriminated cluster, the antibody outside the subset has the same antigen specificity as the antibody in the subset.

Description

Antibody selection method, computer program, and information processing device

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

[0002] The adaptive immune system includes T cell receptors and B cell receptors as molecules that specifically recognize a wide variety of pathogens. Since living organisms cannot predict the pathogens that will invade from the outside world, they respond to a wide variety of pathogens by maintaining the diversity of T cell receptors and B cell receptors. T cell receptors and B cell receptors are composed of a combination of multiple gene regions. For example, human T cell receptors have 10 15 ~10 20 types, 10 for B cell receptors 16 ~10 18 It is believed that there are several types of receptors.

[0003] As a technique for determining a profile of recombinant DNA sequences in T cells and B cells, Patent Document 1 discloses a method for collecting a sample from a subject containing T cells and B cells, spatially isolating individual molecules of genomic DNA from the cells, determining the sequences of the spatially isolated individual molecules, and creating a profile of the recombinant DNA sequence.

[0004] Special Publication No. 2012-508011

[0005] The technique of Patent Document 1, in which recombinant DNA sequence profiles are created for individual T cells and B cells, has limitations in terms of selecting antibodies having specificity for a particular antigen from among a wide variety of antibodies.

[0006] In one aspect, an object of the present invention is to provide an antibody selection method, a computer program, and an information processing device that can select an antibody having specificity for a specific antigen.

[0007] The antibody selection method of the present disclosure acquires sequence data of multiple antibodies, including antibodies with unknown antigen specificity, clusters the acquired sequence data of multiple antibodies based on the similarity of their paratope pseudosequences, uses a subset of antibodies within each cluster to identify clusters that have specificity for a specific antigen, and selects antibodies that have specificity for the specific antigen within the identified cluster by inferring that antibodies outside the subset have the same antigen specificity as the antibodies within the subset.

[0008] In one aspect, antibodies with specificity for a particular antigen can be selected.

[0009] FIG. 1 is an explanatory diagram illustrating an outline of an antibody selection method according to embodiment 1. FIG. 2 is a block diagram illustrating the internal configuration of an information processing device. FIG. 3 is an explanatory diagram illustrating a clustering method according to this embodiment. FIG. 4 is an explanatory diagram illustrating the effectiveness of clustering. FIG. 5 is an explanatory diagram illustrating a clustering result of antibodies including known antibodies and novel antibodies. FIG. 6 is an explanatory diagram illustrating an evaluation result using a cluster representative. FIG. 7 is a flowchart illustrating the procedure of processing executed by an information processing device according to embodiment 1. FIG. 8 is an explanatory diagram illustrating an outline of an antibody selection method according to embodiment 2. FIG. 9 is an explanatory diagram illustrating a BCR data collection process. FIG. 10 is a diagram illustrating a clustering result using an antigen probe. FIG. 11 is an explanatory diagram illustrating an evaluation method. FIG. 12 is an explanatory diagram illustrating an evaluation result of antigen specificity. FIG. 13 is a flowchart illustrating the procedure of processing executed by an information processing device according to embodiment 2.

[0010] The present invention will be described in detail below with reference to the drawings illustrating embodiments. (Embodiment 1) FIG. 1 is an explanatory diagram outlining an antibody selection method according to embodiment 1. This disclosure focuses on a method for selecting novel antibodies specific to a particular antigen. Here, specificity (also referred to as antigen specificity) refers to the property of an antibody that reacts only with a specific antigen. An information processing device 10 according to embodiment 1 obtains amino acid sequence data for antibodies with known antigen specificity and antibodies (new antibodies) with unknown antigen specificity, and then clusters the obtained sequence data to select novel antibodies with antigen specificity for a specific antigen.

[0011] 1, sequence data of antibody α having specificity to known antigen A is shown schematically by a black solid line, sequence data of antibody β having specificity to known antigen B is shown schematically by a gray solid line, and sequence data of the novel antibody is shown schematically by a dashed line. Antibody α is shown schematically by a black circle, antibody β is shown schematically by a gray circle, and the novel antibody is shown schematically by a circle surrounded by a dashed line. The information processing device 10, for example, acquires the sequence data of these antibodies and uses the sequence data of each antibody to cluster the sequence data of each antibody, including antibody α, antibody β, and the novel antibody.

[0012] As a result of clustering, a cluster formed mainly from the sequence data of antibody α may contain sequence data of one or more new antibodies. Furthermore, a cluster formed mainly from the sequence data of antibody β may contain sequence data of one or more other new antibodies. In this case, the information processing device 10 infers that the new antibody included in the cluster of antibody α has specificity for known antigen A, and that the new antibody included in the cluster of antibody β has specificity for known antigen B. In this way, the information processing device 10 can select new antibodies that have specificity for known antigen A and known antigen B.

[0013] 2 is a block diagram illustrating the internal configuration of the information processing device 10. The information processing device 10 is a dedicated or general-purpose computer, and includes a control unit 11, a storage unit 12, a communication unit 13, an operation unit 14, and a display unit 15.

[0014] The control unit 11 includes, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), and a RAM (Random Access Memory). The ROM included in the control unit 11 stores a control program that controls the operation of each hardware unit included in the server device 10. The CPU in the control unit 11 reads and executes the control program stored in the ROM and a computer program (described below) stored in the storage unit 12, and controls the operation of each hardware unit, thereby causing the entire device to function as the information processing device 10 of the present disclosure. The RAM included in the control unit 11 temporarily stores data used during calculation and control.

[0015] In the embodiment, the control unit 11 is configured to include a CPU, a ROM, and a RAM, but the configuration of the control unit 11 is not limited to the above. The control unit 11 may be, for example, one or more control circuits or arithmetic circuits including a GPU (Graphics Processing Unit), an FPGA (Field Programmable Gate Array), a DSP (Digital Signal Processor), a quantum processor, volatile or non-volatile memory, etc. The control unit 11 may also have functions such as a clock that outputs date and time information, a timer that measures the elapsed time from when a measurement start instruction is given until when a measurement end instruction is given, and a counter that counts numbers.

[0016] The storage unit 12 includes a storage device such as a hard disk drive (HDD), a solid state drive (SSD), etc. The storage unit 12 stores various computer programs executed by the control unit 11 and various data used by the control unit 11.

[0017] The computer program stored in the memory unit 12 includes an antibody selection program PG that causes a computer to execute a process of acquiring sequence data of multiple antibodies including antibodies with unknown antigen specificity, clustering the acquired sequence data in terms of similarity of paratope pseudo sequences, using a subset of antibodies within each cluster to identify clusters that have specificity for a specific antigen, and inferring that antibodies outside the subset within the identified cluster have the same antigen specificity as antibodies within the subset, thereby selecting antibodies that have specificity for a specific antigen.

[0018] Each of the computer programs may be a single computer program, or may be a group of programs consisting of multiple computer programs. Furthermore, the computer programs may partially use an existing library. Each of the computer programs may be executed on a single computer, or may be executed cooperatively by multiple computers.

[0019] The computer program may be provided by a non-transitory recording medium RM (program product) on which the computer program is readably recorded. The recording medium RM is a portable memory such as a CD-ROM, a USB memory, or an SD card (SD: Secure Digital). The control unit 11 reads various computer programs from the recording medium RM using a reading device (not shown) and stores the read various computer programs in the memory unit 12. The computer program may also be provided by communication. The control unit 11 acquires the computer program by communication via the communication unit 13 and stores the acquired computer program in the memory unit 12.

[0020] The communication unit 13 includes a communication interface for transmitting and receiving various data to and from an external device. The communication interface of the communication unit 13 may be a communication interface conforming to a communication standard such as Wi-Fi (registered trademark), LAN (Local Area Network), Bluetooth (registered trademark), ZigBee (registered trademark), 3G, 4G, 5G, or LTE (Long Term Evolution). When data to be transmitted is input from the control unit 11, the communication unit 13 transmits the data to the destination external device, and when data transmitted from the external device is received, the communication unit 13 outputs the received data to the control unit 11. In this embodiment, an example of an external device is a terminal device used by a medical professional such as a doctor or a subject.

[0021] The operation unit 14 includes operation devices such as a touch panel, a keyboard, and switches, and receives various operations and settings from a user, etc. The control unit 11 performs appropriate control based on various pieces of operation information provided by the operation unit 14, and stores setting information in the storage unit 12 as necessary.

[0022] The display unit 15 includes a display device such as a liquid crystal display or an organic EL (Electro-Luminescence) display, and displays information to be presented in response to an instruction from the control unit 11 .

[0023] In this embodiment, the information processing device 10 may be a single computer, or may be a computer system configured with multiple computers and peripheral devices, etc. Furthermore, the information processing device 10 may be a virtual machine whose entity is virtualized, or may be a cloud.

[0024] The processing executed by the information processing device 10 is described in detail below. As described above, the information processing device 10 clusters antibodies based on sequence data of known antibodies and novel antibodies. FIG. 3 is an explanatory diagram illustrating a clustering method according to this embodiment. Various antibodies are composed of amino acid sequences and are described by the V gene name, the J gene name, and the location of the complementarity-determining region (CDR) within the amino acid sequence. The complementarity-determining region (CDR) is also referred to as a CDR. The CDR includes three regions (CDR1, CDR2, and CDR3). In this embodiment, the three segments CDR1, CDR2, and CDR3 are pseudo-linked to construct a single pseudo-sequence, and antibody sequence data is clustered based on the similarity of this pseudo-sequence (hereinafter referred to as a paratope pseudo-sequence). This clustering method is also referred to as a CDR clustering method. Note that AIR in the diagram stands for adaptive immune receptor (AIR).

[0025] Clustering by paratope pseudosequences was performed using the MMseqs2 search method with appropriate sequence identity and coverage thresholds. Alignments of paratope pseudosequences with fixed sequence identity and coverage are statistically more significant than CDR3 alignments based on the same criteria due to the longer sequence lengths.

[0026] Figure 4 illustrates the effectiveness of clustering. To verify the effectiveness of clustering using paratope pseudosequences, we utilized the publicly available BCR database (CovAbDab), which contains a large collection of BCRs (B cell receptors) known to have specificity for coronavirus antigens. Among the datasets registered in the database, we excluded nonhuman antibodies and focused on the sequence data of 6,412 antibodies that recognize the receptor-binding domain (RBD) of the SARS-CoV-2 spike protein (hereafter referred to as RBD) and 3,702 antibodies that target other regions of the spike protein (hereafter referred to as non-RBD).

[0027] Paratope pseudosequences were created from the BCR data of each antibody obtained from the BCR database, and then clustering was performed based on the similarity of the paratope pseudosequences, with the sequence identity threshold and coverage threshold set to appropriate values. Of the 1,074 clustered non-singleton clusters, 25 clusters were extracted in descending order of cluster size, and the proportions of RBD and non-RBD in each cluster are shown in the graph in Figure 4. The horizontal axis of the graph in Figure 4 represents the cluster name, and the vertical axis represents the cluster size. The graph in Figure 4 reveals that of the 25 clusters, clusters Cov1 to Cov5, Cov9 to Cov11, Cov15, Cov18, Cov21, and Cov24 are primarily clusters based on RBD, while the other clusters are primarily clusters based on non-RBD.

[0028] Clustering was performed with various sequence identity and coverage thresholds, and the average cluster purity was found to be sensitive to both the sequence identity and coverage thresholds. Here, cluster purity is an index representing the degree to which the sequence data in a cluster matches a specific annotation (in this case, either RBD or non-RBD). For example, the cluster purity of a cluster with the highest RBD frequency (e.g., Cov1) is calculated by RBD / (RBD + non-RBD) × 100 (%), and the cluster purity of a cluster with the highest non-RBD frequency (e.g., Cov6) is calculated by non-RBD / (RBD + non-RBD) × 100 (%).

[0029] The sequence identity threshold was adjusted to achieve a cluster purity of 95% or more for each cluster. The coverage threshold was set to 90% to reduce the possibility of sequences spanning CDR boundaries. For example, when the sequence identity threshold was set to 80% and the coverage threshold was set to 90%, the cluster purity of the top 25 clusters shown in the graph in Figure 4 was 95.3%. Note that the sequence identity threshold and coverage threshold are not limited to the above values, and may be set appropriately so that the cluster purity of each cluster is higher than the set values.

[0030] Alignment of the CDRH sequences (CDRHs are the complementarity-determining regions of the heavy chain) of the top three largest clusters revealed that CDRH3 showed the most variability, but CDRH1 and CDRH2 also showed variability. The average sequence identity of a pair of CDRH3s within a cluster was approximately 10% lower than the lower limit (80%) of a typical clonotype, suggesting that CDR clusters can group antibodies from a clonal population that share paratope antigen specificity.

[0031] Next, an example of application to novel antibodies will be described. Figure 5 is an explanatory diagram illustrating the clustering results of antibodies, including known and novel antibodies. For the sequence data of known antibodies, similar to the above, non-human antibodies were excluded from the dataset registered in the BCR database (CovAbDab), and sequence data of 6,412 antibodies that recognize the receptor binding domain of the SARS-CoV-2 spike protein (RBD) and sequence data of 3,702 antibodies that target other regions of the spike protein (non-RBD) were extracted and used. For the sequence data of novel antibodies, 52,799 BCR sequence data obtained from COVID-19 patients were used. Since it is unknown whether the novel antibodies are RBD or non-RBD antibodies and cannot be labeled as either, hereinafter, the sequence data of the novel antibodies will also be referred to as unlabeled sequence data (or unlabeled).

[0032] As in the case of Figure 4, paratope pseudosequences were created from antibody sequence data containing RBD, non-RBD, and unlabeled sequences. Clustering was then performed based on the similarity of the paratope pseudosequences, with a sequence identity threshold of 80% and a coverage threshold of 90%. Of the 52,799 unlabeled sequence data from COVID-19 patients, 1,863 were clustered with sequence data of known antibodies obtained from the BCR database, resulting in a total of 602 clusters. Of the 1,863 clusters, the 25 largest clusters were extracted, and the proportions of RBD, non-RBD, and unlabeled sequences in each cluster are shown in the graph in Figure 5. The horizontal axis of the graph in Figure 5 represents the cluster name, and the vertical axis represents the cluster size. Referring to the graph in Figure 5, it can be seen that RBD-based clusters such as OUCov1 to OUCov3 and OUCov11 contain unlabeled sequence data, while non-RBD-based clusters such as OUCov4, OUCov6 to OUCov8, OUCov14, OUCov17, and OUCov20 contain other unlabeled sequence data. The results of co-clustering the labeled sequence data (RBD and non-RBD) and the unlabeled sequence data suggest that CDR clustering can be used to detect public antibody responses after SARS-CoV-2 infection.

[0033] Figure 6 is an explanatory diagram illustrating the evaluation results using cluster representatives. A total of five unlabeled BCRs were selected from the RBD-based clusters OUCov1 and 3, and the non-RBD-based clusters OUCov4 and 6, and antibodies were expressed from each of these five BCRs. Hereinafter, antibodies expressed from unlabeled BCRs will also be referred to as unlabeled antibodies. These unlabeled antibodies were used as cluster representatives for each cluster to evaluate their specificity for the full-length SARS-CoV-2 spike protein and RBD-expressing Expi293F cells.

[0034] The evaluation was performed using an enzyme-linked immunosorbent assay (ELISA). ELISA is a type of enzyme immunoassay that uses antibodies to detect or quantify target molecules. The evaluation results are shown in Figure 6. The values ​​(0-100%) in Figure 6 represent the percentage of spike protein or RBD populations that positively bound to unlabeled antibodies. The evaluation results show that all five unlabeled antibodies representing the clusters bound to spike protein, but only unlabeled antibodies expressed from the OUCov1 and 3 clusters bound to RBD. Unlabeled antibodies expressed from the OUCov4 and 6 clusters did not bind to RBD. This indicates that the label assignments predicted from the graph in Figure 5 are correct.

[0035] Furthermore, alignment of CDRHs between the assessed clusters confirmed that the sequences of CDRH1 and CDRH2 are highly conserved but not completely, and that there is greater variability in CDRH3.

[0036] These findings demonstrate that CDR clustering is effective in assigning untagged BCRs to target antigens using a limited dataset of tagged antibodies (a subset within each cluster).

[0037] The operation of the information processing device 10 will be described below. Fig. 7 is a flowchart showing the procedure of processing executed by the information processing device 10 according to embodiment 1. The control unit 11 of the information processing device 10 reads and executes the antibody selection program PG from the storage unit 12, thereby performing the following processing.

[0038] The control unit 11 acquires sequence data of known antibodies and novel antibodies (step S101). Known antibodies are antibodies known to have specificity for a target antigen. In the above example, an antibody that recognizes an RBD corresponds to a known antibody. If a publicly available BCR database exists, such as CovAbDab, the control unit 11 can acquire sequence data of the known antibody by accessing the BCR database via the communication unit 13. The target antigen is not limited to SARS-CoV-2; any antigen can be the target. Furthermore, the target antigen does not have to be a single type; multiple types may be used. When multiple types of antigens are targeted, sequence data of known antibodies with specificity for each type can be acquired. A novel antibody is an antibody whose antigen specificity is unknown. In the above example, an unlabeled antibody corresponds to a novel antibody. Sequence data of the novel antibody is identified using a known analytical method, such as FACS (fluorescence-activated cell sorting) or single-cell immune profiling. In step S101, sequence data of the novel antibody identified using a known analytical method can be acquired. The sequence data of known antibodies is given a label that identifies the type of antigen, and the sequence data of new antibodies is given a label that indicates that the antigen is unknown.

[0039] The control unit 11 clusters the sequence data of the known antibodies and the novel antibodies acquired in step S101 (step S102). Based on the acquired sequence data of each antibody, the control unit 11 creates a paratope pseudosequence by pseudo-linking CDR1, CDR2, and CDR3 segments, and clusters the sequence data from the perspective of similarity of the paratope pseudosequence. The control unit 11 can cluster the sequence data, for example, by setting the sequence identity threshold and coverage threshold to appropriate values ​​and using the MMseqs2 search method.

[0040] The control unit 11 determines clusters of known antibodies based on the clustering results (step S103). By referring to the labels attached to each piece of sequence data, the control unit 11 can determine how much of the sequence data of known antibodies and how much of the sequence data of novel antibodies are contained in each cluster. The control unit 11 determines clusters of known antibodies by calculating the proportion of known antibodies contained in each cluster (cluster purity).

[0041] Based on the clustering results of step S102 and the discrimination results of step S103, the control unit 11 selects a new antibody having specificity for the target antigen (step S104). The control unit 11 references the labels attached to the sequence data to determine whether the new antibody is included in the cluster of known antibodies. If it is determined that the new antibody is included in the cluster of known antibodies, it infers that the antigen specificity of the new antibody is the same as the antigen specificity of the known antibody in the same cluster. Through the above procedure, the control unit 11 can select a new antibody having specificity for the target antigen.

[0042] As described above, in the first embodiment, sequence data of antibodies with known antigen specificity and sequence data of novel antibodies with unknown antigen specificity are clustered, and the antigen specificity of a subset (known antigens) within the cluster is identified, thereby making it possible to select novel antibodies with specificity for a specific antigen.

[0043] (Embodiment 2) In Embodiment 1, sequence data of antibodies with known antigen specificity is acquired in advance and clustered together with sequence data of antibodies with unknown antigen specificity, but it is also possible to acquire only sequence data of antibodies with unknown antigen specificity, perform clustering, and then identify antigen specificity for a subset within the cluster. In Embodiment 2, a configuration will be described in which antigen specificity is identified for a subset within a cluster after clustering.

[0044] 8 is an explanatory diagram outlining an antibody selection method according to embodiment 2. In embodiment 2, sequence data of antibodies with unknown antigen specificity is acquired, and the acquired data is clustered based on the similarity of paratope pseudo sequences. The clustering method is the same as that of embodiment 1. For example, sequence data can be clustered by setting a sequence identity threshold and a coverage threshold and using the MMseqs2 search method.

[0045] Of the clustered non-singleton clusters, 25 clusters are extracted in descending order of cluster size, and the cluster size of each cluster is plotted on a graph, such as that shown in the upper right of Figure 8. At this point, the antigen specificity of the antibody identified by the sequence data in each cluster is unknown.

[0046] In the second embodiment, a small number of sequence data (e.g., 1 to 4) are selected from each cluster, and the antigen specificity of the small number of antibodies having the selected sequence data is identified. The set of the small number of sequence data (or antibodies) selected from each cluster is also called a cluster representative or subset.

[0047] As described in the first embodiment, antibodies having sequence data in the same cluster can be inferred to have the same antigen specificity. Therefore, by identifying the antigen specificity of a cluster representative antibody (an antibody in a subset), the antigen specificity of other antibodies in the same cluster (antibodies outside the subset) can be inferred. For example, as shown in the graph in the lower right of Figure 8, if the antigen specificity is identified using the cluster representative of cluster 1 and found to be specific to antigen A, it can be inferred that antibodies other than the cluster representative in cluster 1 also have specificity to antigen A. Furthermore, if the antigen specificity is identified using the cluster representative of cluster 3 and found to be specific to antigen B, it can be inferred that antibodies other than the cluster representative in cluster 3 also have specificity to antigen B. The same applies to the antigen specificity of antibodies in other clusters.

[0048] Below, we demonstrate the utility of CDR clustering methods in the annotation of antigens associated with other diseases. The DTP vaccine was chosen as an immune perturbation model because of its proven efficacy in providing robust and broad protection against three highly lethal pathogens: diphtheria (D), tetanus (T), and pertussis (P). To this end, BCR data were collected from four healthy donors who received the trivalent DTP vaccine.

[0049] Figure 9 is an explanatory diagram illustrating the BCR data collection process. In this embodiment, peripheral blood mononuclear cells (PBMCs) were collected from donors who received a DTP booster vaccination 7 days after the booster administration. CD19+ CD27+ antigen-experienced B cells were selected from the collected PBMCs, and four distinct B cell subgroups were separated using fluorescently labeled DTP toxin / toxoid as an antigen probe. The four subgroups included three subgroups bound to diphtheria toxin (DT), tetanus toxin (TT), and pertussis toxin (PT), and one subgroup (NB) not bound to any of the toxins.

[0050] The four subgroups were then subjected to single-cell immune profiling and passed through a quality control procedure, resulting in a total of 13,847 BCR sequences consisting of both heavy (H) and light (L) chains.

[0051] Both clonotype and CDR clustering were performed on the obtained BCR sequences, resulting in a total of 11,636 clonotype clusters and 10,802 CDR clusters. CDR clustering yielded more non-singleton clusters than clonotype clustering.

[0052] Of the 4243 BCR sequences belonging to 1198 non-singleton clusters, we found 69 clusters with members assigned to multiple V genes, 46 clusters with members assigned to multiple J genes, and 1 cluster with members assigned to multiple V and J genes.

[0053] A total of 116 (9.68%) clusters were polyclonal, i.e., composed of multiple VDJ clonotypes. These results highlight the distinction between clonotype and CDR clusters, the latter revealing the convergence of different clonotypes onto functionally similar paratopes in antibody responses after DTP vaccination. However, clonotype and CDR clusters were similarly represented, suggesting that CDR usage does not lead to excessive clustering.

[0054] Figure 10 shows the results of clustering using antigen probes. To verify the antigen specificity in clustering using antigen probes, 25 clusters were extracted in descending order of cluster size, and a label indicating the type of antigen was provisionally assigned to each BCR sequence based on the antigen probes (DT, TT, PT, NB) bound during the cell sorting process. The labels assigned to each BCR sequence include DT, TT, PT, and Neg, which does not correspond to any of the above. The numbers 1 to 25 shown in the figure indicate cluster numbers.

[0055] Cluster 1 indicates a cluster containing only BCR sequences bound to TT. Clusters 3 to 5, 7, and 8 indicate clusters containing only BCR sequences that did not bind to DT, TT, or PT. On the other hand, other clusters contain BCR sequences bound to multiple types of antigens. The cluster purity of some clusters was low, and it was observed that antibodies binding to different types of antigens were present in such clusters. The average cluster purity of clusters using provisional labels was 82%, which was lower than the 95% observed in the COVID-19 analysis shown in embodiment 1.

[0056] Discrepancies in cluster purity between CDR clustering and tentative label assignment using antigen probes may be due to (1) nonspecific binding of fluorophores, streptavidin, and antigen purification tags, leading to lax gating strategies and high false positive rates, (2) sticky or polyreactive BCRs, and (3) reduced cluster purity when applied to raw BCR data without validated antibodies.

[0057] To address these clustering results, we sampled a small number of BCR sequences from each cluster, expressed actual antibodies, and evaluated the antigen specificity of each antibody. Figure 11 is an explanatory diagram illustrating the evaluation method. From each cluster, we sampled one to four BCR sequences with different provisional label assignments. For example, cluster 12 contains BCR sequences that bind to DT, TT, and PT, so at least three BCR sequences including these were sampled as a subset. We expressed antibodies (human IgG1 monoclonal antibodies) based on a total of 52 BCR sequences selected from 25 clusters. Indirect ELISA was performed to evaluate the binding of these antibodies to each DTP toxoid (DT, TT, PT) and the SARS-CoV-2 spike protein as an unrelated antigen.

[0058] Figure 12 is an explanatory diagram illustrating the results of the antigen specificity evaluation. Of the 52 antibodies derived from the top 25 clusters, 10 antibodies showed no reactivity to DT, TT, or PT, while 16 antibodies specifically bound to DT, 2 antibodies specifically bound to TT, and 24 antibodies specifically bound to PT. Although some antibodies in Cluster 9 and Cluster 13 tested negative, none of the clusters contained members that tested positive for multiple antigens. Therefore, within the range of sampling performed, the corrected cluster purity was found to be 96%. This is consistent with the COVID-19 evaluation described in Embodiment 1.

[0059] Based on these evaluation results, seven clusters specific for DT, one for TT, and eight for PT were identified, respectively. These results validated the antigen specificity of the CDR clusters and selected novel DTP-binding antibodies with potential neutralizing activity. By mapping all members of 25 large clusters, the specificities of over 800 unknown antibodies were elucidated by evaluating only 52 antibodies.

[0060] 13 is a flowchart showing the procedure of processing executed by the information processing device 10 according to embodiment 2. The control unit 11 of the information processing device 10 reads and executes the antibody selection program PG from the storage unit 12, thereby performing the following processing.

[0061] The control unit 11 acquires sequence data of a new antibody (step S201). In this embodiment, the acquired sequence data is a sequence data in which the target antigen is provisionally labeled by an antigen probe.

[0062] The control unit 11 clusters the sequence data of the novel antibodies acquired in step S201 (step S202). Based on the acquired sequence data of each antibody, the control unit 11 creates a paratope pseudosequence by pseudo-linking CDR1, CDR2, and CDR3 segments, and clusters the sequence data from the perspective of similarity of the paratope pseudosequence. The control unit 11 can cluster the sequence data, for example, by setting the sequence identity threshold and coverage threshold to appropriate values ​​and using the MMseqs2 search method.

[0063] The control unit 11 selects one or more sequence data from the multiple clusters obtained as a result of the clustering (step S203). At this time, if a cluster contains sequence data labeled with multiple antigens, the control unit 11 may select a minimum number of sequence data labeled with different labels. Furthermore, if a cluster contains only one type of label, the control unit 11 may select any one of the sequence data in the cluster.

[0064] For the sequence data selected by the control unit 11, antibodies specified by each sequence data are actually expressed, and the antigen specificity of each antibody is identified using an existing method such as ELISA. The expression of antibodies and the identification of antigen specificity are performed outside the information processing device 10.

[0065] The control unit 11 acquires information on the antigen specificity identified for the antibody (step S204), and based on the acquired information, selects a new antibody having specificity for the target antigen (step S205). The control unit 11 infers the antigen specificity of other antibodies by mapping the antigen specificity of the antibody having the selected sequence data in each cluster (antibodies in the subset) to other antibodies in the cluster (antibodies outside the subset). Through the above procedure, the control unit 11 can select a new antibody having specificity for the target antigen.

[0066] As described above, in the second embodiment, sequence data of novel antibodies with unknown antigen specificity are clustered, and the antigen specificity of a subset within the cluster is identified, thereby making it possible to select novel antibodies with specificity for a target antigen.

[0067] After selecting novel antibodies specific to a target antigen, neutralizing antibodies for each antigen may be identified by observing the neutralizing activity of each antibody. Existing techniques can be used to identify neutralizing antibodies. For example, a cytotoxicity assay may be performed to evaluate the inhibitory effect of a DT-binding antibody on DT-induced VERO cell death. Furthermore, an ELISA assay may be performed to examine the ability of a TT-binding antibody to inhibit TT binding to the GT1b receptor. Furthermore, a CHO-K1 cell line that undergoes morphological changes when exposed to PT may be used to examine neutralization by a PT-binding antibody. Neutralizing antibodies can be identified by observing neutralizing activity using these assays.

[0068] In the first and second embodiments, clustering methods using paratope representation have been described, but if a decrease in sensitivity is acceptable, antibodies may also be clustered using clonotype representation. A clustering method using clonotype representation is a method for clustering antibodies that have the same V and J genes and have similar CDR3 sequences.

[0069] The embodiments disclosed herein are illustrative in all respects and should not be considered limiting. The scope of the present invention is defined by the claims, not by the above meaning, and is intended to include all modifications within the meaning and scope of the claims.

[0070] 10 Information processing device 11 Control unit 12 Memory unit 13 Communication unit 14 Operation unit 15 Display unit PG Antibody selection program

Claims

1. A method for selecting antibodies, the method comprising the steps of: acquiring sequence data of a plurality of antibodies including antibodies whose antigen specificity is unknown; clustering the acquired sequence data of the plurality of antibodies based on the similarity of the paratope pseudosequences; discriminating a cluster having specificity for a particular antigen using a subset of antibodies in each cluster; and, within the discriminated cluster, inferring that antibodies outside the subset have the same antigen specificity as the antibodies in the subset, thereby selecting antibodies having specificity for the particular antigen.

2. The antibody selection method according to claim 1, wherein the plurality of antibodies includes antibodies whose specificity for the particular antigen is known, and the computer executes a process of inferring the antigen specificity of antibodies outside the subset by using the antibodies whose specificity is known in the subset.

3. The method for selecting an antibody according to claim 1, wherein the computer executes the process of identifying the antigen specificity of antibodies within said subset, and inferring the antigen specificity of antibodies outside said subset based on the identification results.

4. The antibody selection method of claim 3, wherein the computer executes a process of expressing one or more antibodies sampled from the subset, and obtaining results of antigen screening for the expressed antibodies, thereby identifying the antigen specificity of the antibodies in the subset.

5. The antibody selection method according to claim 4, wherein the computer executes the following steps: classifying human peripheral blood mononuclear cells collected from multiple donors using an antigen probe; based on the classification results, labeling B cells contained in the human peripheral blood mononuclear cells with a label indicating the type of antigen; and selecting B cells with a different type of label as B cells that produce the one or more antibodies.

6. The antibody selection method according to claim 4, wherein the process of identifying clusters containing antibodies having neutralizing activity against the antigen by observing the neutralizing activity of the antibodies whose antigen specificity has been predicted is carried out by the computer.

7. An antibody selection method according to any one of claims 1 to 6, wherein the paratope pseudo-sequence is constructed by pseudo-linking segments representing three complementarity determining regions of an antigen receptor.

8. A computer program for causing a computer to execute a process of obtaining sequence data of a plurality of antibodies including antibodies of unknown antigen specificity, clustering the obtained sequence data of the plurality of antibodies from the viewpoint of similarity of paratope pseudo sequences, discriminating a cluster having specificity for a particular antigen using a subset of antibodies in each cluster, and selecting an antibody having specificity for the particular antigen by inferring that antibodies outside the subset within the discriminated cluster have the same antigen specificity as the antibodies within the subset.

9. An information processing device comprising at least one processor, which acquires sequence data of a plurality of antibodies including antibodies whose antigen specificity is unknown, clusters the acquired sequence data of the plurality of antibodies in terms of similarity of paratope pseudo sequences, identifies clusters having specificity for a particular antigen using a subset of antibodies within each cluster, and selects antibodies having specificity for the particular antigen within the identified cluster by inferring that antibodies outside the subset have the same antigen specificity as antibodies within the subset.

Images