Antigen-specific binding antibody prediction method and device

Abstract

The application provides an antigen-specific binding antibody prediction method and device, and belongs to the technical field of computer and antibody drug research and development. The method comprises the following steps: determining an actual antigen epitope of a target antigen amino acid sequence; determining N candidate antibody amino acid sequences, determining a predicted antigen epitope corresponding to an i-th candidate antibody amino acid sequence according to the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; wherein N is an integer not less than 1, and i is a positive integer not greater than N; calculating the similarity between the actual antigen epitope and the predicted antigen epitope corresponding to the i-th candidate antibody amino acid sequence, respectively, and sorting the N candidate antibody amino acid sequences according to the similarity to determine an antigen-specific binding antibody. Based on the method, zero-sample antigen-specific binding antibody prediction can be realized under the premise of ensuring prediction accuracy without relying on prior experimental data.

Description

Technical Field

[0001] This application relates to the fields of computer and antibody drug development technology, and in particular to methods for predicting antigen-specific binding antibodies, related devices, electronic devices, computer-readable storage media, and computer program products. Background Technology

[0002] Understanding the interaction between antibodies and their specific antigens is crucial in disease prevention and control and the development of biopharmaceuticals. In particular, predicting the antigenic epitopes that antibodies can specifically bind to (i.e., the parts of the antigen that bind to antibodies) is of great significance for developing effective vaccines and drugs.

[0003] Computational methods for predicting antigen-specific binding antibodies in related technologies include molecular docking and molecular dynamics simulations, candidate antibody scoring models, and the Prodigy method. Among these, candidate antibody scoring models are obtained by training on labeled antibody data and rely on a large amount of prior experimental data to achieve predictions. Summary of the Invention

[0004] This application aims to at least partially address one of the aforementioned technical problems or at least provide a useful commercial option. In view of this, this application provides a method, related apparatus, electronic device, computer-readable storage medium, and computer program product for predicting antigen-specific binding antibodies, which, while ensuring prediction accuracy, can achieve zero-sample antigen-specific binding antibody prediction without relying on prior experimental data.

[0005] In a first aspect, this application proposes a method for predicting antigen-specific binding antibodies. The method includes: determining the actual antigenic epitope of a target antigen amino acid sequence; determining N candidate antibody amino acid sequences; and determining the predicted antigenic epitope corresponding to the i-th candidate antibody amino acid sequence based on the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; wherein N is an integer not less than 1, and i is a positive integer not greater than N; calculating the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the i-th candidate antibody amino acid sequence; and sorting the N candidate antibody amino acid sequences according to the similarity to determine the antigen-specific binding antibody.

[0006] This application provides a novel scheme for predicting antigen-specific antibodies. It determines the actual antigenic epitope of the target antigen's amino acid sequence and the predicted antigenic epitope corresponding to the candidate antibody's amino acid sequence, and predicts the antigen-specific binding antibody based on the similarity between the two. This antigen-specific antibody prediction scheme does not rely on known three-dimensional structures of antigens and antibodies or large amounts of experimental data. While ensuring accuracy, it can achieve zero-sample prediction, is effectively used for screening antigen-corresponding antibodies against novel pathogens, is easy to implement, and has greater adaptability.

[0007] In some possible implementations, the aforementioned determination of the actual antigenic epitope of the target antigen amino acid sequence includes: inputting the target antigen amino acid sequence and the ligand amino acid sequence into a structure prediction tool to determine the structure of the target antigen-ligand complex; wherein the target antigen amino acid sequence and the ligand amino acid sequence are at least partially specifically bound; based on the target antigen-ligand complex structure, determining the amino acids whose distance from the target antigen amino acid sequence to the ligand amino acid sequence is less than the aforementioned predetermined threshold, and determining the actual antigenic epitope of the target antigen amino acid sequence.

[0008] By inputting the target antigen and ligand amino acid sequences into a structure prediction tool, the structure of the target antigen-ligand complex is determined, and the actual antigenic epitope is identified based on this structure. This method overcomes the shortcomings of traditional techniques that rely on experimental data and structural information for antigenic epitope identification. By calculating the binding relationship between the target antigen and ligand amino acid sequences, it provides a more accurate and effective basis for prediction.

[0009] In some other possible implementations, the aforementioned determination of the actual antigenic epitope of the target antigen amino acid sequence includes: inputting the target antigen amino acid sequence into an antigenic epitope prediction tool to determine the actual antigenic epitope of the target antigen amino acid sequence.

[0010] An epitope prediction tool was introduced into the step of determining the actual epitope of the target antigen's amino acid sequence. This method avoids dependence on large amounts of experimental data and structural information, directly identifying the actual epitope of the target antigen through the prediction tool, thereby simplifying the prediction process and improving efficiency. This step solves the problems of cumbersome and data-dependent prediction of epitopes through experiments or protein structure in traditional methods, making the prediction of antigen-specific binding antibodies more efficient and rapid.

[0011] In some possible implementations, determining the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody based on the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody includes: inputting the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody into a structure prediction tool to determine the structure of the target antigen-i-th candidate antibody complex; and based on the structure of the target antigen-i-th candidate antibody complex, determining amino acids whose distance between the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody is less than a predetermined threshold, and determining the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody.

[0012] In determining the predicted antigenic epitope based on the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody, the structure of the target antigen-antibody complex is calculated by inputting a structure prediction tool, and the interaction between the antibody and the antigenic epitope is identified based on this complex structure. This method solves the problem of accurately predicting the interaction pattern between antigen and antibody in existing antibody prediction technologies. Traditional methods usually rely on empirical formulas or experimental data, but these methods are difficult to accurately reflect the actual binding between the antibody and the target antigen, and their prediction effect is poor for novel antigens and antibodies. By introducing a structure prediction tool, this method can provide more accurate predictions based on structural information even without a large amount of experimental data.

[0013] In some possible implementations, the similarity between the aforementioned actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the aforementioned i-th candidate antibody is calculated, including: calculating the similarity between the aforementioned actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the aforementioned i-th candidate antibody based on the number of amino acid overlaps in the aforementioned actual antigenic epitope and the number of amino acids in the aforementioned actual antigenic epitope.

[0014] By quantifying the similarity between actual and predicted antigenic epitopes based on the number of amino acid overlaps between the actual and predicted antigenic epitopes, and the number of amino acids in the actual antigenic epitope, candidate antibodies can be ranked more objectively and accurately.

[0015] In some possible implementations, the similarity between the aforementioned actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the aforementioned i-th candidate antibody is calculated, including: calculating the similarity between the aforementioned actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the aforementioned i-th candidate antibody based on the ratio of the number of amino acid overlaps in the aforementioned actual antigenic epitope and the number of amino acids in the aforementioned predicted antigenic epitope.

[0016] Calculating similarity based on the ratio of the number of amino acid overlaps to the number of amino acids in the actual antigenic epitope can avoid uneven distribution of amino acid weights, making the similarity assessment more accurate and enabling a more objective and precise ranking of candidate antibodies.

[0017] In some possible implementations, the aforementioned calculation of the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody includes: inputting the amino acid sequences of the actual antigenic epitope and the predicted antigenic epitope into a structural similarity prediction tool, and calculating the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody.

[0018] Structural similarity prediction tools can take into account the spatial configuration and structural features of antigenic epitopes, thereby improving the accuracy of similarity assessment.

[0019] In some possible implementations, the aforementioned sorting of the aforementioned N candidate antibody amino acid sequences based on the aforementioned similarity to determine the antigen-specific binding antibody includes: sorting the aforementioned N candidate antibody amino acid sequences in descending order of similarity to obtain a first queue; screening out target candidate antibody amino acid sequences from the aforementioned first queue to obtain a second queue, wherein the target candidate antibody amino acid sequences contain interacting amino acids whose distance from the aforementioned predicted antigen epitope amino acid sequence is less than the aforementioned predetermined threshold, and the aforementioned interacting amino acids include the heavy chain complementarity-determining region amino acids of the aforementioned target candidate antibody; and determining the antigen-specific binding antibody based on the aforementioned second queue.

[0020] By introducing threshold screening and considering specific amino acids that interact with antigen epitopes (heavy chain complementarity-determining region amino acids), the relevance and specificity of candidate antibodies during the screening process are improved, thereby enabling more precise selection of antibodies with high affinity and specificity.

[0021] In some implementations, the aforementioned heavy chain variable region is the heavy chain third complementarity determination region HCDR3.

[0022] By confining the interacting amino acids to the HCDR3 region of the antibody, the screening efficiency and accuracy for antibodies with actual functional binding capabilities are improved. As the most critical region for antigen recognition within the antibody's variable region, HCDR3 ensures that the screened antibodies not only match the antigen in overall structure, but also that their key binding regions actually interact with the antigen epitopes, thereby enhancing the antibody's functional reliability and clinical applicability.

[0023] In some possible implementations, the aforementioned distance is determined as follows: based on the coordinates of the α-carbon atom of each amino acid in the target antigen amino acid sequence and the ligand amino acid sequence; or based on the coordinates of the α-carbon atom of each amino acid in the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; or based on the coordinates of the α-carbon atom of each amino acid in the target candidate antibody and the predicted antigen epitope amino acid sequence.

[0024] By calculating distances using the coordinates of the α-carbon atom of amino acids, accurate assessment of the true three-dimensional spatial relationships between key amino acids in antigen epitope recognition or antibody screening can be achieved, avoiding judgment biases caused by amino acid sequence position or protein conformation errors. Furthermore, it can effectively improve the consistency between the spatial accuracy and physical meaning of antigen-antibody structure prediction and screening, further enhancing the accuracy and functional relevance of screening results in practical applications, and contributing to the acquisition of antibodies with higher binding specificity.

[0025] In some possible implementations, the aforementioned predetermined threshold is determined by: determining the predetermined threshold based on the interaction force between the aforementioned target antigen amino acid sequence and the aforementioned ligand amino acid sequence; or determining the predetermined threshold based on the interaction force between the aforementioned target antigen amino acid sequence and the aforementioned i-th candidate antibody amino acid sequence; or determining the predetermined threshold based on the interaction force between the aforementioned target candidate antibody and the aforementioned predicted antigen epitope amino acid sequence.

[0026] By dynamically determining the distance threshold by incorporating intermolecular interaction forces, an adaptive and biologically relevant structural screening criterion is provided. This criterion can more accurately distinguish candidate antibodies with actual binding ability, reduce the probability of false positives, and help screen for antibodies that can effectively bind to target antigens under physiological conditions, thereby improving the specificity and functional relevance of the prediction results.

[0027] In a second aspect, this application proposes an antigen-specific binding antibody prediction device, comprising: a first determining module for determining the actual antigenic epitope of a target antigen amino acid sequence; a second determining module for determining N candidate antibody amino acid sequences, and determining the predicted antigenic epitope corresponding to the i-th candidate antibody amino acid sequence based on the aforementioned target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; wherein N is an integer not less than 1, and i is a positive integer not greater than N; and a third determining module for calculating the similarity between the aforementioned actual antigenic epitope and the predicted antigenic epitope corresponding to the i-th candidate antibody amino acid sequence, and sorting the aforementioned N candidate antibody amino acid sequences according to the aforementioned similarity to determine the antigen-specific binding antibody.

[0028] In a third aspect of this application, an electronic device is provided, comprising: a processor and a memory; the memory for storing a computer program; and the processor for executing the computer program to implement the antigen-specific binding antibody prediction method as described in the first aspect.

[0029] In a fourth aspect of this application, a computer-readable storage medium is provided that stores computer instructions or programs that, when executed on a computer, cause the antigen-specific binding antibody prediction method of the first aspect to be performed.

[0030] In a fifth aspect of this application, a computer program product is proposed, comprising computer instructions that, when some or all of the computer instructions are executed on a computer, cause the antigen-specific binding antibody prediction method of the first aspect to be performed.

[0031] The aforementioned electronic devices, computer-readable storage media, and computer program products achieve highly efficient automation and improve prediction efficiency and accuracy through the automatic execution of antigen-specific binding antibody prediction methods via computer instructions. Furthermore, the instruction-based nature of these methods ensures high consistency and reliability in complex experimental scenarios.

[0032] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a schematic diagram of a system architecture according to an embodiment of this application;

[0035] Figure 2 This is a flowchart illustrating an antigen-specific binding antibody prediction method according to an embodiment of this application;

[0036] Figure 3 This is a schematic diagram of the SARS-CoV-2 antigen ligand and antigen-antibody co-crystal structure according to an embodiment of this application;

[0037] Figure 4 This is a schematic diagram of an antigen-specific binding antibody screening process based on amino acids in the heavy chain complementarity-determining region according to an embodiment of this application;

[0038] Figure 5 This is a schematic diagram of an antigen-specific binding antibody screening process based on the heavy chain third complementarity-determining region HCDR3 according to an embodiment of this application.

[0039] Figure 6 This is a schematic diagram illustrating an antigenic epitope based on ligands and antibodies according to another embodiment of this application;

[0040] Figure 7 This is a schematic diagram of an antigen-specific binding antibody prediction device according to an embodiment of this application;

[0041] Figure 8 This is a schematic block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0044] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0045] Before introducing the technical solution of this application, the relevant knowledge of this application will be introduced below:

[0046] In this application, "antigen" refers to a molecule that can be recognized by the immune system and trigger an immune response. It can be a protein, polypeptide, carbohydrate, lipid, or nucleic acid. An antigen provides a specific binding site in a complex and is typically composed of specific domains or epitopes on its surface.

[0047] In this embodiment of the application, an "antigenic epitope," also known as an antigenic determinant, is a specific portion of an antigen molecule that binds to an antibody or T-cell receptor. Epitopes can be linear (composed of a continuous amino acid sequence) or conformational (determined by the three-dimensional structure of a protein).

[0048] In this application, "novel pathogen" refers to a recently emerged, spread, or identified pathogenic microorganism, including viruses, bacteria, fungi, or parasites. These pathogens may lead to new disease outbreaks or the spread of infectious diseases.

[0049] In this application, a "ligand" is a molecule capable of specifically binding to an antigen. In immunology, ligands are often antibodies (or fragments thereof), T-cell receptors, chemical probes, or other biomolecules. The structure of a ligand determines its ability to recognize and bind to antigens.

[0050] In this embodiment, "antibody" refers to immunoglobulin, whose main function is to recognize and bind to specific antigens. Antibodies have high specificity, capable of precisely binding to specific parts of antigens (antigenic epitopes), thereby helping the immune system recognize and neutralize pathogens (such as viruses and bacteria) or abnormal cells (such as cancer cells). Common antibodies consist of two heavier chains and two lighter chains linked by interchain disulfide bonds. The amino acid composition and sequence of the constant region of the immunoglobulin heavy chain differ, thus their antigenicity also differs. Immunoglobulins can be classified into five classes: IgM, IgD, IgG, IgA, and IgE. The same class of immunoglobulins can also be divided into different subclasses based on their amino acid composition, such as IgG1, IgG2, IgG3, and IgG4. Immunoglobulin light chains are divided into κ chains or λ chains based on their constant regions. In some examples, the antibody heavy chain variable region may further include a heavy chain constant region containing human or mouse IgG1, IgG2, IgG3, IgG4 or variants thereof; the antibody light chain variable region may further include a light chain constant region containing human or mouse κ, λ chains or variants thereof.

[0051] The sequence of approximately 110 amino acids near the N-terminus of both the antibody heavy and light chains varies considerably and is known as the variable region (Fv region); the remaining amino acid sequences near the C-terminus are relatively stable and are called the constant region. The variable region includes three hypervariable regions (HVRs) and four relatively conserved backbone regions (FRs). The three hypervariable regions determine the antibody's specificity and are also called complementarity-determining regions (CDRs). Each light chain variable region (LCVR) and heavy chain variable region (HCVR) consists of three CDRs and four FRs, arranged in the following order from the amino terminus to the carboxyl terminus: FR1, CDR1, FR2, CDR2, FR3, CDR3, FR4. The three CDRs of the light chain refer to LCDR1, LCDR2, and LCDR3; the three CDRs of the heavy chain refer to HCDR1, HCDR2, and HCDR3.

[0052] In this application, an "antigen-ligand complex" refers to a complex formed by an antigen and a molecule (ligand) that binds to it. The ligand can be an antibody, a receptor, or other molecules that bind to the antigen. This complex can help in the study of antigen recognition and immune responses.

[0053] In the embodiments of this application, "antigen-candidate antibody complex" refers to a complex formed by the binding of a candidate antibody and an antigen. This binding is usually specific, with the antibody's variable region recognizing a specific epitope of the antigen. This binding process is one of the key steps in the immune response.

[0054] In the embodiments of this application, the "α-carbon of an amino acid" is the central carbon atom connected to an amino group (NH2), a carboxyl group (COOH), a hydrogen atom (H), and a side chain (R group). The α-carbon is a key component of the protein backbone, participating in the formation of peptide bonds and influencing the three-dimensional structure of proteins.

[0055] In this application, "interaction force" refers to the force of attraction or repulsion between substances or particles due to their properties. It is the basis of the interaction between substances and is usually realized through force fields, the interaction of electrons, protons, or other particles between molecules. Interaction forces can be physical (such as gravity, electrostatic force) or chemical (such as chemical bonds, van der Waals forces, etc.). Different types of interaction forces determine the relative positions, structures, and stability of molecules, atoms, or particles.

[0056] In the embodiments of this application, "actual antigenic epitope" refers to the amino acid sequence of the target antigen at the binding site of the target antigen and ligand; "predicted antigenic epitope" refers to the amino acid sequence of the target antigen at the binding site of the target antigen and candidate antibody.

[0057] In this embodiment, "ANARCI" is a program for numbering antibody and T-cell receptor sequences (https: / / academic.oup.com / bioinformatics / article / 32 / 2 / 298 / 1743894). It uses HMMER to search for immunoglobulin superfamily (IgSF) domains in the input sequence, and then numbers the found domains according to a user-specified numbering scheme (e.g., IMGT, AHo, Martin, etc.), helping researchers accurately identify and compare the CDR regions of antibodies. This standardized approach plays a crucial role in antibody engineering, vaccine design, and other biomedical applications. By specifying a numbering scheme, researchers can better understand the binding characteristics of antibodies to antigens and perform corresponding sequence optimization and design.

[0058] In this embodiment, the "International Immunogenomics Information System (IMGT)" is a standard database and analysis system specifically designed for studying antibody and T-cell receptor gene sequences. It defines a set of internationally accepted rules and standards for the sequence and structural analysis of antibodies or T-cell receptors. Key aspects of the IMGT rules include: a. Standardized Nomenclature: IMGT provides a systematic nomenclature scheme for consistently describing the genes, sequences, and regions of immunoglobulins (such as antibodies) and T-cell receptors. b. Partitioning System: IMGT standardizes the partitioning of the variable region of antibodies, defining the frame region (FR) and complementarity-determining region (CDR). In antibodies, CDRs are key regions for antigen binding. IMGT defines three CDR segments: CDR1, CDR2, and CDR3. CDR3 is the longest and most diverse segment, typically directly involved in antigen binding. c. Sequence Numbering: IMGT uses an internationally recognized sequence numbering system to standardize the variable region (V region) of antibodies into a common reference frame. This numbering method makes comparison and annotation between different antibody sequences more convenient and accurate. d. Structure-based annotation: By integrating sequence and three-dimensional structural information, IMGT provides detailed structural annotations, which helps to understand the functional mechanism of antibodies.

[0059] The technical problems to be solved, the inventive concept and the system architecture of the embodiments of this application will be described below:

[0060] As mentioned above, various computational methods are currently available for predicting antigen-specific binding antibodies, such as molecular docking and molecular dynamics simulations, candidate antibody scoring models based on deep learning methods, and the Prodigy method. However, each of these methods has certain limitations, such as reliance on previous experimental data, poor generalization ability, and insufficient prediction accuracy.

[0061] Specifically, molecular docking and molecular dynamics simulations aim to predict the three-dimensional conformation of antigen-antibody complexes and calculate their binding energies (usually characterized by docking scores). Molecular docking software (such as AutoDock, HADDOck, and ZDOCK) is commonly used to predict antigen-antibody interaction patterns and binding sites, while molecular dynamics simulations delve into the dynamic behavior during antigen-antibody binding. However, this method heavily relies on protein co-crystal structures, and due to the scarcity of protein crystallization data, it is often difficult to obtain sufficient information to screen for antigen-specific antibodies.

[0062] Deep learning methods, on the other hand, train labeled antibody data to build scoring models for evaluating candidate antibodies. While this method can predict antibody characteristics to some extent, it is prone to overfitting due to the scarcity of labeled data. This is especially true when the deep learning model is highly complex; insufficient data leads to weak model generalization ability, making it difficult to train an efficient and accurate model.

[0063] The Prodigy method predicts protein-protein binding affinity and dissociation constants by calculating the number of contact points between proteins, the types and quantities of various interactions (such as charged-charged, charged-polar, charged-nonpolar, and polar-polar interactions), and the proportion of non-interacting surface (NIS) residues. This method calculates protein-protein interactions based on empirically derived formulas, but it only applies to protein interaction patterns similar to those in the training data. For protein interactions that have not been observed before, the prediction effect is poor, and it is difficult to extend to novel antigen-antibody interactions.

[0064] To address the technical problems of the aforementioned prior art, this application proposes a method for predicting antigen-specific binding antibodies. This method can predict antigen-specific binding antibodies based on the similarity between actual and predicted antigenic epitopes under zero-sample conditions, avoiding the high dependence of molecular docking and molecular dynamics simulation methods on large amounts of experimental data or protein co-crystal structures. The zero-sample prediction strategy of this method does not rely on complex deep learning models, nor does it require training the model with prior data, overcoming the problems of overfitting, insufficient generalization ability, and poor model performance caused by training the model with scarce labeled data in traditional methods. Compared to methods such as Prodigy that calculate protein-protein interactions based on empirical formulas, the zero-sample prediction strategy of this method also avoids the limitation that empirical formulas are only applicable to similar protein interaction modes, and can effectively predict protein interactions that have not occurred before, expanding its application scope. Overall, this method enhances the generalization ability of prediction, enabling it to maintain high prediction accuracy even when facing novel antigens or antibodies.

[0065] In some embodiments of this application, the system architecture is as follows: Figure 1 As shown.

[0066] Figure 1 This is a schematic diagram of a system architecture according to an embodiment of this application. The system architecture involves a computing device 10 and may also include a network, storage device, etc.

[0067] The computing device 10 is used to run a program for predicting antigen-specific binding antibodies. Specifically, the computing device 10 receives the actual antigenic epitope 11 of the target antigen amino acid sequence and the predicted antigenic epitope 12 corresponding to the amino acid sequence of the i-th candidate antibody, executes the antigen-specific binding antibody prediction scheme, and outputs the antigen-specific binding antibody 13.

[0068] refer to Figure 1 The antigen-specific binding antibody prediction scheme executed by the computing device 10 may be to calculate the similarity between the actual antigen epitope and the predicted antigen epitope corresponding to the amino acid sequence of the i-th candidate antibody, and sort the N candidate antibody amino acid sequences according to the similarity to determine the antigen-specific binding antibody.

[0069] The computing device 10 can be a server.

[0070] For example, the server can be a rack server, blade server, tower server, or cabinet server, etc. The server can be a standalone test server or a test server cluster composed of multiple test servers.

[0071] In this embodiment, the computing device 10 is connected to the storage device via a network. The network can be an intranet, the Internet, Global System for Mobile Communications (GSM), Wideband Code Division Multiple Access (WCDMA), 4G, 5G, Bluetooth, Wi-Fi, or a voice communication network, or other wireless or wired networks. The storage device can be any available medium accessible to the computing device 10, or a data storage device such as a server or data center that integrates one or more available media. This available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0072] The technical solution of this application will be described in detail below:

[0073] Figure 2 This is a flowchart illustrating an antigen-specific binding antibody prediction method according to an embodiment of this application, wherein the execution entity of the aforementioned method can be a server, such as... Figure 1 The computing device 10; or an electronic device, computer-readable storage medium, or computer program product storing the relevant algorithm, etc. The aforementioned method is illustrated using a server as the executing entity. (See reference...) Figure 2 The aforementioned methods include:

[0074] S210, determine the actual antigenic epitopes of the target antigen amino acid sequence;

[0075] In some implementations, determining the actual antigenic epitope of the target antigen amino acid sequence includes: inputting the target antigen amino acid sequence and the ligand amino acid sequence into a structure prediction tool to determine the target antigen-ligand complex structure; wherein the target antigen amino acid sequence and the ligand amino acid sequence bind at least partially specifically; based on the target antigen-ligand complex structure, determining amino acids whose distance from the target antigen amino acid sequence to the ligand amino acid sequence is less than a predetermined threshold, and determining the actual antigenic epitope of the target antigen amino acid sequence. Calculating the amino acid spatial distance between the target antigen and ligand through the target antigen-ligand complex structure allows for a deeper understanding of the interaction between the target antigen and ligand, effectively improving the accuracy of actual antigenic epitope estimation.

[0076] It is understandable that the aforementioned structural prediction tools can be selected from open-source tools in this field, such as AlphaFold-Multimer (AFM) and tfold-ag.

[0077] For example, AlphaFold-Multimer was used as a structure prediction tool to predict the complex structure of the amino acid sequences of SARS-CoV-2 (green, target antigen) and its receptor angiotensin-converting enzyme 2 (ACE2) (gray, ligand), such as... Figure 3 As shown in Figure A. Further, based on the complex structure formed by SARS-CoV-2 and ACE2, the amino acids in SARS-CoV-2 and ACE2 with a distance less than a predetermined threshold were calculated; these amino acids are the actual antigenic epitopes of SARS-CoV-2.

[0078] In some implementations, the distance is determined based on the coordinates of the α-carbon atom of each amino acid in the target antigen and ligand amino acid sequences. Calculating the distance between amino acids based on the coordinates of the α-carbon atom of each amino acid helps to more accurately identify key interaction regions between the target antigen and ligand, effectively identifying actual binding sites and improving the accuracy of actual antigen epitope estimation.

[0079] For example, the distance referred to can be the Euclidean distance, specifically, the formula for calculating the Euclidean distance is as follows:

[0080]

[0081] Among them, (x i ,y i ,z i (x) represents the coordinate of the α-carbon in the i-th amino acid of the target antigen sequence. t ,y t ,z t) represents the coordinate of the α carbon of the t-th amino acid in the amino acid sequence of the ligand.

[0082] Determining distances using the Euclidean distance formula allows for an accurate measurement of the three-dimensional spatial proximity between amino acid residues, which helps in the rapid identification of the actual antigenic epitopes of the target antigen.

[0083] In some implementations, the predetermined threshold is determined based on the interaction force between the target antigen amino acid sequence and the ligand amino acid sequence. Determining the predetermined threshold based on the interaction force between the target antigen and the ligand can improve the accuracy and reliability of actual antigen epitope estimation. By using interaction forces, the predetermined threshold can be dynamically adjusted to better reflect actual binding affinity and interaction strength, thus helping to reduce misjudgments caused by setting unreasonable thresholds.

[0084] It is understandable that the aforementioned interaction forces can be weak interaction forces, such as van der Waals forces.

[0085] In some implementations, determining the actual antigenic epitope of the target antigen amino acid sequence includes: inputting the target antigen amino acid sequence into an antigenic epitope prediction tool to determine the actual antigenic epitope of the target antigen amino acid sequence. By inputting the target antigen amino acid sequence into the antigenic epitope prediction tool, the actual antigenic epitope can be identified quickly and accurately.

[0086] The aforementioned antigenic epitope prediction tools can be selected from open-source tools in this field, such as those mentioned in https: / / www.nature.com / articles / s41598-022-18021-1.

[0087] In some implementations, the target antigen and ligand referred to in this application are both proteins.

[0088] S220, determine N candidate antibody amino acid sequences, and determine the predicted antigen epitope corresponding to the i-th candidate antibody amino acid sequence based on the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; where N is an integer not less than 1, and i is a positive integer not greater than N;

[0089] In some implementations, determining the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody based on the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody includes: inputting the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody into a structure prediction tool to determine the structure of the target antigen-i-th candidate antibody complex; and based on the structure of the target antigen-i-th candidate antibody complex, determining amino acids whose distance between the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody is less than a predetermined threshold, and determining the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody. Traditional methods typically rely on empirical formulas or experimental data, which are difficult to accurately reflect the actual binding between the antibody and the target antigen, and have poor prediction performance for novel antigens and antibodies. The method of this application uses a protein structure prediction tool to simulate the structure of the antigen-candidate antibody complex, and then predicts the antigenic epitope based on the simulated complex structure to achieve the prediction of antigen-specific antibodies, significantly improving prediction accuracy and efficiency. It does not rely on known three-dimensional structures of antigens and antibodies or a large amount of experimental data, and has superior prediction performance for novel antigens and antibodies, with the potential for rapid response to novel pathogens.

[0090] It should be noted that the descriptions of the aforementioned structural prediction tools are the same as above, and will not be repeated here.

[0091] For example, AlphaFold-Multimer was used as a structure prediction tool to predict the complex structure of SARS-CoV-2 (green, target antigen) and candidate antibodies (red for heavy chain and yellow for light chain) amino acid sequences, such as... Figure 3 As shown in B. Further, based on the complex structure formed by SARS-CoV-2 and candidate antibodies, the amino acids in SARS-CoV-2 and candidate antibodies with a distance less than a predetermined threshold were calculated; these amino acids are the predicted antigenic epitopes of SARS-CoV-2. Figure 3 A and Figure 3 The result of B fusion is as follows Figure 3 As shown in C, the accuracy of the candidate antibody prediction can be seen more intuitively.

[0092] In some implementations, the distance is determined based on the coordinates of the α-carbon atom of each amino acid in the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody. Calculating the distance between amino acids based on the coordinates of the α-carbon atom of each amino acid helps to more accurately identify key interaction regions between the target antigen and the candidate antibody, effectively identifying actual binding sites and improving the reliability and accuracy of antigen epitope prediction.

[0093] For example, the distance referred to can be the Euclidean distance, specifically, the formula for calculating the Euclidean distance is as follows:

[0094]

[0095] Among them, (x i ,y i ,z i (x) represents the coordinate of the α-carbon in the i-th amino acid of the target antigen sequence. j ,y j ,z j ) represents the coordinate of the α carbon of the j-th amino acid in the amino acid sequence of the candidate antibody.

[0096] Determining distances using the Euclidean distance formula allows for an accurate measurement of the three-dimensional spatial proximity between amino acid residues, which helps in the rapid identification of predicted antigenic epitopes of target antigens.

[0097] In some implementations, the predetermined threshold is determined based on the interaction force between the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody. Determining the predetermined threshold based on the interaction force between the target antigen and the candidate antibody can improve the accuracy and reliability of the predicted antigen epitope estimation. By using interaction forces, the predetermined threshold can be dynamically adjusted to better reflect actual binding affinity and interaction strength, thus helping to reduce misjudgments caused by setting unreasonable thresholds. Furthermore, it allows for more accurate differentiation of which target antigen-candidate antibody combinations are likely to form stable bindings, thereby improving the specificity of antibody screening.

[0098] It should be noted that the descriptions of the aforementioned interaction forces are the same as above, and will not be repeated here.

[0099] In some implementations, the target antigen and candidate antibody referred to in this application are both proteins.

[0100] S230, calculate the similarity between the actual antigen epitope and the predicted antigen epitope corresponding to the amino acid sequence of the i-th candidate antibody, and sort the N candidate antibody amino acid sequences according to the similarity to determine the antigen-specific binding antibody.

[0101] The antigen-specific binding antibody in this application refers to an antibody that specifically binds to the antigenic epitope of a target antigen. It is understood that the antigen-specific antibody in this application may be one or more.

[0102] In some implementations, the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated separately. This includes calculating the similarity between the actual antigenic epitope and the predicted antigenic epitope based on the number of amino acid overlaps in the actual and predicted antigenic epitopes, and the total number of amino acids in the actual antigenic epitope. By introducing a similarity calculation method based on the number of amino acid overlaps and the total number of amino acids, the overlap rate of amino acids between the actual and predicted antigenic epitopes is quantified, thereby more effectively ranking candidate antibodies and improving the accuracy and specificity of antibody prediction.

[0103] In some implementations, the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated separately. This includes: calculating the similarity between the actual antigenic epitope and the predicted antigenic epitope based on the ratio of the number of amino acid overlaps in the actual antigenic epitope to the number of amino acids in the predicted antigenic epitope. For antigenic epitopes of different lengths, calculating the similarity based on the ratio of the number of amino acid overlaps to the number of amino acids in the actual antigenic epitope achieves a more balanced similarity assessment, better reflecting their similarity and effectively avoiding the unbalanced assessment relying solely on the number of amino acid overlaps, thereby improving the accuracy and reliability of the calculation results. Furthermore, using a ratio further improves the universality of the method, allowing for more objective comparisons when dealing with different types of antigenic epitopes, ultimately contributing to improving the quality and accuracy of antibody prediction.

[0104] In some implementations, N candidate antibodies are scored based on the similarity of amino acids in the actual and predicted antigenic epitopes, and then ranked based on the scores. For example, the formula for calculating similarity can be expressed as:

[0105]

[0106] Here, "Score" represents the similarity score; "number of overlapping amino acids" represents the number of amino acid overlaps between the actual and predicted antigenic epitopes; and "number of amino acids in true epitope" represents the number of amino acids in the actual antigenic epitope.

[0107] By calculating similarity using the ratio of the number of amino acid overlaps to the number of amino acids in the actual antigenic epitope, a more balanced similarity assessment is achieved. This better reflects the similarity between the predicted and actual antigenic epitopes, effectively avoiding the unbalanced assessment that relies solely on the number of amino acid overlaps, thereby improving the accuracy and reliability of the calculation results.

[0108] In some implementations, the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated. This includes inputting the amino acid sequences of the actual and predicted antigenic epitopes into a structural similarity prediction tool, and then calculating the similarity between the actual and predicted antigenic epitopes corresponding to the amino acid sequence of the i-th candidate antibody. By introducing a structural similarity prediction tool, more accurate similarity calculations can be performed based on the three-dimensional structural information of the antigenic epitope amino acids, thereby improving the accuracy and reliability of antigenic epitope matching. The calculation of structural similarity considers not only the arrangement of the amino acid sequences but also their spatial structure and conformational information, making the similarity assessment more comprehensive and realistic. This method enhances the accuracy of the prediction results, ultimately improving the quality and reliability of antibody prediction, especially when dealing with antigenic epitopes with large structural differences, where it has significant advantages.

[0109] It is understood that the aforementioned structural similarity prediction tools are selected from open-source tools in this field, such as https: / / www.pnas.org / doi / 10.1073 / pnas.0906146106.

[0110] refer to Figure 4 In some implementations, the step of sorting the N candidate antibody amino acid sequences according to the similarity to determine the antigen-specific binding antibody includes: sorting the N candidate antibody amino acid sequences in descending order of similarity to obtain a first queue; screening target candidate antibody amino acid sequences from the first queue to obtain a second queue, wherein the target candidate antibody amino acid sequences contain interacting amino acids whose distance to the predicted antigen epitope amino acid sequence is less than a predetermined threshold, and the interacting amino acids include the heavy chain complementarity-determining region (CDR) amino acids of the target candidate antibody; and determining the antigen-specific binding antibody based on the second queue. Both the heavy chain and light chain CDRs of an antibody play important roles in antigen recognition and binding, but in many cases, the heavy chain CDR is considered to play a dominant role in antigen recognition. The method in this application innovatively considers the heavy chain CDR amino acids among the interacting amino acids, effectively increasing the accuracy of prediction and contributing to a better understanding of antibody-antigen interactions.

[0111] It should be noted that the phrase "determining antigen-specific binding antibodies according to the second queue" in this application can be understood as all candidate antibodies in the second queue being antigen-specific binding antibodies; or it can be understood as ranking the candidate antibodies in the second queue based on similarity and selecting the top N candidate antibodies as antigen-specific antibodies, where N is a positive integer, such as 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10, etc.

[0112] In some implementations, this application also includes similarity correction. For example, the correction method can be expressed as:

[0113]

[0114] Among them, Score final This represents the corrected similarity score; Score represents the similarity score; aa ab This indicates antibody amino acids that are less than a predetermined threshold compared to the predicted antigen epitope amino acid.

[0115] That is, if aa ab In the heavy chain complementarity-determining region amino acids of the target candidate antibody, the score... final For Score; if aa ab If the amino acid is not in the heavy chain complementarity-determining region of the target candidate antibody, then the score is... final It is 0.

[0116] The above correction methods can improve the accuracy and specificity of antibody prediction and help to better understand the interaction between antibodies and antigens.

[0117] In some implementations, the variable region of the heavy chain is the heavy chain third complementarity-determining region (HCDR3). The CDRs of the antibody heavy chain are considered to play a dominant role in antigen recognition, especially HCDR3. This is because HCDR3 exhibits the greatest diversity in length, sequence, and conformation, enabling antibodies to recognize and bind to a wide range of different antigens. Therefore, considering the amino acid information of the antibody heavy chain CDR region when predicting antigenic epitopes can improve prediction accuracy and contribute to a better understanding of antibody-antigen interactions.

[0118] For ease of understanding, please refer to Figure 5 The following is a detailed description of the steps for determining antigen-specific binding antibodies using a specific example:

[0119] Based on any of the above embodiments, actual and predicted antigenic epitopes are determined, and the similarity between them is calculated. N candidate antibodies are then sequentially arranged based on their similarity scores to obtain a first queue. The score of each candidate antibody in the first queue is its similarity score. The antibodies in the first queue are further screened based on the following conditions: the target antigen and the candidate antibody contain an amino acid (i.e., an interacting amino acid) below a predetermined threshold, and the interacting amino acid includes HCDR3 of the candidate antibody. If this condition is met, the antibody is retained and enters the second queue, where its score is the same as its original similarity score. If the condition is not met, the similarity score is 0, and the candidate antibody is discarded. The N candidate antibodies are screened separately to obtain the second queue. Based on the scores of each candidate antibody in the second queue, antigen-specific antibodies are determined.

[0120] The N candidate antibody amino acid sequences are sorted in descending order of similarity to obtain a first queue. Target candidate antibody amino acid sequences are then selected from the first queue to obtain a second queue. These target candidate antibody amino acid sequences contain interacting amino acids whose distance to the predicted antigen epitope amino acid sequence is less than a predetermined threshold. These interacting amino acids include the heavy chain complementarity-determining region (CDR) amino acids of the target candidate antibody. Based on the second queue, antigen-specific binding antibodies are determined. Both the heavy and light chain CDRs of antibodies play important roles in antigen recognition and binding, but in many cases, the heavy chain CDR is considered to play a dominant role in antigen recognition. The method in this application innovatively considers the heavy chain CDR amino acids among the interacting amino acids, effectively increasing the accuracy of prediction and contributing to a better understanding of antibody-antigen interactions.

[0121] In some implementations, the inventors of this application use the ANARCI tool and the IMGT rules to align the antibody sequence and find the CDR region of the antibody.

[0122] In some implementations, the distance is determined based on the coordinates of the α-carbon atom of each amino acid in the target candidate antibody and the predicted antigen epitope amino acid sequence. Interacting amino acids are determined based on this distance, and the target candidate antibody is further screened by checking whether the heavy chain complementarity-determining region amino acid of the target candidate antibody is present in the interacting amino acids, thereby improving the accuracy and specificity of antibody prediction.

[0123] For example, the distance referred to can be the Euclidean distance, specifically, the formula for calculating the Euclidean distance is as follows:

[0124]

[0125] Among them, (xn ,y n ,z n (x) represents the coordinate of the α-carbon in the nth amino acid of the target candidate antibody amino acid sequence. m ,y m ,z m The prediction threshold is determined based on the interaction between the target candidate antibody and the predicted antigen epitope amino acid sequence.

[0126] This application determines a distance threshold based on the interaction forces (such as van der Waals forces) between the target candidate antibody and the predicted antigenic epitope amino acid sequence. A distance threshold less than or equal to the threshold indicates an interaction between the target candidate antibody and the predicted antigenic epitope amino acid sequence, while a distance threshold greater than the threshold indicates no interaction. Based on the amino acids less than or equal to the distance threshold, the amino acids in the target candidate antibody that interact with the predicted antigenic epitope (i.e., interacting amino acids) are identified. These interacting amino acids can then be used to further refine the candidate antibody selection, thereby improving the specificity of the target candidate antibody screening.

[0127] It should be noted that the descriptions of the aforementioned interaction forces are the same as above, and will not be repeated here.

[0128] For ease of understanding, please refer to Figure 6 The following section details how the technical solution of this application is implemented, using SARS-CoV-2 functional antibody screening as an example:

[0129] Figure 6 (Left) shows the structure of the target antigen-ligand complex, where the blue protein is SARS-CoV-2, the green protein is ACE2, and the globular epitope is the antigenic amino acid defined in this application whose distance from the ligand is less than a predetermined threshold range, i.e., the actual antigenic epitope. Figure 6 (Right) shows the structure of the target antigen-target candidate antibody complex, where the blue protein is SARS-CoV-2, the red is the heavy chain of the candidate antibody, the yellow is the light chain of the candidate antibody, the blue globular epitopes are antigen amino acids that are less than a predetermined threshold range, and the red are antibody amino acids that are less than a predetermined threshold range from the antigen. The antibody amino acids include the complementarity-determining region amino acids of the heavy chain of the target candidate antibody.

[0130] The technical solutions described above can be applied to antibody drug development, screening antibodies that can bind to specific sites on antigens. In real-world scenarios, high-throughput sequencing technology can rapidly determine tens of thousands of antibody sequences. However, expressing, purifying, and determining the binding sites of all these antibodies would cost tens of millions of dollars and take several years, significantly hindering the progress of antibody drug development. Using machine learning or deep learning algorithms, which rely on labeled data for model training, is prone to overfitting and has weak generalization ability. Especially with high-complexity deep learning models, insufficient data makes it difficult to train efficient models. However, antibody screening using the method described in this application can achieve superior results compared to traditional molecular docking, molecular dynamics simulations, and deep learning even with zero-sample prediction. Furthermore, it improves the efficiency of antibody screening, shortens drug development time, and saves experimental costs.

[0131] Furthermore, the inventors used a test set to compare the predictive performance of the scheme in this application with existing antibody prediction schemes, and the results are shown in Table 1. The test set included 217 antibodies, 180 negative samples (candidate antibodies whose amino acids interacting with the antigen epitope do not contain the heavy chain complementarity-determining region amino acids of the candidate antibody), and 37 positive samples. The accuracy of the predictions was evaluated using AUC values. It can be seen that the antigen-specific antibody prediction based on the scheme in this application has significantly improved prediction accuracy.

[0132] Table 1

[0133]

[0134]

[0135] The methods provided in the embodiments of this application have been described above. The antigen-specific binding antibody prediction device provided in the embodiments of this application will be described below.

[0136] Figure 7 This is a schematic diagram of an antigen-specific binding antibody prediction device 700 according to one embodiment of this application, as shown below. Figure 7 As shown, the device 700 includes: a first determining module 710, a second determining module 720, and a third determining module 730, wherein,

[0137] The first determining module 710 is used to determine the actual antigenic epitope of the target antigen amino acid sequence; the second determining module 720 is used to determine N candidate antibody amino acid sequences, and determine the predicted antigenic epitope corresponding to the i-th candidate antibody amino acid sequence based on the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; wherein, N is an integer not less than 1, and i is a positive integer not greater than N; the third determining module 730 is used to calculate the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the i-th candidate antibody amino acid sequence, and sort the N candidate antibody amino acid sequences according to the similarity to determine the antigen-specific binding antibody.

[0138] In some possible implementations, determining the actual antigenic epitope of the target antigen amino acid sequence includes: inputting the target antigen amino acid sequence and the ligand amino acid sequence into a structure prediction tool to determine the target antigen-ligand complex structure; wherein the target antigen amino acid sequence and the ligand amino acid sequence bind at least partially specifically; and based on the target antigen-ligand complex structure, determining amino acids whose distance from the target antigen amino acid sequence to the ligand amino acid sequence is less than a predetermined threshold to determine the actual antigenic epitope of the target antigen amino acid sequence.

[0139] In some possible implementations, determining the actual antigenic epitope of the target antigen amino acid sequence includes: inputting the target antigen amino acid sequence into an antigenic epitope prediction tool to determine the actual antigenic epitope of the target antigen amino acid sequence.

[0140] In some implementations, determining the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody based on the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody includes: inputting the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody into a structure prediction tool to determine the structure of the target antigen-i-th candidate antibody complex; and based on the structure of the target antigen-i-th candidate antibody complex, determining amino acids whose distance between the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody is less than a predetermined threshold, and determining the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody.

[0141] In some possible implementations, the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated, including: calculating the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody based on the number of amino acid overlaps in the actual antigenic epitope and the number of amino acids in the actual antigenic epitope.

[0142] In some possible implementations, the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated, including: based on the ratio of the number of amino acid overlaps in the actual antigenic epitope and the predicted antigenic epitope to the number of amino acids in the actual antigenic epitope, the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated.

[0143] In some possible implementations, calculating the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody includes: inputting the amino acid sequences of the actual antigenic epitope and the predicted antigenic epitope into a structural similarity prediction tool, and calculating the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody.

[0144] In some implementations, the step of sorting the N candidate antibody amino acid sequences according to the similarity to determine the antigen-specific binding antibody includes: sorting the N candidate antibody amino acid sequences in descending order of similarity to obtain a first queue; screening target candidate antibody amino acid sequences from the first queue to obtain a second queue, wherein the target candidate antibody amino acid sequences contain interacting amino acids that are less than a predetermined threshold distance from the predicted antigen epitope amino acid sequence, and the interacting amino acids include the heavy chain complementarity-determining region amino acids of the target candidate antibody; and determining the antigen-specific binding antibody based on the second queue.

[0145] In some implementations, the heavy chain variable region is the heavy chain third complementarity determination region HCDR3.

[0146] In some implementations, the distance is determined by: determining the distance based on the coordinates of the α-carbon atom of each amino acid in the target antigen amino acid sequence and the ligand amino acid sequence; or determining the distance based on the coordinates of the α-carbon atom of each amino acid in the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; or determining the distance based on the coordinates of the α-carbon atom of each amino acid in the target candidate antibody and the predicted antigen epitope amino acid sequence.

[0147] In some implementations, the predetermined threshold is determined by: determining the predetermined threshold based on the interaction force between the target antigen amino acid sequence and the ligand amino acid sequence; or determining the predetermined threshold based on the interaction force between the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; or determining the predetermined threshold based on the interaction force between the target candidate antibody and the predicted antigen epitope amino acid sequence.

[0148] It should be understood that the device embodiments and method embodiments can correspond to each other, and similar descriptions can be found in the method embodiments. To avoid repetition, further details are omitted here. Specifically, Figure 7 The device 700 shown can perform Figure 2 The corresponding method embodiments, and the foregoing and other operations and / or functions of each module in the device 700 are respectively for implementing Figure 2 For the sake of brevity, the corresponding processes in each method are not described in detail here.

[0149] The apparatus 700 of this application embodiment has been described above from the perspective of functional modules in conjunction with the accompanying drawings. It should be understood that this functional module can be implemented in hardware, in software instructions, or in a combination of hardware and software modules. Specifically, the steps of the method embodiments in this application can be completed by integrated logic circuits in the processor's hardware and / or by software instructions. The steps of the method disclosed in this application embodiment can be directly embodied as being executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. Optionally, the software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps in the above method embodiments.

[0150] Figure 8 This is a schematic block diagram of an electronic device 800 provided in an embodiment of this application. The electronic device 800 may be the aforementioned training device or execution device, but is not limited thereto. Figure 8 As shown, the electronic device 800 may include:

[0151] The system includes a memory 810 and a processor 820. The memory 810 stores a computer program 830 and transfers the computer program 830 to the processor 820. In other words, the processor 820 can retrieve and run the computer program 830 from the memory 810 to implement the methods described in the embodiments of this application.

[0152] For example, the processor 820 can be used to execute the steps in the above method according to the instructions in the computer program 830.

[0153] In some embodiments of this application, the processor 820 may include, but is not limited to:

[0154] General-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0155] In some embodiments of this application, the memory 810 includes, but is not limited to:

[0156] Volatile memory and / or non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and Direct Rambus RAM (DR RAM).

[0157] In some embodiments of this application, the computer program 830 may be divided into one or more modules, which are stored in the memory 810 and executed by the processor 820 to perform the method provided in this application. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program 830 in the electronic device.

[0158] like Figure 8 As shown, the electronic device 800 may further include:

[0159] Transceiver 840, which can be connected to processor 820 or memory 810.

[0160] The processor 820 can control the transceiver 840 to communicate with other devices; specifically, it can send information or data to other devices or receive information or data sent by other devices. The transceiver 840 may include a transmitter and a receiver. The transceiver 840 may further include antennas, and the number of antennas may be one or more.

[0161] It should be understood that the various components in the electronic device 800 are connected through a bus system, which includes a data bus, a power bus, a control bus, and a status signal bus.

[0162] According to one aspect of this application, a computer-readable storage medium is provided that stores computer instructions or programs thereon, which, when executed by a computer, enable the computer to perform the methods of the above-described method embodiments. Alternatively, embodiments of this application also provide a computer program product containing instructions that, when executed by a computer, cause the computer to perform the methods of the above-described method embodiments.

[0163] According to another aspect of this application, a computer program product or computer program is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method described in the above-described method embodiments.

[0164] In other words, when implemented using software, it can be implemented wholly or partially in the form of a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., digital video disc (DVD)), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0165] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0166] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0167] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to implement the solution of this embodiment according to actual needs. For example, the functional modules in the various embodiments of this application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0168] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting antigen-specific binding antibodies, characterized in that, include: Identify the actual antigenic epitopes in the amino acid sequence of the target antigen; N candidate antibody amino acid sequences are determined. Based on the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence, the predicted antigen epitope corresponding to the i-th candidate antibody amino acid sequence is determined; where N is an integer not less than 1, and i is a positive integer not greater than N. The similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated respectively. The amino acid sequences of the N candidate antibodies are sorted according to the similarity to determine the antigen-specific binding antibody.

2. The method according to claim 1, characterized in that, The determination of the actual antigenic epitopes of the target antigen amino acid sequence includes: The target antigen amino acid sequence and the ligand amino acid sequence are input into a structure prediction tool to determine the structure of the target antigen-ligand complex; wherein the target antigen amino acid sequence and the ligand amino acid sequence bind at least partially specifically. Based on the structure of the target antigen-ligand complex, amino acids whose distance from the target antigen amino acid sequence to the ligand amino acid sequence is less than a predetermined threshold are identified, and the actual antigenic epitopes of the target antigen amino acid sequence are determined.

3. The method according to claim 1, characterized in that, The determination of the actual antigenic epitopes of the target antigen amino acid sequence includes: The target antigen amino acid sequence is input into the antigen epitope prediction tool to determine the actual antigen epitopes of the target antigen amino acid sequence.

4. The method according to claim 1, characterized in that, The step of determining the predicted antigen epitope corresponding to the amino acid sequence of the i-th candidate antibody based on the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody includes: The amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody are input into a structure prediction tool to determine the structure of the target antigen-i-th candidate antibody complex. Based on the structure of the target antigen-ith candidate antibody complex, determine the amino acids whose distance between the amino acid sequence of the target antigen and the amino acid sequence of the ith candidate antibody is less than a predetermined threshold, and determine the predicted antigen epitope corresponding to the amino acid sequence of the ith candidate antibody.

5. The method according to any one of claims 1-4, characterized in that, Calculate the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody, including: Based on the number of amino acid overlaps in the actual antigenic epitope and the predicted antigenic epitope, and the number of amino acids in the actual antigenic epitope, the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated.

6. The method according to claim 5, characterized in that, Calculate the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody, including: Based on the actual antigenic epitope and the ratio of the number of amino acid overlaps in the predicted antigenic epitope to the number of amino acids in the actual antigenic epitope, the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody is calculated.

7. The method according to any one of claims 1-4, characterized in that, Calculate the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody, including: The amino acid sequences of the actual antigenic epitope and the predicted antigenic epitope are input into a structural similarity prediction tool to calculate the similarity between the actual antigenic epitope and the predicted antigenic epitope corresponding to the amino acid sequence of the i-th candidate antibody.

8. The method according to claim 1, characterized in that, The step of sorting the amino acid sequences of the N candidate antibodies according to the similarity to determine the antigen-specific binding antibody includes: The N candidate antibody amino acid sequences are sorted according to their similarity from high to low to obtain the first queue; The amino acid sequences of target candidate antibodies are screened from the first queue to obtain the second queue. The amino acid sequences of the target candidate antibodies contain interacting amino acids that are less than a predetermined threshold distance from the amino acid sequence of the predicted antigen epitope. The interacting amino acids include the amino acids in the heavy chain complementarity-determining region of the target candidate antibody. Based on the second cohort, antigen-specific binding antibodies are identified.

9. The method according to claim 8, characterized in that, The variable region of the heavy chain is the third complementarity determination region HCDR3 of the heavy chain.

10. The method according to claim 2, 4, or 8, characterized in that, The distance is determined in the following way: The distance is determined based on the coordinates of the α-carbon atom of each amino acid in the target antigen amino acid sequence and the ligand amino acid sequence; or The distance is determined based on the coordinates of the α-carbon atom of each amino acid in the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody; or The distance is determined based on the coordinates of the α-carbon atom of each amino acid in the target candidate antibody and the predicted antigen epitope amino acid sequence.

11. The method according to claim 2, 4, or 8, characterized in that, The predetermined threshold is determined in the following manner: The predetermined threshold is determined based on the interaction force between the target antigen amino acid sequence and the ligand amino acid sequence; or The predetermined threshold is determined based on the interaction force between the amino acid sequence of the target antigen and the amino acid sequence of the i-th candidate antibody; or The predetermined threshold is determined based on the interaction between the target candidate antibody and the predicted antigen epitope amino acid sequence.

12. An antigen-specific binding antibody prediction device, characterized in that, include: The first determining module is used to determine the actual antigenic epitopes of the target antigen's amino acid sequence; The second determining module is used to determine N candidate antibody amino acid sequences, and to determine the predicted antigen epitope corresponding to the i-th candidate antibody amino acid sequence based on the target antigen amino acid sequence and the i-th candidate antibody amino acid sequence; where N is an integer not less than 1, and i is a positive integer not greater than N; The third determining module is used to calculate the similarity between the actual antigen epitope and the predicted antigen epitope corresponding to the amino acid sequence of the i-th candidate antibody, and to sort the N candidate antibody amino acid sequences according to the similarity to determine the antigen-specific binding antibody.

13. An electronic device, characterized in that, include: Processor and memory; The memory is used to store computer programs; The processor is configured to execute the computer program to implement the antigen-specific binding antibody prediction method as described in any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions or programs that, when executed on a computer, cause the antigen-specific binding antibody prediction method as described in any one of claims 1 to 11 to be performed.

15. A computer program product, characterized in that, The computer program product includes computer instructions that, when some or all of the computer instructions are run on a computer, cause the antigen-specific binding antibody prediction method as described in any one of claims 1 to 11 to be executed.

Images