Method, apparatus, and medium for predicting nanobody-antigen interaction

Abstract

The application relates to a nanobody and antigen interaction prediction method and device, computer equipment and a storage medium. The method comprises the following steps: obtaining a to-be-tested nanobody sequence and a to-be-tested antigen sequence; generating a nanobody embedding vector and an antigen embedding vector respectively by using a first embedding generator according to the to-be-tested nanobody sequence and the to-be-tested antigen sequence; generating a nanobody-antigen pair embedding vector by using a second embedding generator according to the to-be-tested nanobody sequence and the to-be-tested antigen sequence; fusing the nanobody embedding vector and the nanobody-antigen pair embedding vector to obtain nanobody fusion features, and fusing the antigen embedding vector and the nanobody-antigen pair embedding vector to obtain antigen fusion features; and predicting the interaction between the to-be-tested nanobody sequence and the to-be-tested antigen sequence according to the nanobody fusion features and the antigen fusion features. The method can improve the nanobody and antigen interaction prediction accuracy.

Description

Technical Field

[0001] This application relates to the field of bioinformatics, and in particular to a method, apparatus, computer device, and storage medium for predicting the interaction between nanobodies and antigens. Background Technology

[0002] Nanobodies, also known as variable regions (VHHs) of heavy chain antibodies, are unique single-domain fragments derived from naturally occurring heavy-chain-only antibodies found in animals such as camels. They possess relatively small molecular weights (12-15 kDa) and only three complementarity-determining regions (CDR1, CDR2, and CDR3), offering significant advantages over traditional antibodies in terms of tissue penetration, blood-brain barrier permeability, stability, and production cost. In recent years, nanobodies have garnered significant attention due to their wide range of applications in the biomedical field, encompassing sensitive detection methods and novel therapeutic strategies. As nanobodies research continues to deepen, the ongoing release of relevant public databases has driven advancements in algorithmic research. In the prediction of nanobodies-antigen interactions, computational methods have undergone a paradigm shift from traditional approaches to machine learning and deep learning models.

[0003] However, in existing technologies, predictions based on machine learning and deep learning models rely solely on energy scores or sequence information. These methods fail to fully utilize multimodal data, resulting in limited predictive power and lower-than-expected prediction accuracy. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, and storage medium for predicting nanobody-antigen interactions that can improve the accuracy of predicting nanobody-antigen interactions, in order to address the aforementioned technical problems.

[0005] Firstly, a method for predicting the interaction between nanobodies and antigens is provided, the method comprising:

[0006] Obtain the sequence of the nanobody to be tested and the sequence of the antigen to be tested;

[0007] Based on the nanobody sequence and the antigen sequence to be tested, a first embedding generator is used to generate nanobody embedding vectors and antigen embedding vectors, respectively; wherein, the first embedding generator is a protein embedding generator.

[0008] Based on the nanobody sequence to be tested and the antigen sequence to be tested, a nanobody antigen pair embedding vector is generated using a second embedding generator; wherein, the second embedding generator is a structure prediction embedding generator.

[0009] The nanobody embedding vector and the nanobody antigen pair embedding vector are fused to obtain the nanobody fusion feature, and the antigen embedding vector and the nanobody antigen pair embedding vector are fused to obtain the antigen fusion feature.

[0010] The interaction between the nanobody sequence and the antigen sequence to be tested is predicted based on the fusion characteristics of nanobodies and antigens.

[0011] In some embodiments, both the nanobody embedding vector and the nanobody antigen pair embedding vector include a first-dimensional parameter, a second-dimensional parameter, and a third-dimensional parameter; fusing the nanobody embedding vector and the nanobody antigen pair embedding vector yields a nanobody fusion feature, including:

[0012] The order of the second and third dimension parameters of the nanobody embedding vector is interchanged to obtain the adjusted nanobody embedding vector.

[0013] By swapping the order of the second and third dimension parameters of the nanoantibody antigen pair embedding vector, the adjusted nanoantibody antigen pair embedding vector is obtained.

[0014] The antibody fusion weights are calculated based on the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector.

[0015] The nanobody fusion characteristics are calculated based on the nanobody embedding vector, the nanobody antigen pair embedding vector, and the antibody fusion weight.

[0016] In some embodiments, the first dimension parameter represents the batch size, the second dimension parameter represents the longest length of the sequence, and the third dimension parameter represents the latent dimension of the feature.

[0017] In some embodiments, calculating antibody fusion weights based on the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector includes:

[0018] The adjusted nanobody embedding vector and the adjusted nanobody antigen-to-embedding vector are summed to obtain the summed feature.

[0019] The summed features are sequentially processed by first convolution, batch normalization, and activation function to obtain intermediate output data.

[0020] The intermediate output data is sequentially processed by a second convolution, batch normalization, and activation function to obtain the antibody fusion weights.

[0021] In some embodiments, the antibody fusion weight includes a first-dimensional parameter, a second-dimensional parameter, and a third-dimensional parameter; the nanobody fusion features are calculated based on the nanobody embedding vector, the nanobody antigen pair embedding vector, and the antibody fusion weight, including:

[0022] The order of the second and third dimension parameters of the antibody fusion weights is swapped to obtain the adjusted antibody fusion weights.

[0023] The nanobody fusion characteristics are calculated based on the adjusted antibody fusion weights, nanobody embedding vectors, and nanobody antigen pair embedding vectors.

[0024] In some embodiments, both the antigen embedding vector and the nanobody antigen pair embedding vector include a first dimension parameter, a second dimension parameter, and a third dimension parameter; fusing the antigen embedding vector and the nanobody antigen pair embedding vector yields an antigen fusion feature, including:

[0025] The order of the second and third dimension parameters of the antigen embedding vector is swapped to obtain the adjusted antigen embedding vector;

[0026] By swapping the order of the second and third dimension parameters of the nanoantibody antigen pair embedding vector, the adjusted nanoantibody antigen pair embedding vector is obtained.

[0027] The antigen fusion weights are calculated based on the adjusted antigen embedding vector and the adjusted nanobody antigen pair embedding vector.

[0028] Antigen fusion characteristics are calculated based on antigen embedding vectors, nanobody antigen pair embedding vectors, and antigen fusion weights.

[0029] In some embodiments, predicting the interaction between the target nanobody sequence and the target antigen sequence based on nanobody fusion characteristics and antigen fusion characteristics includes:

[0030] The nanobody fusion feature and the antigen fusion feature are spliced ​​together to obtain the spliced ​​feature;

[0031] The spliced ​​features are then subjected to first pooling, second pooling, and third pooling, respectively.

[0032] The output of the first pooling is processed by the first perception mechanism, the output of the second pooling is processed by the second perception mechanism, and the output of the third pooling is processed by the third perception mechanism to obtain the first processing result, the second processing result, and the third processing result, respectively.

[0033] Based on the first, second, and third processing results, the predicted interaction between the nanobody sequence to be tested and the antigen sequence to be tested is obtained.

[0034] Secondly, a device for predicting the interaction between a nanobody and an antigen is provided, the device comprising:

[0035] The data acquisition module is used to acquire the sequence of the nanobody to be tested and the sequence of the antigen to be tested.

[0036] The first embedding generation module is used to generate nanobody embedding vectors and antigen embedding vectors respectively based on the nanobody sequence to be tested and the antigen sequence to be tested using a first embedding generator; wherein, the first embedding generator is an embedding generator for a protein language model.

[0037] The second embedding generation module is used to generate a nanobody antigen pair embedding vector based on the nanobody sequence to be tested and the antigen sequence to be tested using a second embedding generator; wherein, the second embedding generator is the embedding generator of the structure prediction model;

[0038] The feature fusion module is used to fuse the nanobody embedding vector and the nanobody antigen pair embedding vector to obtain the nanobody fusion feature, and to fuse the antigen embedding vector and the nanobody antigen pair embedding vector to obtain the antigen fusion feature.

[0039] An integrated decision module is used to predict the interaction between the target nanobody sequence and the target antigen sequence based on the nanobody fusion characteristics and antigen fusion characteristics.

[0040] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method described in the first aspect.

[0041] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, wherein the computer program, when executed by a processor, implements the steps of the method described in the first aspect.

[0042] The aforementioned method, apparatus, computer equipment, and storage medium for predicting nanobody-antigen interactions, wherein the nanobody-antigen interaction prediction method, by fusing the nanobody embedding vector and antigen embedding vector extracted by the protein embedding generator (i.e., the first embedding generator) with the nanobody-antigen pair embedding vector extracted by the structure prediction embedding generator (i.e., the second embedding generator), can obtain deep features that integrate protein sequence information and antibody-antigen complex spatial structure information, thereby improving the accuracy of predicting the interaction between nanobody and antigen. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating the method for predicting the interaction between nanobodies and antigens in some embodiments;

[0044] Figure 2This is a flowchart illustrating the steps of fusing nanobody embedding vectors and nanobody antigen embedding vectors in some embodiments to obtain nanobody fusion feature steps.

[0045] Figure 3 This is a schematic diagram illustrating the technical implementation of the nanobody-antigen interaction prediction architecture in some embodiments;

[0046] Figure 4 This is a structural block diagram of the nanobody-antigen interaction prediction device in some embodiments;

[0047] Figure 5 This is a diagram showing the internal structure of a computer device in some embodiments. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0049] The nanobody-antigen interaction prediction method provided in this application can be applied to computer devices. These computer devices can be terminal devices, including, but are not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices.

[0050] In some embodiments, such as Figure 1 As shown, a method for predicting the interaction between nanobodies and antigens is provided, and this method is applied to... Figure 1 Taking the terminal in the example, the explanation includes the following steps:

[0051] Step S102: Obtain the sequence of the nanobody to be tested and the sequence of the antigen to be tested.

[0052] In this step, the nanobody sequence to be tested refers to the amino acid or nucleic acid sequence of the nanobody that meets the prediction requirements. Its source can typically include, but is not limited to, camels or be obtained through artificial synthesis and modification. The antigen sequence to be tested refers to a biomolecule sequence that has the potential to serve as a binding target for the nanobody, including but not limited to the amino acid sequence of a protein or peptide, or its encoding nucleic acid sequence. Its binding characteristics with the nanobody need to be predicted and evaluated using this method. The nanobody sequence to be tested and the antigen sequence to be tested have complementary structures, and the pairing combinations they form can characterize potential antibody-antigen interactions.

[0053] Step S104: Based on the nanobody sequence to be tested and the antigen sequence to be tested, a first embedding generator is used to generate a nanobody embedding vector and an antigen embedding vector, respectively; wherein, the first embedding generator is a protein embedding generator.

[0054] The first embedding generator can be a protein embedding generator, which can be a protein embedding generator based on a protein language model, including but not limited to a PLM embedding generator (Protein Language Model Embedding Generator).

[0055] Specifically, there can be one or more first embedding generators. For example, in the case of multiple first embedding generators, the nanobody sequence to be tested and the antigen sequence to be tested can be input into their respective first embedding generators, so that the nanobody embedding vector and the antigen embedding vector can be output by their respective first embedding generators.

[0056] For example, the specific process of generating nanobody embedding vectors and antigen embedding vectors using the first embedding generator can be referred to as follows:

[0057] ,

[0058] ,

[0059] in, This represents the protein embedding generator based on the protein language model, i.e., the first embedding generator; and These represent the sequences of the nanobody to be tested and the antigen to be tested, respectively. and These represent the PLM nanobody embedding vector and the PLM antigen embedding vector generated based on the protein embedding generator, respectively.

[0060] Step S106: Based on the nanobody sequence to be tested and the antigen sequence to be tested, generate a nanobody antigen pair embedding vector using a second embedding generator; wherein, the second embedding generator is a structure prediction embedding generator.

[0061] The second embedding generator is a structure prediction embedding generator, which can be an embedding generator based on a structure prediction model, including but not limited to an SPM embedding generator (Structure Prediction Model EmbeddingGenerator).

[0062] Specifically, the sequence of the nanobody to be tested and the sequence of the antigen to be tested, i.e., the sequence of the nanobody-antigen pair to be tested, can be input into the embedding generator based on the structure prediction model, i.e., input into the second embedding generator, and the nanobody-antigen pair embedding vector is output by the second embedding generator.

[0063] For example, the specific process of generating a nanobody-antigen pair embedding vector using a second embedding generator based on the nanobody sequence to be tested and the antigen sequence to be tested can be referred to as follows:

[0064] ,

[0065] in, This represents the embedding generator based on the structural prediction model, i.e., the second embedding generator; This represents a combination of the nanobody sequence to be tested and the antigen sequence to be tested; This represents the SPM nanobody antigen pair embedding vector output by the second embedding generator.

[0066] Step S108: The nanobody embedding vector and the nanobody antigen pair embedding vector are fused to obtain the nanobody fusion feature, and the antigen embedding vector and the nanobody antigen pair embedding vector are fused to obtain the antigen fusion feature.

[0067] In this step, the nanobody embedding vector and the antigen embedding vector can be fused with the nanobody antigen pair embedding vector respectively. For example, feature fusion can be performed based on the attention fusion mechanism to obtain the nanobody fusion feature obtained by fusing the nanobody embedding vector and the nanobody antigen pair embedding vector, and the antigen fusion feature obtained by fusing the antigen embedding vector and the nanobody antigen pair embedding vector.

[0068] The specific process of fusing the nanobody embedding vector and the antigen embedding vector with the nanobody-antigen pair embedding vector can be found below:

[0069] ,

[0070] ,

[0071] in, This indicates the attention fusion mechanism; and These represent the fusion characteristics of the nanobody and the antigen fusion characteristics obtained after fusion, respectively.

[0072] Step S110: Based on the nanobody fusion characteristics and antigen fusion characteristics, predict the interaction between the nanobody sequence to be tested and the antigen sequence to be tested.

[0073] In this step, based on the fusion features of the nanobody and antigen after feature fusion, and further combined with different pooling strategies, different data that can characterize the interaction prediction results between the nanobody sequence to be tested and the antigen sequence to be tested are output according to the task output requirements. For example, the output value for the regression task is a continuous value, and the output value for the classification task is a discrete value of 0 or 1.

[0074] In the above-mentioned method for predicting the interaction between nanobodies and antigens, the nanobodies embedding vector and the antigen embedding vector extracted by the protein embedding generator (i.e., the first embedding generator) are fused with the nanobodies-antigen pair embedding vector extracted by the structure prediction embedding generator (i.e., the second embedding generator). This allows for the acquisition of deep features that integrate protein sequence information and antibody-antigen complex spatial structure information, thereby improving the accuracy of predicting the interaction between nanobodies and antigens.

[0075] In some embodiments, reference may be made to Figure 2 As shown, both the nanobody embedding vector and the nanobody antigen pair embedding vector include a first-dimensional parameter, a second-dimensional parameter, and a third-dimensional parameter. Fusing the nanobody embedding vector and the nanobody antigen pair embedding vector yields the nanobody fusion feature, which may specifically include:

[0076] Step S202: Swap the order of the second and third dimension parameters of the nanobody embedding vector to obtain the adjusted nanobody embedding vector.

[0077] Step S204: Swap the order of the second and third dimension parameters of the nanoantibody antigen pair embedding vector to obtain the adjusted nanoantibody antigen pair embedding vector.

[0078] In the two steps described above, based on the attention fusion mechanism First, the order of the second and third dimension parameters in the nanobody embedding vector and the nanobody antigen pair embedding vector can be interchanged:

[0079] ,

[0080] ,

[0081] in, This is the first dimension parameter, which can represent the batch size. This is the second dimension parameter, which can represent the longest length of the sequence; for example, it can be 1000. It is the third dimension parameter, which can represent the hidden dimension of the feature. and These are the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector, respectively, after swapping the second and third dimension parameters.

[0082] Step S206: Calculate the antibody fusion weight based on the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector.

[0083] In this step, the fusion weights can be obtained by processing the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector through convolution, batch normalization, activation functions, etc.

[0084] Step S208: Calculate the nanobody fusion characteristics based on the nanobody embedding vector, the nanobody antigen pair embedding vector, and the antibody fusion weight.

[0085] In this step, the weights of each component feature are calculated based on the nanobody embedding vector, the nanobody antigen pair embedding vector, and the calculated antibody fusion weights, thereby obtaining the nanobody fusion features.

[0086] In some embodiments, calculating antibody fusion weights based on the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector includes: summing the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector to obtain a summed feature; sequentially performing a first convolution, batch normalization, and activation function processing on the summed feature to obtain intermediate output data; and sequentially performing a second convolution, batch normalization, and activation function processing on the intermediate output data to obtain the antibody fusion weights. For example, this can be implemented using the following formula:

[0087] ,

[0088] in, and These represent the first 1D convolution (first convolution processing) and the second 1D convolution (second convolution processing), respectively. Representative batch normalization processing; This indicates the processing of the Rectified Linear Unit (ReLU); This indicates the processing of the sigmoid function; This indicates the fusion weight.

[0089] In this embodiment, through processing operations such as two-level convolutional networks, batch normalization, and activation functions, nonlinear fusion and weight learning are performed on two types of features. This can adaptively allocate antibody fusion weights according to the importance of sequence and structural information, further improving the discriminative ability of feature combinations.

[0090] In some embodiments, the antibody fusion weight includes a first dimension parameter, a second dimension parameter, and a third dimension parameter; calculating the nanobody fusion feature based on the nanobody embedding vector, the nanobody antigen pair embedding vector, and the antibody fusion weight includes: swapping the order of the second dimension parameter and the third dimension parameter of the antibody fusion weight to obtain an adjusted antibody fusion weight; and calculating the nanobody fusion feature based on the adjusted antibody fusion weight, the nanobody embedding vector, and the nanobody antigen pair embedding vector.

[0091] For example, the following formula can be used as a reference:

[0092] ,

[0093] ,

[0094] in, It is the adjusted antibody fusion weight after swapping the second and third dimension parameters; Represents the tensor product; This refers to the fusion characteristics of nanobody obtained after fusion.

[0095] In this embodiment, by dynamically adjusting the dimensions of the weight matrix to adapt to the spatial structure of the embedded vector, it is possible to achieve precise alignment and weighted fusion of features from different sources at the dimensional level, thereby enhancing the spatial representation capability of the fused features.

[0096] In some embodiments, both the antigen embedding vector and the nanobody antigen pair embedding vector include a first dimension parameter, a second dimension parameter, and a third dimension parameter. Fusing the antigen embedding vector and the nanobody antigen pair embedding vector to obtain antigen fusion features includes: swapping the order of the second and third dimension parameters of the antigen embedding vector to obtain an adjusted antigen embedding vector; swapping the order of the second and third dimension parameters of the nanobody antigen pair embedding vector to obtain an adjusted nanobody antigen pair embedding vector; calculating antigen fusion weights based on the adjusted antigen embedding vector and the adjusted nanobody antigen pair embedding vector; and calculating antigen fusion features based on the antigen embedding vector, the nanobody antigen pair embedding vector, and the antigen fusion weights.

[0097] In this embodiment, the specific implementation process of fusing the antigen embedding vector and the nanobody antigen pair embedding vector to obtain the antigen fusion feature can be referred to in the specific process of fusing the nanobody embedding vector and the nanobody antigen pair embedding vector to obtain the nanobody fusion feature, which will not be repeated here.

[0098] In some embodiments, predicting the interaction between the target nanobody sequence and the target antigen sequence based on nanobody fusion characteristics and antigen fusion characteristics includes: splicing the nanobody fusion characteristics and antigen fusion characteristics to obtain spliced ​​characteristics; performing first pooling, second pooling, and third pooling on the spliced ​​characteristics respectively; processing the output of the first pooling with a first sensing mechanism, processing the output of the second pooling with a second sensing mechanism, and processing the output of the third pooling with a third sensing mechanism to obtain first processing results, second processing results, and third processing results respectively; and obtaining the prediction result of the interaction between the target nanobody sequence and the target antigen sequence based on the first processing result, second processing result, and third processing result.

[0099] In this embodiment, after the nanobody fusion feature and the antigen fusion feature are spliced ​​together, they can be processed through various pooling strategies, such as preferably three pooling strategies, to obtain more information-rich protein characterization data. Then, they are processed through a multilayer perceptron, and finally the average value is taken as the final output.

[0100] For example, the following formula can be used as a reference:

[0101] ,

[0102] in, This indicates tensor (feature) concatenation; , , These represent minimum pooling (first pooling), average pooling (second pooling), and maximum pooling (third pooling), respectively. , , These represent multilayer perceptrons.

[0103] In this embodiment, by using multiple pooling strategies and corresponding perception mechanisms in parallel processing, key information from multi-scale features can be extracted and integrated, thereby capturing interaction patterns more comprehensively and improving the accuracy and reliability of prediction results.

[0104] It should be understood that, although Figures 1 to 2The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figures 1 to 2 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0105] In some embodiments, reference may be made to Figure 3 As shown, Figure 3 The diagram illustrates the technical implementation of the nanobody-antigen interaction prediction architecture in some embodiments. The amino acid sequences of the nanobody to be tested (e.g., V, Q, L, A, etc.) and the amino acid sequences of the antigen to be tested (e.g., L, N, N, I, etc.) are first fed into a first embedding generator. This generator consists of a protein language model (PLM) and a stacked Transformer layer structure, which, after encoding, generates a single-modality feature embedding vector. Simultaneously, the two amino acid sequences are fed into a second embedding generator in residue pairs. This generator consists of a structure prediction model (SPM) and a stacked Transformer layer structure, which, after encoding, generates another single-modality feature embedding vector. Both types of feature embeddings are simultaneously fed into an attention fusion module. This module uses an attention pooling layer (APM) to weight and aggregate the bimodal features, and combines residual connections to retain the original feature information, generating a more fused feature representation. Finally, the fused features are input into an integration decision module, which calculates and outputs the predicted binding affinity between the nanobody and the antigen.

[0106] In some embodiments, such as Figure 4 As shown, a device for predicting the interaction between a nanobody and an antigen is provided, comprising: a data acquisition module 410, a first embedding generation module 420, a second embedding generation module 430, a feature fusion module 440, and an integration decision module 450, wherein:

[0107] Data acquisition module 410 is used to acquire the sequence of the nanobody to be tested and the sequence of the antigen to be tested;

[0108] The first embedding generation module 420 is used to generate nanobody embedding vectors and antigen embedding vectors respectively based on the nanobody sequence to be tested and the antigen sequence to be tested using a first embedding generator; wherein, the first embedding generator is an embedding generator for a protein language model.

[0109] The second embedding generation module 430 is used to generate a nanobody antigen pair embedding vector based on the nanobody sequence to be tested and the antigen sequence to be tested using a second embedding generator; wherein, the second embedding generator is the embedding generator of the structure prediction model.

[0110] The feature fusion module 440 is used to fuse the nanobody embedding vector and the nanobody antigen pair embedding vector to obtain nanobody fusion features, and to fuse the antigen embedding vector and the nanobody antigen pair embedding vector to obtain antigen fusion features;

[0111] The integrated decision module 450 is used to predict the interaction between the nanobody sequence to be tested and the antigen sequence to be tested based on the nanobody fusion characteristics and the antigen fusion characteristics.

[0112] In some embodiments, the feature fusion module 440 is specifically used to interchange the order of the second and third dimension parameters of the nanobody embedding vector to obtain an adjusted nanobody embedding vector; interchange the order of the second and third dimension parameters of the nanobody antigen pair embedding vector to obtain an adjusted nanobody antigen pair embedding vector; calculate the antibody fusion weight based on the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector; and calculate the nanobody fusion feature based on the nanobody embedding vector, the nanobody antigen pair embedding vector, and the antibody fusion weight.

[0113] In some embodiments, the feature fusion module 440 is further configured to sum the adjusted nanobody embedding vector and the adjusted nanobody antigen-to-embedding vector to obtain a summed feature; perform a first convolution, batch normalization, and activation function processing on the summed feature in sequence to obtain intermediate output data; and perform a second convolution, batch normalization, and activation function processing on the intermediate output data in sequence to obtain antibody fusion weights.

[0114] In some embodiments, the feature fusion module 440 is further configured to interchange the order of the second-dimensional parameters and the third-dimensional parameters of the antibody fusion weight to obtain the adjusted antibody fusion weight; and to calculate the nanobody fusion feature based on the adjusted antibody fusion weight, the nanobody embedding vector, and the nanobody antigen pair embedding vector.

[0115] In some embodiments, the feature fusion module 440 is further configured to: interchange the order of the second and third dimension parameters of the antigen embedding vector to obtain an adjusted antigen embedding vector; interchange the order of the second and third dimension parameters of the nanoantibody antigen pair embedding vector to obtain an adjusted nanoantibody antigen pair embedding vector; calculate the antigen fusion weight based on the adjusted antigen embedding vector and the adjusted nanoantibody antigen pair embedding vector; and calculate the antigen fusion feature based on the antigen embedding vector, the nanoantibody antigen pair embedding vector, and the antigen fusion weight.

[0116] In some embodiments, the integration decision module 450 is specifically used to splice the nanobody fusion feature and the antigen fusion feature to obtain a spliced ​​feature; perform a first pooling, a second pooling, and a third pooling on the spliced ​​feature; process the output of the first pooling with a first sensing mechanism, process the output of the second pooling with a second sensing mechanism, and process the output of the third pooling with a third sensing mechanism to obtain a first processing result, a second processing result, and a third processing result, respectively; and obtain a prediction result of the interaction between the nanobody sequence to be tested and the antigen sequence to be tested based on the first processing result, the second processing result, and the third processing result.

[0117] Specific limitations regarding the nanobody-antigen interaction prediction device can be found in the limitations of the nanobody-antigen interaction prediction method described above, and will not be repeated here. Each module in the aforementioned nanobody-antigen interaction prediction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0118] In some embodiments, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5As shown, the computer device includes a processor, memory, network interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for predicting the interaction between nanobody and antigen. The display screen can be a liquid crystal display (LCD) or an e-ink display. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0119] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0120] In some embodiments, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following steps: acquiring a nanobody sequence to be tested and a antigen sequence to be tested; generating a nanobody embedding vector and an antigen embedding vector respectively using a first embedding generator based on the nanobody sequence to be tested and the antigen sequence to be tested; wherein the first embedding generator is a protein embedding generator; generating a nanobody-antigen pair embedding vector using a second embedding generator based on the nanobody sequence to be tested and the antigen sequence to be tested; wherein the second embedding generator is a structure prediction embedding generator; fusing the nanobody embedding vector and the nanobody-antigen pair embedding vector to obtain a nanobody fusion feature; fusing the antigen embedding vector and the nanobody-antigen pair embedding vector to obtain an antigen fusion feature; and predicting the interaction between the nanobody sequence to be tested and the antigen sequence to be tested based on the nanobody fusion feature and the antigen fusion feature.

[0121] In some embodiments, when the processor executes the computer program, it further implements the following steps: swapping the order of the second and third dimension parameters of the nanobody embedding vector to obtain an adjusted nanobody embedding vector; swapping the order of the second and third dimension parameters of the nanobody antigen pair embedding vector to obtain an adjusted nanobody antigen pair embedding vector; calculating antibody fusion weights based on the adjusted nanobody embedding vectors and the adjusted nanobody antigen pair embedding vectors; and calculating nanobody fusion features based on the nanobody embedding vectors, the nanobody antigen pair embedding vectors, and the antibody fusion weights.

[0122] In some embodiments, when the processor executes the computer program, it further performs the following steps: summing the adjusted nanobody embedding vector and the adjusted nanobody antigen-to-embedding vector to obtain a summed feature; sequentially performing a first convolution, batch normalization, and activation function processing on the summed feature to obtain intermediate output data; and sequentially performing a second convolution, batch normalization, and activation function processing on the intermediate output data to obtain antibody fusion weights.

[0123] In some embodiments, when the processor executes the computer program, it further performs the following steps: swapping the order of the second-dimensional parameters and the third-dimensional parameters of the antibody fusion weight to obtain the adjusted antibody fusion weight; and calculating the nanobody fusion feature based on the adjusted antibody fusion weight, the nanobody embedding vector, and the nanobody antigen pair embedding vector.

[0124] In some embodiments, when the processor executes the computer program, it further implements the following steps: swapping the order of the second and third dimension parameters of the antigen embedding vector to obtain an adjusted antigen embedding vector; swapping the order of the second and third dimension parameters of the nanoantibody antigen pair embedding vector to obtain an adjusted nanoantibody antigen pair embedding vector; calculating the antigen fusion weight based on the adjusted antigen embedding vector and the adjusted nanoantibody antigen pair embedding vector; and calculating the antigen fusion feature based on the antigen embedding vector, the nanoantibody antigen pair embedding vector, and the antigen fusion weight.

[0125] In some embodiments, when the processor executes the computer program, it further performs the following steps: splicing the nanobody fusion feature and the antigen fusion feature to obtain a spliced ​​feature; performing a first pooling, a second pooling, and a third pooling on the spliced ​​feature; processing the output of the first pooling with a first sensing mechanism, processing the output of the second pooling with a second sensing mechanism, and processing the output of the third pooling with a third sensing mechanism to obtain a first processing result, a second processing result, and a third processing result, respectively; and obtaining a prediction result of the interaction between the nanobody sequence to be tested and the antigen sequence to be tested based on the first processing result, the second processing result, and the third processing result.

[0126] In some embodiments, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it performs the following steps: acquiring a nanobody sequence to be tested and a antigen sequence to be tested; generating a nanobody embedding vector and an antigen embedding vector respectively using a first embedding generator based on the nanobody sequence to be tested and the antigen sequence to be tested; wherein the first embedding generator is a protein embedding generator; generating a nanobody-antigen pair embedding vector using a second embedding generator based on the nanobody sequence to be tested and the antigen sequence to be tested; wherein the second embedding generator is a structure prediction embedding generator; fusing the nanobody embedding vector and the nanobody-antigen pair embedding vector to obtain a nanobody fusion feature, and fusing the antigen embedding vector and the nanobody-antigen pair embedding vector to obtain an antigen fusion feature; and predicting the interaction between the nanobody sequence to be tested and the antigen sequence to be tested based on the nanobody fusion feature and the antigen fusion feature.

[0127] In some embodiments, when the computer program is executed by the processor, it further implements the following steps: swapping the order of the second and third dimension parameters of the nanobody embedding vector to obtain an adjusted nanobody embedding vector; swapping the order of the second and third dimension parameters of the nanobody antigen pair embedding vector to obtain an adjusted nanobody antigen pair embedding vector; calculating antibody fusion weights based on the adjusted nanobody embedding vectors and the adjusted nanobody antigen pair embedding vectors; and calculating nanobody fusion features based on the nanobody embedding vectors, the nanobody antigen pair embedding vectors, and the antibody fusion weights.

[0128] In some embodiments, when the computer program is executed by the processor, it further performs the following steps: summing the adjusted nanobody embedding vector and the adjusted nanobody antigen-to-embedding vector to obtain a summed feature; sequentially performing a first convolution, batch normalization, and activation function processing on the summed feature to obtain intermediate output data; and sequentially performing a second convolution, batch normalization, and activation function processing on the intermediate output data to obtain antibody fusion weights.

[0129] In some embodiments, when the computer program is executed by the processor, it further performs the following steps: swapping the order of the second-dimensional parameters and the third-dimensional parameters of the antibody fusion weight to obtain the adjusted antibody fusion weight; and calculating the nanobody fusion feature based on the adjusted antibody fusion weight, the nanobody embedding vector, and the nanobody antigen pair embedding vector.

[0130] In some embodiments, when the computer program is executed by the processor, it further implements the following steps: swapping the order of the second and third dimension parameters of the antigen embedding vector to obtain an adjusted antigen embedding vector; swapping the order of the second and third dimension parameters of the nanoantibody antigen pair embedding vector to obtain an adjusted nanoantibody antigen pair embedding vector; calculating the antigen fusion weight based on the adjusted antigen embedding vector and the adjusted nanoantibody antigen pair embedding vector; and calculating the antigen fusion feature based on the antigen embedding vector, the nanoantibody antigen pair embedding vector, and the antigen fusion weight.

[0131] In some embodiments, when the computer program is executed by the processor, it further implements the following steps: splicing the nanobody fusion feature and the antigen fusion feature to obtain a spliced ​​feature; performing a first pooling, a second pooling, and a third pooling on the spliced ​​feature; processing the output of the first pooling with a first sensing mechanism, processing the output of the second pooling with a second sensing mechanism, and processing the output of the third pooling with a third sensing mechanism to obtain a first processing result, a second processing result, and a third processing result, respectively; and obtaining a prediction result of the interaction between the nanobody sequence to be tested and the antigen sequence to be tested based on the first processing result, the second processing result, and the third processing result.

[0132] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0133] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0134] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the characters in this article generally indicate that the preceding and following related objects have an "or" relationship.

[0135] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

[0136] It should be noted that, in the embodiments of this application, data related to user information or user data must be obtained and processed only after the user's authorization and consent. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

Claims

1. A method for predicting the interaction between nanobodies and antigens, characterized in that, The method includes: Obtain the sequence of the nanobody to be tested and the sequence of the antigen to be tested; Based on the nanobody sequence to be tested and the antigen sequence to be tested, a first embedding generator is used to generate nanobody embedding vectors and antigen embedding vectors, respectively; wherein, the first embedding generator is an embedding generator for a protein language model. Based on the nanobody sequence to be tested and the antigen sequence to be tested, a nanobody antigen pair embedding vector is generated using a second embedding generator; wherein, the second embedding generator is the embedding generator of the structure prediction model; The nanobody embedding vector and the nanobody-antigen pair embedding vector are fused to obtain a nanobody fusion feature, and the antigen embedding vector and the nanobody-antigen pair embedding vector are fused to obtain an antigen fusion feature. Each of the nanobody embedding vector and the nanobody-antigen pair embedding vector includes a first-dimensional parameter, a second-dimensional parameter, and a third-dimensional parameter. The process of fusing the nanobody embedding vector and the nanobody-antigen pair embedding vector to obtain the nanobody fusion feature includes: swapping the order of the second-dimensional and third-dimensional parameters of the nanobody embedding vector to obtain an adjusted nanobody embedding vector; swapping the order of the second-dimensional and third-dimensional parameters of the nanobody-antigen pair embedding vector to obtain an adjusted nanobody-antigen pair embedding vector; calculating antibody fusion weights based on the adjusted nanobody embedding vector and the adjusted nanobody-antigen pair embedding vector; and calculating the nanobody fusion feature based on the nanobody embedding vector, the nanobody-antigen pair embedding vector, and the antibody fusion weights. Based on the nanobody fusion characteristics and the antigen fusion characteristics, the interaction between the nanobody sequence to be tested and the antigen sequence to be tested is predicted.

2. The method according to claim 1, characterized in that, The first dimension parameter represents the batch size, the second dimension parameter represents the longest length of the sequence, and the third dimension parameter represents the latent dimension of the feature.

3. The method according to claim 1, characterized in that, The step of calculating the antibody fusion weight based on the adjusted nanobody embedding vector and the adjusted nanobody antigen pair embedding vector includes: The adjusted nanobody embedding vector and the adjusted nanobody antigen-pair embedding vector are summed to obtain the summed feature; The summed features are sequentially subjected to first convolution processing, batch normalization processing, and activation function processing to obtain intermediate output data; The intermediate output data is sequentially subjected to a second convolution process, a batch normalization process, and an activation function process to obtain the antibody fusion weights.

4. The method according to claim 1, characterized in that, The antibody fusion weight includes a first-dimensional parameter, a second-dimensional parameter, and a third-dimensional parameter; the calculation of the nanobody fusion feature based on the nanobody embedding vector, the nanobody antigen pair embedding vector, and the antibody fusion weight includes: The order of the second and third dimension parameters of the antibody fusion weights is interchanged to obtain the adjusted antibody fusion weights. The nanobody fusion feature is calculated based on the adjusted antibody fusion weight, the nanobody embedding vector, and the nanobody antigen pair embedding vector.

5. The method according to claim 1, characterized in that, Both the antigen embedding vector and the nanobody antigen pair embedding vector include a first-dimensional parameter, a second-dimensional parameter, and a third-dimensional parameter; the fusion of the antigen embedding vector and the nanobody antigen pair embedding vector to obtain antigen fusion features includes: The order of the second and third dimension parameters of the antigen embedding vector is interchanged to obtain the adjusted antigen embedding vector; The order of the second and third dimension parameters of the nanoantibody antigen pair embedding vector is interchanged to obtain the adjusted nanoantibody antigen pair embedding vector; The antigen fusion weight is calculated based on the adjusted antigen embedding vector and the adjusted nanobody antigen pair embedding vector; The antigen fusion feature is calculated based on the antigen embedding vector, the nanoantibody antigen pair embedding vector, and the antigen fusion weight.

6. The method according to claim 1, characterized in that, The step of predicting the interaction between the target nanobody sequence and the target antigen sequence based on the nanobody fusion characteristics and the antigen fusion characteristics includes: The nanobody fusion feature and the antigen fusion feature are spliced ​​together to obtain the spliced ​​feature; The splicing features are then subjected to first pooling, second pooling, and third pooling, respectively. The output of the first pooling is processed by the first perception mechanism, the output of the second pooling is processed by the second perception mechanism, and the output of the third pooling is processed by the third perception mechanism to obtain the first processing result, the second processing result, and the third processing result, respectively. Based on the first processing result, the second processing result, and the third processing result, the predicted results of the interaction between the nanobody sequence to be tested and the antigen sequence to be tested are obtained.

7. A device for predicting the interaction between nanobodies and antigens, characterized in that, The device includes: The data acquisition module is used to acquire the sequence of the nanobody to be tested and the sequence of the antigen to be tested. The first embedding generation module is used to generate nanobody embedding vectors and antigen embedding vectors respectively based on the nanobody sequence to be tested and the antigen sequence to be tested using a first embedding generator; wherein, the first embedding generator is an embedding generator for a protein language model. The second embedding generation module is used to generate a nanobody antigen pair embedding vector based on the nanobody sequence to be tested and the antigen sequence to be tested using a second embedding generator; wherein, the second embedding generator is the embedding generator of the structure prediction model; A feature fusion module is used to fuse the nanobody embedding vector and the nanobody-antigen pair embedding vector to obtain a nanobody fusion feature, and to fuse the antigen embedding vector and the nanobody-antigen pair embedding vector to obtain an antigen fusion feature. Each nanobody embedding vector and the nanobody-antigen pair embedding vector includes a first-dimensional parameter, a second-dimensional parameter, and a third-dimensional parameter. Specifically, the feature fusion module is used to interchange the order of the second-dimensional and third-dimensional parameters of the nanobody embedding vector to obtain an adjusted nanobody embedding vector; interchange the order of the second-dimensional and third-dimensional parameters of the nanobody-antigen pair embedding vector to obtain an adjusted nanobody-antigen pair embedding vector; calculate antibody fusion weights based on the adjusted nanobody embedding vectors and the adjusted nanobody-antigen pair embedding vectors; and calculate the nanobody fusion feature based on the nanobody embedding vectors, the nanobody-antigen pair embedding vectors, and the antibody fusion weights. An integrated decision module is used to predict the interaction between the nanobody sequence to be tested and the antigen sequence to be tested based on the nanobody fusion characteristics and the antigen fusion characteristics.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images