Methods, apparatuses, and devices for predicting antibodies and antigen binding epitopes and storage media

Abstract

The application relates to a method, device and equipment for predicting an antibody and an antigen binding epitope and a storage medium. The method comprises the following steps: obtaining a structure of a target antigen and a structure of a target antibody; determining surface residues on the structure of the target antigen; cutting the structure of the target antigen into a plurality of antigen fragments based on the surface residues; predicting the affinity between the target antibody and each antigen fragment according to the structure of the antigen fragment and the structure of the target antibody, and obtaining affinity data; determining an antigen fragment bound by the target antibody according to the affinity data, and taking the antigen fragment as a binding epitope of the target antibody and the target antigen. The scheme provided in the application can combine antigen cutting technology and affinity prediction technology, efficiently and accurately predict the binding epitope of the antibody and the antigen, and provide strong support for in-depth research on the antibody.

Description

Technical Field

[0001] This application relates to the field of computational structural biology, and in particular to methods, apparatus and devices for predicting antibody and antigen-binding epitopes, as well as storage media. Background Technology

[0002] In the field of antibody drug development, rapidly and efficiently identifying the binding epitopes of antibodies and antigens is a key technical challenge. Accurate epitope information is of great guiding significance for in-depth research on antibody mechanisms of action, optimization of antibody performance, and assessment of antibody development potential.

[0003] Among related technologies, commonly used experimental methods such as enzyme-linked immunosorbent assay (ELISA), surface plasmon resonance (SPR), hydrogen / deuterium exchange (HDX), and cryo-electron microscopy (cryo-EM), although capable of measuring antibody-antigen binding, have the following significant limitations:

[0004] (1) Long experimental cycle and high cost

[0005] The above-mentioned experimental methods typically require complex pretreatment steps such as sample preparation and probe labeling, resulting in lengthy experimental procedures that can take weeks or even months from construction to detection. Furthermore, the experimental costs are quite high due to the consumption of large amounts of reagents and reliance on specialized instruments.

[0006] (2) Limited ability to locate positions

[0007] The experimental methods described above mainly focus on the binding of antibodies to the entire antigen, making it difficult to accurately locate the binding epitope region of the antibody on the antigen. This poses a challenge to in-depth research on the mechanism of antibody action.

[0008] (3) Limited high-throughput screening capability

[0009] In the early stages of antibody discovery, it is often necessary to identify promising molecules from a large library of candidate antibodies, but traditional experimental methods are inefficient and cannot meet this need.

[0010] Therefore, developing new methods for epitope prediction based on computer simulation to compensate for the shortcomings of experimental methods has important theoretical significance and application value. Summary of the Invention

[0011] To address or partially address the problems existing in related technologies, this application provides a method, apparatus, device, and storage medium for predicting antibody-antigen binding epitopes. This method combines antigen cleavage technology and affinity prediction technology to efficiently and accurately predict antibody-antigen binding epitopes, providing strong support for in-depth antibody research.

[0012] The first aspect of this application provides a method for predicting antibody-antigen binding epitopes, comprising:

[0013] Obtain the structures of the target antigen and target antibody;

[0014] Identify the surface residues on the structure of the target antigen;

[0015] Based on the surface residues, the structure of the target antigen is divided into multiple antigen fragments;

[0016] Based on the structure of the antigen fragment and the structure of the target antibody, the affinity of the target antibody to each antigen fragment is predicted to obtain affinity data;

[0017] Based on the affinity data, the antigen fragment to which the target antibody binds is determined, and this antigen fragment is used as the binding epitope of the target antibody and the target antigen.

[0018] As an optional embodiment, determining the surface residues on the structure of the target antigen includes:

[0019] Calculate the solvent-accessible surface area of ​​each residue in the structure of the target antigen;

[0020] Based on the comparison between the solvent-accessible surface area of ​​the residue and a threshold, it is determined whether the residue is a surface residue.

[0021] As an optional embodiment, the step of slicing the target antigen into multiple antigen fragments based on the surface residues includes:

[0022] Randomly select a surface residue as a cutting point, and place the cutting point in the first position of the cutting order;

[0023] Outside the preset cutting range of the m-th cutting point, the surface residue closest to the m-th cutting point is calculated and used as the cutting point, and the cutting order of this cutting point is arranged at the (m+1)-th position, where m≥1;

[0024] Using each of the aforementioned slicing points as the center, and following the preset slicing range, the slicing of all slicing points is completed sequentially starting from the first slicing point, generating multiple antigen fragments.

[0025] As an optional embodiment, after dividing the target antigen into multiple antigen fragments based on the surface residues, the method further includes:

[0026] Calculate the similarity between at least two adjacent antigen fragments;

[0027] If the similarity reaches a preset value, the antigen fragment farthest from the first segmentation point is taken as a redundant antigen fragment and removed.

[0028] As an optional embodiment, calculating the similarity of at least two adjacent antigen fragments includes:

[0029] The total number is the number of all surface residues contained in the antigen fragment closest to the first cleavage point among at least two adjacent antigen fragments;

[0030] The ratio of the number of identical surface residues in at least two adjacent antigen fragments to the total number of such residues is calculated, and this ratio is used as the similarity score.

[0031] As an optional embodiment, the step of predicting the affinity of the target antibody for each antigen fragment based on the structure of the antigen fragment and the structure of the target antibody, and obtaining affinity data, includes:

[0032] The structure of the antigen fragment and the target antibody are expressed using graph convolution to generate the structural features of the antigen fragment and the target antibody;

[0033] The structural features of the antigen fragment and the target antibody are fused using an attention mechanism to generate a fusion feature;

[0034] Based on the fusion characteristics, the binding probability of the antigen fragment and the target antibody is mapped.

[0035] As an optional embodiment, the step of determining the antigen fragment to which the target antibody binds based on the affinity data, and using the antigen fragment as the binding epitope of the target antibody and the target antigen, includes:

[0036] The antigen fragment whose binding probability satisfies a preset probability is selected as a candidate antigen fragment;

[0037] The structures of spatially adjacent antigen fragments in the candidate antigen fragments are fused, and the fused candidate antigen fragments are used as the binding epitopes of the target antibody and the target antigen.

[0038] A second aspect of this application provides an antibody and antigen binding epitope prediction device, comprising:

[0039] The acquisition module is used to acquire the structures of the target antigen and the target antibody;

[0040] A first determining module is used to determine surface residues on the structure of the target antigen;

[0041] A segmentation module is used to segment the structure of the target antigen into multiple antigen fragments based on the surface residues;

[0042] An affinity prediction module is used to predict the affinity between the target antibody and each of the antigen fragments based on the structure of the antigen fragments and the structure of the target antibody, thereby obtaining affinity data.

[0043] The second determining module is used to determine the antigen fragment that the target antibody binds to based on the affinity data, and to use the antigen fragment as the binding epitope of the target antibody and the target antigen.

[0044] A third aspect of this application provides an electronic device, comprising:

[0045] Processor; and

[0046] A memory that stores executable code, which, when executed by the processor, causes the processor to perform the method described above.

[0047] A fourth aspect of this application provides a computer-readable storage medium having executable code stored thereon, which, when executed by a processor of an electronic device, causes the processor to perform the method described above.

[0048] The technical solution provided in this application may include the following beneficial effects:

[0049] This application, based on the surface residues of the target antigen structure, divides the target antigen structure into multiple antigen fragments. This residue-level division is equivalent to dividing the target antigen into multiple epitope regions. Then, the affinity of the divided antigen fragments with the target antigen is predicted to determine the antigen fragment that the target antibody binds to, which is equivalent to determining the epitope region that the target antibody binds to. Therefore, this application performs calculations and predictions at the residue level, integrating antigen and antibody structural information, enabling a more comprehensive characterization of antibody-antigen interactions and thus improving the accuracy of epitope prediction. Furthermore, by combining antigen division technology and affinity prediction technology, it can efficiently and accurately predict the antibody-antigen binding epitopes, providing strong support for in-depth antibody research.

[0050] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0051] The above and other objects, features and advantages of this application will become more apparent from the more detailed description of exemplary embodiments thereof in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments thereof.

[0052] Figure 1 This is a flowchart illustrating the method for predicting antibody-antigen binding epitopes as shown in the embodiments of this application;

[0053] Figure 2 This is another flowchart illustrating the method for predicting antibody-antigen binding epitopes as shown in the embodiments of this application;

[0054] Figure 3 This is a schematic diagram of the structure of the antibody and antigen binding epitope prediction device shown in the embodiments of this application;

[0055] Figure 4 This is a schematic diagram of the structure of an electronic device shown in an embodiment of this application. Detailed Implementation

[0056] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to make this application more thorough and complete, and to fully convey the scope of this application to those skilled in the art.

[0057] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0058] It should be understood that although the terms "first," "second," "third," etc., may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0059] Among related technologies, commonly used experimental methods such as enzyme-linked immunosorbent assay (ELISA), surface plasmon resonance (SPR), hydrogen / deuterium exchange (HDX), and cryo-electron microscopy (Cryo-EM) can measure the binding of antibodies and antigens, but they have several obvious drawbacks: long experimental cycles and high costs; limited ability to locate epitopes; and limited high-throughput screening capabilities.

[0060] To address the aforementioned issues, this application provides a method for predicting antibody-antigen binding epitopes. This method combines antigen fragmentation and affinity prediction techniques to efficiently and accurately predict antibody-antigen binding epitopes, providing strong support for in-depth antibody research.

[0061] The technical solutions of the embodiments of this application are described in detail below with reference to the accompanying drawings.

[0062] Figure 1 This is a flowchart illustrating the method for predicting antibody-antigen binding epitopes as shown in the embodiments of this application.

[0063] See Figure 1 This application provides a method for predicting antibody-antigen binding epitopes, including steps S1 to S5:

[0064] S1: Obtain the structures of the target antigen and target antibody.

[0065] The structure of the target antigen / antibody can be obtained in the following way: the structure of the target antigen / antibody and its receptor complex is obtained from the PDB database, and after repairing the structure of the complex, the structure of the target antigen / antibody is isolated from it.

[0066] The target antigen and target antibody in the embodiments of this application can be proteins, and the structures of the target antigen and target antibody can be protein structures predicted by experimental means (such as cryo-electron microscopy, X-ray diffraction) or computational methods (such as AlphaFold). It should also be noted that the structure of the target antigen can also be modeled using related modeling methods, and this application does not limit this.

[0067] In this application, the target antigen and target antibody can bind, and the specific binding epitopes of the target antigen and target antibody need to be predicted using the prediction method in the embodiments of this application.

[0068] S2: Identify the surface residues on the structure of the target antigen.

[0069] The embodiments of this application can determine the surface residues on the structure of the target antigen by: calculating the solvent-accessible surface area of ​​each residue on the structure of the target antigen; and determining whether the residue is a surface residue based on the comparison result of the solvent-accessible surface area of ​​the residue with a threshold.

[0070] The embodiments of this application can use bioinformatics tools (such as biopython and pymol) to calculate the solution-accessible surface area of ​​each residue on the structure of the target antigen.

[0071] S3: Based on surface residues, the structure of the target antigen is divided into multiple antigen fragments.

[0072] The embodiments of this application can segment the target antigen in the following manner: randomly select a surface residue as a segmentation point and place the segmentation order of the segmentation point at position 1; outside the preset segmentation range of the m-th segmentation point, calculate the surface residue closest to the m-th segmentation point and place it as a segmentation point, and place the segmentation order of this segmentation point at position m+1, where m≥1; with each segmentation point as the center, according to the preset segmentation range, start from the 1st segmentation point and sequentially complete the segmentation of all segmentation points, generating multiple antigen fragments.

[0073] S4: Based on the structure of the antigen fragment and the structure of the target antibody, predict the affinity of the target antibody to each antigen fragment and obtain affinity data.

[0074] The embodiments of this application can predict affinity in the following way: express the structure of the antigen fragment and the target antibody using graph convolution to generate structural features of the antigen fragment and the target antibody; fuse the structural features of the antigen fragment and the target antibody through an attention mechanism to generate fusion features; and map the binding probability of the antigen fragment and the target antibody based on the fusion features.

[0075] S5: Based on affinity data, determine the antigen fragment that the target antibody binds to, and use this antigen fragment as the binding epitope between the target antibody and the target antigen.

[0076] The binding epitopes of the target antibody and the target antigen can be determined in the following ways: antigen fragments with binding probabilities that meet a preset probability are selected as candidate antigen fragments; the structures of spatially adjacent antigen fragments in the candidate antigen fragments are fused, and the fused candidate antigen fragments are selected as the binding epitopes of the target antibody and the target antigen.

[0077] This application's embodiments are based on the surface residues on the target antigen structure, dividing the target antigen structure into multiple antigen fragments. This residue-level segmentation is equivalent to dividing the target antigen into multiple epitope regions. Then, the affinity of the segmented antigen fragments with the target antigen is predicted to determine the antigen fragment that the target antibody binds to, which is equivalent to determining the epitope region that the target antibody binds to. Therefore, this application's embodiments perform calculations and predictions at the residue level, integrating antigen and antibody structure information, which can more comprehensively characterize antibody-antigen interactions, thereby improving the accuracy of epitope prediction. Furthermore, by combining antigen segmentation technology and affinity prediction technology, it can efficiently and accurately predict the antibody-antigen binding epitopes, providing strong support for in-depth antibody research.

[0078] Figure 2 This is another flowchart illustrating the method for predicting antibody and antigen binding epitopes as shown in the embodiments of this application.

[0079] See Figure 2This application provides a method for predicting antibody-antigen binding epitopes, comprising the following steps:

[0080] S10: Obtain the structures of the target antigen and target antibody.

[0081] The structure of the target antigen / antibody can be obtained in the following way: the structure of the target antigen / antibody and its receptor complex is obtained from the PDB database, and after repairing the structure of the complex, the structure of the target antigen / antibody is isolated from it.

[0082] The target antigen and target antibody in the embodiments of this application can be proteins, and the structures of the target antigen and target antibody can be protein structures predicted by experimental means (such as cryo-electron microscopy, X-ray diffraction) or computational methods (such as AlphaFold). It should also be noted that the structure of the target antigen can also be modeled using related modeling methods, and this application does not limit this.

[0083] In this application, the target antigen and target antibody can bind, and the specific binding epitopes of the target antigen and target antibody need to be predicted using the prediction method in the embodiments of this application.

[0084] S20: Calculate the solvent-accessible surface area of ​​each residue in the structure of the target antigen.

[0085] Calculate the solvent-accessible surface area (SASA) of each amino acid residue in the structure of the target antigen, and compare the solvent-accessible surface area with the threshold.

[0086] The embodiments of this application can use bioinformatics tools (such as biopython and pymol) to calculate the solution accessible surface area (SASA) of each residue on the structure of the target antigen.

[0087] S21: Determine whether a residue is a surface residue based on the comparison between the solvent-accessible surface area of ​​the residue and the threshold.

[0088] In this application, the solvent-accessible surface area (SASA) of a residue can be compared with a threshold. If the solvent-accessible surface area (SASA) of a residue is greater than the threshold, then the residue is determined to be a surface residue.

[0089] Among them, the side chain groups of residues with a solvent-accessible surface area (SASA) greater than the threshold extend toward the outside of the target antigen structure, occupy a certain surface space of the target antigen structure, and can directly contact the external environmental solution. These residues are surface residues and have the potential to interact with the target antibody.

[0090] Furthermore, the threshold value in this embodiment can be set according to actual needs, for example, it can be... This application does not impose any limitations on this.

[0091] S30: Randomly select a surface residue as the cutting point and place the cutting point in the first position.

[0092] Since the target antigen binds to different antibody epitopes, when slicing the target antigen, a surface residue can be randomly selected as the slicing start point, and the slicing start point is ranked first in the slicing order.

[0093] S31: Outside the preset cutting range of the m-th cutting point, calculate the surface residue closest to the m-th cutting point and use it as the cutting point, and place the cutting order of this cutting point at the (m+1)-th position, where m≥1.

[0094] In this embodiment, step S31 starts from m=1. Outside the preset segmentation range of the first segmentation point, the surface residue closest to the first segmentation point is calculated and used as the segmentation point, and this segmentation point is ranked second in the segmentation order. Then, outside the preset segmentation range of the second segmentation point, the surface residue closest to the second segmentation point is calculated and used as the segmentation point, and this segmentation point is ranked third in the segmentation order. This process is repeated to calculate all segmentation points and determine the order of all segmentation points. By sorting the segmentation, the efficiency of segmentation can be improved. Furthermore, the segmentation method of this embodiment can avoid antigen fragment redundancy in advance. In addition, segmenting in sequence makes it faster and more efficient to remove redundancy from adjacent antigen fragments later.

[0095] In this embodiment, the distance between surface residues can be the Euclidean distance between the CA atoms of two surface residues.

[0096] In addition, when there are multiple surface residues that are closest to the m-th cut point, one of these surface residues can be randomly selected as the cut point, and the remaining surface residues are not sorted.

[0097] S32: Using each segmentation point as the center, according to the preset segmentation range, start from the first segmentation point and sequentially complete the segmentation of all segmentation points, generating multiple antigen fragments.

[0098] In this embodiment, the segmentation can be centered on the segmentation point, dividing the structure into a cylinder. The preset segmentation range can be: using the line connecting the geometric center of the target antigen structure and the segmentation point as the central axis of the cylinder, and setting the rotation radius of the cylinder as follows: The height of the cylinder can be set to...

[0099] S33: Calculate the similarity between at least two neighboring antigen fragments.

[0100] The similarity in this application embodiment can be calculated in the following ways:

[0101] The total number of all surface residues contained in the antigen fragment closest to the first cleavage point among at least two neighboring antigen fragments is taken as the total number; the ratio of the number of identical surface residues contained in at least two neighboring antigen fragments to the total number is calculated, and this ratio is taken as the similarity.

[0102] Since neighboring antigen fragments have similar structures and functions and contain highly similar types of surface residues, redundancy can be removed from antigen fragments in order to reduce the computational load for affinity prediction.

[0103] S34: If the similarity reaches the preset value, the antigen fragment farthest from the first segmentation point is taken as a redundant antigen fragment and removed.

[0104] Since the antigen fragments obtained by cutting at the first cleavage point will definitely be retained, this embodiment of the application removes the antigen fragments farthest from the first cleavage point and retains the fragments close to the first cleavage point, so that the retained antigen fragments are relatively concentrated, which facilitates subsequent fusion processing.

[0105] The preset values ​​in the embodiments of this application can be empirical values, such as 50% and 60%, and this application does not limit them.

[0106] S40: Graph convolution is used to express the structure of antigen fragments and target antibodies, generating structural features of antigen fragments and target antibodies.

[0107] This application embodiment can utilize an affinity prediction model, such as the PECAN architecture, to predict the affinity between antigen fragments and target antibodies. The structure of each antigen fragment and the structure of the target antibody are used as features. All features formed by antigen fragments and target antibodies are input into the affinity prediction model. The graph convolution module expresses the structure of the antigen fragment and the structure of the target antibody, generating structural features of the antigen fragment and the target antibody.

[0108] S41: The structural features of the antigen fragment and the target antibody are fused through an attention mechanism to generate a fusion feature.

[0109] After the graph convolution module expresses the structure features of the antigen fragment and the target antibody, the attention mechanism module inputs them into the attention mechanism module. The attention mechanism module fuses the structure features of the antigen fragment and the target antibody to generate fused features.

[0110] S42: Based on fusion features, the binding probability of antigen fragments and target antibodies is mapped.

[0111] After fusion, the affinity prediction model maps the binding probability of the antigen fragment and the target antibody based on the fusion characteristics. The binding probability can be a binding score, for example, 0 to 1, with higher scores indicating higher affinity.

[0112] S50: Select antigen fragments whose binding probability satisfies the preset probability as candidate antigen fragments.

[0113] When the binding probability reaches the preset probability, that is, when the binding score reaches the preset score, it indicates that the antigen fragment has a high affinity for the antibody, a strong binding force, and a high probability that the antibody will bind to the antigen fragment.

[0114] S51: The structures of spatially adjacent antigen fragments in the candidate antigen fragments are fused, and the fused candidate antigen fragments are used as binding epitopes for the target antibody and the target antigen.

[0115] Since there may be spatially adjacent antigen fragments in the candidate antigen fragments, in order to reduce the amount of prediction data, this application embodiment fuses the structures of spatially adjacent antigen fragments in the candidate antigen fragments. This can be done by supplementing the fragment structures cut off between spatially adjacent antigen fragments in the candidate antigen fragments, fusing them to form a more complete binding epitope, preserving the integrity of the target antigen binding epitope, and thus preserving the integrity of the binding epitope function.

[0116] This application first obtains the structure of the target antigen and the structure of the target antibody, then determines the surface residues on the target antigen structure. This can be done by calculating the solvent-accessible surface area (SASA) of each residue on the target antigen structure. Based on the comparison between the SASA and a threshold, it is determined whether a residue is a surface residue. Since the side chain groups of residues with a SASA greater than the threshold extend outwards from the target antigen structure, occupying a certain surface space of the target antigen structure and being able to directly contact the external environmental solution, the calculated surface residues have the potential to interact with the target antibody. Then, based on the surface residues, the target antigen structure is divided into multiple antigen fragments. A surface residue can be randomly selected as a cleavage point, and the cleavage point is ranked first in the cleavage order. Outside the preset cleavage range of the m-th cleavage point, the surface residue closest to the m-th cleavage point is calculated and selected as a cleavage point, and the cleavage order of this cleavage point is ranked (m+1), where m≥1. With each cleavage point as the center, according to the preset cleavage range, the cleavage of all cleavage points is completed sequentially starting from the first cleavage point, generating multiple antigen fragments. By sorting and segmenting, the efficiency of segmentation can be improved. Furthermore, the segmentation method in this embodiment can avoid antigen fragment redundancy beforehand. Sequential segmentation also allows for faster and more efficient removal of redundancy from neighboring antigen fragments. After segmenting the antigen fragments, the similarity between at least two neighboring antigen fragments can be calculated. If the similarity reaches a preset value, the antigen fragment furthest from the first segmentation point is considered redundant and removed. The similarity calculation can use the total number of surface residues in the antigen fragment closest to the first segmentation point among the at least two neighboring antigen fragments as the total number; the ratio of identical surface residues in the at least two neighboring antigen fragments to the total number is calculated and used as the similarity. This embodiment removes redundant antigen fragments through redundancy calculation, significantly reducing the computational load for subsequent affinity prediction. Then, graph convolution is used to express the structure of the antigen fragment and the target antibody, generating structural features of the antigen fragment and the target antibody. An attention mechanism is used to fuse the structural features of the antigen fragment and the target antibody, generating fusion features. Based on the fusion features, the binding probability of the antigen fragment and the target antibody is mapped. This application embodiment uses an affinity prediction model to predict the binding probability of an antigen fragment (epitope region) to a target antibody, thereby screening for binding epitopes. Finally, antigen fragments with binding probabilities satisfying a preset probability are selected as candidate antigen fragments. Spatially adjacent antigen fragments within these candidate fragments are fused, and the fused candidate antigen fragment is used as the binding epitope for both the target antibody and the target antigen. This application embodiment obtains a more complete binding epitope by fusing spatially adjacent candidate antigen fragments, preserving the integrity of the target antigen binding epitope and thus preserving its functional integrity.

[0117] Because the embodiments of this application perform residue-level cleavage of the target antigen structure, which is equivalent to dividing the target antigen into multiple epitope regions, and then predict the affinity between the cleaved antigen fragments and the target antigen to determine the antigen fragments that the target antibody binds to, and fuse the bindable antigen fragments, the integrity of the structure and function of the epitope regions where the target antibody and the target antigen bind is ensured. Therefore, the embodiments of this application have the following advantages:

[0118] (1) Improved the accuracy of epitope prediction

[0119] The embodiments of this application integrate the structural information of antibodies and antigens, which can more comprehensively characterize antibody-antigen interactions, thereby improving the accuracy of epitope prediction.

[0120] (2) Significantly improve forecasting efficiency

[0121] Compared to experimental methods, the prediction method in this application embodiment can efficiently complete large-scale epitope prediction in a very short time, greatly saving time and resources.

[0122] (3) High-throughput screening that supports antibody design

[0123] Efficient epitope prediction enables rapid screening of a large number of candidate antibodies, providing strong support for antibody design.

[0124] (4) Reduce the experimental costs of antibody development

[0125] By using computational prediction, the workload of experimental screening can be significantly reduced, thereby lowering the overall experimental cost of antibody development.

[0126] (5) Efficient epitope prediction was achieved.

[0127] Traditional experimental methods for determining antibody-binding epitopes are complex and time-consuming. The epitope prediction method based on a computational model in this application can quickly and efficiently determine antibody-binding epitopes, providing strong support for in-depth antibody research.

[0128] Corresponding to the aforementioned application function implementation method embodiments, this application also provides an antibody and antigen binding epitope prediction device, an electronic device, and corresponding embodiments.

[0129] Figure 3 This is a schematic diagram of the structure of the antibody and antigen binding epitope prediction device shown in the embodiments of this application.

[0130] See Figure 3 This application also provides an antibody and antigen binding epitope prediction device, including an acquisition module 300, a first determination module 310, a segmentation module 320, an affinity prediction module 330, and a second determination module 340.

[0131] The acquisition module 300 is used to acquire the structure of the target antigen and the target antibody.

[0132] The first determining module 310 is used to determine surface residues on the structure of the target antigen. Specifically, the first determining module 310 can be used to calculate the solvent-accessible surface area of ​​each residue on the structure of the target antigen; based on the comparison result of the solvent-accessible surface area of ​​the residue with a threshold, it determines whether the residue is a surface residue. Embodiments of this application can utilize bioinformatics tools (such as BioPython and PyMOL) to calculate the solution-accessible surface area of ​​each residue on the structure of the target antigen.

[0133] The segmentation module 320 is used to segment the structure of the target antigen into multiple antigen fragments based on surface residues. Specifically, the segmentation module 320 can randomly select a surface residue as a segmentation point and place it in the first segmentation order; outside the preset segmentation range of the m-th segmentation point, calculate the surface residue closest to the m-th segmentation point and select it as a segmentation point, placing it in the (m+1)-th segmentation order, where m ≥ 1; and, using each segmentation point as the center, sequentially segment all segments according to the preset segmentation range, starting from the first segmentation point, generating multiple antigen fragments.

[0134] In this embodiment of the application, the segmentation module 320 can also be used to calculate the similarity of at least two neighboring antigen fragments; if the similarity reaches a preset value, the antigen fragment farthest from the first segmentation point is taken as a redundant antigen fragment and the redundant antigen fragment is removed.

[0135] In this embodiment of the application, the segmentation module 320 can also be used to take the total number of all surface residues contained in the antigen fragment closest to the first segmentation point among the at least two neighboring antigen fragments as the total number; calculate the ratio of the number of identical surface residues contained in the at least two neighboring antigen fragments to the total number, and use the ratio as the similarity.

[0136] The affinity prediction module 330 is used to predict the affinity between the target antibody and each antigen fragment based on the structure of the antigen fragment and the structure of the target antibody, thereby obtaining affinity data. Specifically, the affinity prediction module 330 can use graph convolution to express the structure of the antigen fragment and the target antibody, generating structural features of the antigen fragment and the target antibody; it can also fuse the structural features of the antigen fragment and the target antibody through an attention mechanism to generate fusion features; and based on the fusion features, it can map the binding probability of the antigen fragment and the target antibody.

[0137] The second determining module 340 is used to determine the antigen fragment that the target antibody binds to based on affinity data, and to use the antigen fragment as the binding epitope of the target antibody and the target antigen. Specifically, the second determining module 340 can be used to select antigen fragments whose binding probability meets a preset probability as candidate antigen fragments; to fuse the structures of spatially adjacent antigen fragments among the candidate antigen fragments, and to use the fused candidate antigen fragment as the binding epitope of the target antibody and the target antigen.

[0138] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated further here.

[0139] Figure 4 This is a schematic diagram of the structure of an electronic device shown in an embodiment of this application.

[0140] See Figure 4 The electronic device 400 includes a memory 410 and a processor 420.

[0141] The processor 420 can be a Central Processing Unit (CPU), or other 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. The general-purpose processor can be a microprocessor or any conventional processor.

[0142] Memory 410 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. ROM may store static data or instructions required by processor 420 or other modules of the computer. Permanent storage devices may be read-write storage devices. Permanent storage devices may be non-volatile storage devices that retain stored instructions and data even when the computer is powered off. In some embodiments, permanent storage devices use mass storage devices (e.g., magnetic or optical disks, flash memory) as permanent storage devices. In other embodiments, permanent storage devices may be removable storage devices (e.g., floppy disks, optical drives). System memory may be a read-write storage device or a volatile read-write storage device, such as dynamic random access memory. System memory may store some or all of the instructions and data required by the processor during operation. Furthermore, memory 410 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (e.g., DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, memory 410 may include a removable storage device that is readable and / or writable, such as a laser disc (CD), a read-only digital multifunction optical disc (e.g., DVD-ROM, dual-layer DVD-ROM), a read-only Blu-ray disc, a high-density optical disc, a flash memory card (e.g., SD card, mini SD card, Micro-SD card, etc.), a magnetic floppy disk, etc. Computer-readable storage media do not contain carrier waves or transient electronic signals transmitted wirelessly or via wired connections.

[0143] The memory 410 stores executable code, which, when processed by the processor 420, can cause the processor 420 to execute part or all of the methods described above.

[0144] Furthermore, the method according to this application can also be implemented as a computer program or computer program product, which includes computer program code instructions for performing some or all of the steps in the method described above.

[0145] Alternatively, this application may be implemented as a computer-readable storage medium (or a non-transitory machine-readable storage medium or a machine-readable storage medium) storing executable code (or computer program or computer instruction code) thereon, which, when executed by a processor of an electronic device (or server, etc.), causes the processor to perform part or all of the steps of the methods described above according to this application.

[0146] The various embodiments of this application have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for predicting antibody-antigen binding epitopes, characterized in that, include: Obtain the structures of the target antigen and target antibody; Identify the surface residues on the structure of the target antigen; Based on the surface residues, the structure of the target antigen is divided into multiple antigen fragments; It includes: randomly selecting a surface residue as a slicing point and arranging the slicing order of the slicing point as the first position; calculating the surface residue closest to the m-th slicing point outside the preset slicing range of the m-th slicing point and arranging it as the slicing point, and arranging the slicing order of the m+1-th slicing point, where m≥1; slicing all slicing points sequentially from the first slicing point according to the preset slicing range, with each slicing point as the center, and generating multiple antigen fragments; calculating the similarity between at least two neighboring antigen fragments; if the similarity reaches a preset value, then the antigen fragment farthest from the first slicing point is taken as a redundant antigen fragment and the redundant antigen fragment is removed; Based on the structure of the antigen fragment and the structure of the target antibody, the affinity of the target antibody to each antigen fragment is predicted to obtain affinity data; Based on the affinity data, the antigen fragment to which the target antibody binds is determined, and this antigen fragment is used as the binding epitope of the target antibody and the target antigen.

2. The method according to claim 1, characterized in that, The determination of surface residues on the structure of the target antigen includes: Calculate the solvent-accessible surface area of ​​each residue in the structure of the target antigen; Based on the comparison between the solvent-accessible surface area of ​​the residue and a threshold, it is determined whether the residue is a surface residue.

3. The method according to claim 1, characterized in that, The calculation of the similarity between at least two neighboring antigen fragments includes: The total number is the number of all surface residues contained in the antigen fragment closest to the first cleavage point among at least two adjacent antigen fragments; The ratio of the number of identical surface residues in at least two adjacent antigen fragments to the total number of such residues is calculated, and this ratio is used as the similarity score.

4. The method according to claim 1, characterized in that, The step of predicting the affinity of the target antibody for each antigen fragment based on the structure of the antigen fragment and the structure of the target antibody, and obtaining affinity data, includes: The structure of the antigen fragment and the target antibody are expressed using graph convolution to generate the structural features of the antigen fragment and the target antibody; The structural features of the antigen fragment and the target antibody are fused using an attention mechanism to generate a fusion feature; Based on the fusion characteristics, the binding probability of the antigen fragment and the target antibody is mapped.

5. The method according to claim 4, characterized in that, The step of determining the antigen fragment to which the target antibody binds based on the affinity data, and using this antigen fragment as the binding epitope of the target antibody and the target antigen, includes: The antigen fragment whose binding probability satisfies a preset probability is selected as a candidate antigen fragment; The structures of spatially adjacent antigen fragments in the candidate antigen fragments are fused, and the fused candidate antigen fragments are used as the binding epitopes of the target antibody and the target antigen.

6. A device for predicting antibody and antigen binding epitopes, characterized in that, include: The acquisition module is used to acquire the structures of the target antigen and the target antibody; A first determining module is used to determine surface residues on the structure of the target antigen; A segmentation module is used to segment the structure of the target antigen into multiple antigen fragments based on the surface residues; It includes: randomly selecting a surface residue as a slicing point and arranging the slicing order of the slicing point as the first position; calculating the surface residue closest to the m-th slicing point outside the preset slicing range of the m-th slicing point and arranging it as the slicing point, and arranging the slicing order of the m+1-th slicing point, where m≥1; slicing all slicing points sequentially from the first slicing point according to the preset slicing range, with each slicing point as the center, and generating multiple antigen fragments; calculating the similarity between at least two neighboring antigen fragments; if the similarity reaches a preset value, then the antigen fragment farthest from the first slicing point is taken as a redundant antigen fragment and the redundant antigen fragment is removed; An affinity prediction module is used to predict the affinity between the target antibody and each of the antigen fragments based on the structure of the antigen fragments and the structure of the target antibody, thereby obtaining affinity data. The second determining module is used to determine the antigen fragment that the target antibody binds to based on the affinity data, and to use the antigen fragment as the binding epitope of the target antibody and the target antigen.

7. An electronic device, characterized in that, include: processor; as well as A memory having executable code stored thereon, which, when executed by the processor, causes the processor to perform the method as described in any one of claims 1-5.

8. A computer-readable storage medium having executable code stored thereon, which, when executed by a processor of an electronic device, causes the processor to perform the method as described in any one of claims 1-5.

Images