Influenza antigen prediction model processing method, device and equipment and readable storage medium

Abstract

The invention relates to an influenza antigen prediction model processing method and device, equipment and a readable storage medium. The method comprises the following steps: acquiring a training sample set; the training sample set comprises a plurality of matrix groups, and each matrix group comprises a protein sequence feature matrix and an antigen distance tag vector corresponding to the protein sequence feature matrix; determining a local weight according to a local key protein sequence feature vector in the protein sequence feature matrix included in each matrix group; connecting the protein sequence feature matrixes included in each matrix group in series to obtain a global protein sequence feature matrix, and determining a global weight according to global key protein sequence feature vectors in the global protein sequence feature matrix; according to the local weight and the global weight, constructing a preset influenza antigen prediction model; the preset influenza antigen prediction model is used for outputting an antigen prediction result of the to-be-detected protein sequence feature matrix. By adopting the method, the robustness of the influenza antigen prediction model can be improved.

Description

Technical Field

[0001] The present application relates to the field of machine learning technology, and in particular to a method, device, equipment and readable storage medium for processing an influenza antigen prediction model. Background Art

[0002] Influenza viruses are highly variable, especially influenza A viruses, which can cause new virus strains to emerge through antigenic drift and antigenic shift, thus causing influenza pandemics. Antigenic evolution is the main challenge in vaccine preparation. Therefore, developing more accurate antigen prediction methods to more accurately predict the evolutionary trend of influenza virus antigens is of great significance for responding to influenza virus epidemics and vaccine research and development and production. This can not only improve the timeliness of vaccines, but also better provide effective immune protection for the public. In traditional technology, antigens can be predicted by Bayesian models based on data such as antibody concentration and average affinity of antibodies to viral antigens. However, this method is difficult to accurately capture subtle changes in antigen evolution, resulting in low robustness of Bayesian models in the face of local changes in antigens. Summary of the invention

[0003] Based on this, it is necessary to provide an influenza antigen prediction model processing method, device, equipment and readable storage medium that can improve the robustness of the influenza antigen prediction model in response to the above technical problems.

[0004] In a first aspect, the present application provides a method for processing an influenza antigen prediction model, comprising:

[0005] Acquire a training sample set; the training sample set includes a plurality of matrix groups, the matrix groups include a protein sequence feature matrix and an antigen distance label vector corresponding to the protein sequence feature matrix; the protein sequence feature matrix includes a plurality of protein sequence feature vectors;

[0006] Determining local weights according to local key protein sequence feature vectors in the protein sequence feature matrices included in each of the matrix groups;

[0007] The protein sequence feature matrices included in each of the matrix groups are connected in series to obtain a global protein sequence feature matrix, and a global weight is determined according to a global key protein sequence feature vector in the global protein sequence feature matrix;

[0008] A preset influenza antigen prediction model is constructed according to the local weights and the global weights; the preset influenza antigen prediction model is used to output the antigen prediction result of the protein sequence feature matrix to be detected.

[0009] In one embodiment, determining the local weight according to the local key protein sequence feature vectors in the protein sequence feature matrix included in each matrix group includes:

[0010] For any matrix group in each of the matrix groups, according to the inner product value of the local key protein sequence feature vector in the protein sequence feature matrix included in the any matrix group and the antigen distance label vector, determine the local sub-weight corresponding to the any matrix group;

[0011] The local weight is determined according to the local sub-weights corresponding to the matrix groups.

[0012] In one embodiment, determining the local weight according to the local sub-weights corresponding to each of the matrix groups includes:

[0013] Constructing a local key protein sequence feature matrix according to the local key protein sequence feature vectors in the protein sequence feature matrix included in any one of the matrix groups;

[0014] Determine a conversion matrix according to the inverse matrix of the local key protein sequence feature matrix and the normalized feature vector corresponding to the local key protein sequence feature vector;

[0015] The local weight is determined according to the local sub-weights corresponding to each of the matrix groups and the transformation matrix.

[0016] In one embodiment, before determining the local weights according to the local key protein sequence feature vectors in the protein sequence feature matrices included in each of the matrix groups, the method further comprises:

[0017] For any matrix group in each of the matrix groups, determining a normalized feature vector corresponding to each protein sequence feature vector included in the any matrix group;

[0018] The local key protein sequence feature vectors in the protein sequence feature matrix included in any matrix group are determined according to the inner products of the antigen distance label vectors included in any matrix group and the normalized feature vectors corresponding to each protein sequence feature vector.

[0019] In one embodiment, determining the local key protein sequence feature vector in the protein sequence feature matrix included in any matrix group according to the inner product of the normalized feature vector corresponding to each protein sequence feature vector respectively included in the antigen distance label vector included in any matrix group comprises:

[0020] Determine the inner product modulus square value corresponding to each protein sequence feature vector in any matrix group according to the modulus square of the inner product of the normalized feature vector corresponding to each protein sequence feature vector respectively included in any matrix group;

[0021] According to the inner product modulus square values ​​corresponding to the respective protein sequence feature vectors, the protein sequence feature vector corresponding to the largest inner product modulus square value is determined as the local key protein sequence feature vector;

[0022] The steps of determining the other protein sequence feature vectors among the protein sequence feature vectors except the local key protein sequence feature vector as new protein sequence feature vectors, returning the modulus square of the inner product of the normalized feature vector corresponding to each protein sequence feature vector respectively with the antigen distance label vector included in any matrix group, and determining the modulus square value of the inner product corresponding to each protein sequence feature vector in any matrix group.

[0023] In one embodiment, before acquiring the training sample set, the method further includes:

[0024] Acquire multiple protein sequences, generate similarity matrices corresponding to the multiple protein sequences by multiple sequence alignment, and determine the similarity matrix as a protein sequence feature matrix; the similarity matrix includes similarities between any two protein sequences in the multiple protein sequences;

[0025] Obtaining a hemagglutination inhibition reaction value matrix corresponding to the multiple protein sequences; the hemagglutination inhibition reaction value matrix includes reaction values ​​of the multiple protein sequences interacting with multiple sera respectively;

[0026] The antigen distance label vector corresponding to the protein sequence feature matrix is ​​determined according to the maximum value of the reaction value in the hemagglutination inhibition reaction value matrix and the ratio between the maximum value and each of the reaction values.

[0027] In a second aspect, the present application also provides an influenza antigen prediction model processing device, comprising:

[0028] An acquisition module is used to acquire a training sample set; the training sample set includes a plurality of matrix groups, the matrix groups include a protein sequence feature matrix and an antigen distance label vector corresponding to the protein sequence feature matrix; the protein sequence feature matrix includes a plurality of protein sequence feature vectors;

[0029] A weight determination module, used to determine a local weight according to a local key protein sequence feature vector in the protein sequence feature matrix included in each of the matrix groups;

[0030] The weight determination module is further used to connect the protein sequence feature matrices included in each of the matrix groups in series to obtain a global protein sequence feature matrix, and determine the global weight according to the global key protein sequence feature vector in the global protein sequence feature matrix;

[0031] The model determination module is used to construct a preset influenza antigen prediction model according to the local weights and the global weights; the preset influenza antigen prediction model is used to output the antigen prediction result of the protein sequence feature matrix to be detected.

[0032] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0033] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0034] In a fifth aspect, the present application also provides a computer program product, including a computer program, which implements the steps of the above method when executed by a processor.

[0035] The above-mentioned influenza antigen prediction model processing method, device, equipment and readable storage medium obtain a training sample set, wherein the training sample set includes multiple matrix groups, the matrix groups include a protein sequence feature matrix composed of multiple protein sequence feature vectors and an antigen distance label vector corresponding to the protein sequence feature matrix; determine local weights according to local key protein sequence feature vectors in the protein sequence feature matrix included in each matrix group; connect the protein sequence feature matrices included in each matrix group in series to obtain a global protein sequence feature matrix, and determine global weights according to the global key protein sequence feature vectors in the global protein sequence feature matrix; construct a preset influenza antigen prediction model according to the local weights and the global weights, and the preset influenza antigen prediction model is used to output the antigen prediction result of the protein sequence feature matrix to be detected, and can determine local weights according to each matrix group in the training sample set, and determine the overall global weight according to each matrix group after series connection, so that the influenza antigen prediction model can adapt to local changes in the data through local weights, and can also achieve a global perspective on the data through global weights. The combination of local weights and global weights can improve the adaptability and generalization ability of the influenza antigen prediction model, and improve the robustness of the influenza antigen prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0037] FIG1 is an application environment diagram of an influenza antigen prediction model processing method in one embodiment;

[0038] FIG2 is a schematic flow chart of a method for processing an influenza antigen prediction model in one embodiment;

[0039] FIG3 is a schematic diagram of performance evaluation of an influenza antigen prediction model in one embodiment;

[0040] FIG4 is a schematic flow chart of another influenza antigen prediction model processing method according to an embodiment;

[0041] FIG5 is a structural block diagram of an influenza antigen prediction model processing device in one embodiment;

[0042] FIG. 6 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0044] The influenza antigen prediction model processing method provided in the embodiment of the present application can be applied to the application environment shown in Figure 1. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The terminal 102 obtains a training sample set; the training sample set includes a plurality of matrix groups, the matrix group includes a protein sequence feature matrix and an antigen distance label vector corresponding to the protein sequence feature matrix; the protein sequence feature matrix includes a plurality of protein sequence feature vectors; the terminal 102 determines the local weight according to the local key protein sequence feature vectors in the protein sequence feature matrix included in each matrix group; the terminal 102 connects the protein sequence feature matrices included in each matrix group in series to obtain a global protein sequence feature matrix, and determines the global weight according to the global key protein sequence feature vectors in the global protein sequence feature matrix; the terminal 102 constructs a preset influenza antigen prediction model based on the local weight and the global weight; the preset influenza antigen prediction model is used to output the antigen prediction result of the protein sequence feature matrix to be detected. The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car devices, projection devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. Head-mounted devices may be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server 104 may be an independent physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides cloud computing services.

[0045] In an exemplary embodiment, as shown in FIG2 , a method for processing an influenza antigen prediction model is provided, which is described by taking the method applied to the terminal 102 in FIG1 as an example, and includes:

[0046] Step S202, obtaining a training sample set.

[0047] The training sample set includes multiple matrix groups, the matrix groups include a protein sequence feature matrix and an antigen distance label vector corresponding to the protein sequence feature matrix; the protein sequence feature matrix includes multiple protein sequence feature vectors.

[0048] In the specific implementation, the value measured by the hemagglutination inhibition test (HI) of the influenza virus can be obtained, which is referred to as the HI titer value. The HI titer value is used to measure the hemagglutination inhibition test reaction value of different viruses and each serum. Each HI titer value can correspond to a virus-antiserum pair. The virus can refer to a circulating or tested virus isolate, and the antiserum can be produced based on a certain organism after infection. The antigenic difference of each virus-antiserum pair can be calculated by the HI titer value. The antigenic difference can be defined as the 2-fold dilution difference between the homologous titer value and the heterologous titer value. The antigenic difference can be used as a quantitative indicator to reflect the change in the antigenicity of the virus. The HI titer value can be used to calculate the antigenic distance between each virus to generate an antigenic distance label vector in the training sample set.

[0049] For each virus-antiserum pair, the amino acid sequence of the hemagglutinin (HA) protein corresponding to each virus (abbreviated as "HA sequence") can be obtained. Hemagglutinin is an important glycoprotein on the surface of influenza virus and one of the main antigens. It is often used in influenza vaccine development and antigen prediction. The amino acid sequence of the HA protein of each virus-antiserum pair can be used to generate the protein sequence feature matrix in the training sample set.

[0050] In an exemplary embodiment, before obtaining the training sample set, it also includes: obtaining multiple protein sequences, generating similarity matrices corresponding to the multiple protein sequences through multiple sequence alignment, and determining the similarity matrix as a protein sequence feature matrix; the similarity matrix includes the similarities between any two protein sequences in the multiple protein sequences; obtaining a hemagglutination inhibition reaction value matrix corresponding to the multiple protein sequences; the hemagglutination inhibition reaction value matrix includes reaction values ​​of multiple protein sequences interacting with multiple sera respectively; and determining the antigen distance label vector corresponding to the protein sequence feature matrix based on the maximum value of the reaction value in the hemagglutination inhibition reaction value matrix and the ratio between the maximum value and each reaction value.

[0051] Alternatively, the protein sequence may be the HA sequence described above.

[0052] Optionally, similarity matrices corresponding to multiple protein sequences are generated by multiple sequence alignment. The similarity matrix X is generated by aligning the HA sequences using a pattern-induced multiple sequence alignment (PIMA coding) function. The similarity matrix includes the similarities between any two protein sequences in the multiple protein sequences, wherein each element X[i][j] represents the similarity between the i-th and j-th protein sequences, and the similarity matrix X is used as the protein sequence feature matrix X in the training sample set.

[0053] The hemagglutination inhibition reaction value matrix includes reaction values ​​of multiple protein sequences interacting with multiple sera respectively, and may be the above-mentioned HI titer value.

[0054] In the specific implementation, the antigen distance label vector corresponding to the protein sequence feature matrix is ​​determined according to the maximum value of the reaction value in the hemagglutination inhibition reaction value matrix and the ratio between the maximum value and each reaction value. First, the matrix of the hemagglutination inhibition test reaction values ​​of different viruses and each serum can be completed by the low-rank matrix completion method to complete the missing data and low reaction values, and then the antigen distance vector Y between each virus is calculated by the following formula.

[0055] ;

[0056] Where Y is the antigen distance vector, is a matrix of the hemagglutination inhibition test reaction values ​​(HI titer values) of different viruses and each serum; this formula can be used to determine the antigen distance label vector in the training sample set.

[0057] Optionally, a portion containing antigenic epitopes may be cut out from the HA protein sequence as a protein sequence feature matrix in the training sample set. For example, the protein sequence feature matrix is ​​determined based on the HA1 subunit in the HA sequence, with the signal peptide in the HA1 subunit fragment as the starting position and the cleavage site between the HA1 subunit and the HA2 subunit as the ending position, to determine the sequence fragment cut out from the HA1 subunit, because the HA1 subunit forms the globular head of the HA protein sequence and may contain antigenically important key epitopes of the influenza virus.

[0058] Step S204, determining local weights according to local key protein sequence feature vectors in the protein sequence feature matrices included in each matrix group.

[0059] The local weight may be a weight determined according to the local key protein sequence feature vectors of each matrix group and used to reflect the local change of the feature.

[0060] The local key protein sequence feature vector is the protein sequence feature vector with the highest correlation with the antigen distance label vector in the protein sequence feature matrix, and is the most representative protein sequence feature vector for capturing antigen evolution information.

[0061] Optionally, the local key protein sequence feature vector can be determined by evaluating the contribution of each protein sequence feature vector to antigen prediction, such as information gain, entropy and other indicators; or, the correlation between each protein sequence feature vector and the antigen distance label vector is calculated, and the protein sequence feature vector with the highest correlation is selected to determine it as the local key protein sequence feature vector.

[0062] Optionally, the local weight is determined based on the local key protein sequence feature vectors in the protein sequence feature matrices included in each matrix group. The local weight can be determined based on the correlation between the local key protein sequence feature vectors included in each matrix group and the antigen distance label vector. For example, the local weight can be determined based on the inner product of the local key protein sequence feature vectors included in each matrix group and the antigen distance label vector.

[0063] Step S206, concatenating the protein sequence feature matrices included in each matrix group to obtain a global protein sequence feature matrix, and determining a global weight according to the global key protein sequence feature vector in the global protein sequence feature matrix.

[0064] In a specific implementation, before obtaining the global key protein sequence feature vector in the global protein sequence feature matrix, the antigen distance label vectors included in each matrix group may be concatenated to obtain a global antigen distance label vector.

[0065] Among them, the global key protein sequence feature vector is the protein sequence feature vector with the highest correlation with the global antigen distance label vector in the global protein sequence feature matrix, and is the most representative protein sequence feature vector for capturing antigen evolution information.

[0066] The global weight may be a weight determined according to the entire training sample set and used to reflect the global change of the feature.

[0067] In a specific implementation, the protein sequence feature matrices included in each matrix group can be connected in series by columns to form a global protein sequence feature matrix containing complete sample feature information.

[0068] The global key protein sequence feature vector is the protein sequence feature vector with the highest correlation with the antigen distance label vector in the global protein sequence feature matrix, and is the most representative protein sequence feature vector for capturing antigen evolution information.

[0069] Optionally, the global key protein sequence feature vector can be determined by evaluating the contribution of each protein sequence feature vector in the global protein sequence feature matrix to antigen prediction, such as information gain, entropy and other indicators; or, the correlation between each protein sequence feature vector in the global protein sequence feature matrix and the antigen distance label vector is calculated, and the protein sequence feature vector with the highest correlation is selected to determine it as the global key protein sequence feature vector.

[0070] Optionally, the global weight is determined according to the global key protein sequence feature vector in the global protein sequence feature matrix. The global weight may be determined by an optimization algorithm such as gradient descent or genetic algorithm.

[0071] Optionally, the global weight is determined based on the global key protein sequence feature vector in the global protein sequence feature matrix. The global weight can be determined based on the correlation between the global key protein sequence feature vector in the global protein sequence feature matrix and the antigen distance label vector. For example, the global weight can be determined based on the inner product of the second key protein sequence feature in the global protein sequence feature matrix and the antigen distance label vector.

[0072] Step S208: construct a preset influenza antigen prediction model based on the local weights and the global weights.

[0073] Among them, the preset influenza antigen prediction model is used to output the antigen prediction results of the protein sequence feature matrix to be detected.

[0074] Optionally, the local weight and the global weight may be fused to obtain a comprehensive weight, such as taking a weighted average of the local weight and the global weight to obtain a comprehensive weight, and a preset influenza antigen prediction model is constructed based on the comprehensive weight.

[0075] Optionally, the influenza antigen prediction model may be a machine learning model such as a random forest model, without specific limitation.

[0076] Among them, the preset influenza antigen prediction model is used to output the antigen prediction results of the protein sequence feature matrix to be detected. Optionally, the protein sequence feature matrix to be detected can be a protein sequence feature matrix determined according to the HA protein sequence of the virus to be detected, and the antigen prediction results can be used to output the key protein sequence feature vector of the protein sequence feature matrix to be detected and the antigen distance vector between each virus. The key protein sequence feature vector can be used to reflect the key points related to antigenicity in the virus HA protein sequence, that is, the important amino acid positions in the virus HA protein sequence; the antigen distance can be used to reflect the antigenic differences between different viruses, and to identify the progressive changes (antigenic drift) and mutations (antigenic shift) of the virus antigenicity, that is, the antigenic characteristics of the virus HA protein, which are important for predicting the trend of virus antigen variation and selecting appropriate vaccine strains.

[0077] In the above-mentioned influenza antigen prediction model processing method, a training sample set is obtained, wherein the training sample set includes multiple matrix groups, the matrix groups include a protein sequence feature matrix composed of multiple protein sequence feature vectors and an antigen distance label vector corresponding to the protein sequence feature matrix; local weights are determined according to local key protein sequence feature vectors in the protein sequence feature matrix included in each matrix group; the protein sequence feature matrices included in each matrix group are connected in series to obtain a global protein sequence feature matrix, and global weights are determined according to the global key protein sequence feature vectors in the global protein sequence feature matrix; according to the local weights and the global weights, a preset influenza antigen prediction model is constructed, and the preset influenza antigen prediction model is used to output the antigen prediction result of the protein sequence feature matrix to be detected, and the local weights can be determined according to each matrix group in the training sample set, and the overall global weights can be determined according to each matrix group after the series connection, so that the influenza antigen prediction model can adapt to local changes in the data through local weights, and can also achieve a global perspective on the data through global weights. The combination of local weights and global weights can improve the adaptability and generalization ability of the influenza antigen prediction model, and improve the robustness of the influenza antigen prediction model.

[0078] In an exemplary embodiment, local weights are determined based on local key protein sequence feature vectors in the protein sequence feature matrix included in each matrix group, including: for any matrix group in each matrix group, based on the inner product value of the local key protein sequence feature vector in the protein sequence feature matrix included in any matrix group and the antigen distance label vector, determining the local sub-weight corresponding to any matrix group; determining the local weight based on the local sub-weights corresponding to each matrix group.

[0079] The inner product value can be used to reflect the correlation between the local key protein sequence feature vector and the antigen distance label vector. In other words, it is the degree of influence of the local key protein sequence feature vector on the antigen distance label vector. In the machine learning model, weights can be used to measure the importance of feature vectors. The larger the weight, the greater the influence of the feature vector on the prediction result. Therefore, the local weight can be determined based on the inner product value. For example, the larger the inner product value, the larger the determined local weight. Determining the local weight through the inner product value allows the influenza antigen prediction model to be unrestricted by outliers during the training process.

[0080] In a specific implementation, the inner product value of the local key protein sequence feature vector and the antigen distance label vector in the protein sequence feature matrix included in any matrix group can be used as the local sub-weight corresponding to any matrix group. The local sub-weights corresponding to each matrix group can be combined into a set, and the set of local sub-weights can be used as the local weight.

[0081] In an exemplary embodiment, before determining the local weights based on the local key protein sequence feature vectors in the protein sequence feature matrix included in each matrix group, it also includes: for any matrix group in each matrix group, determining the normalized feature vector corresponding to each protein sequence feature vector included in any matrix group; determining the local key protein sequence feature vector in the protein sequence feature matrix included in any matrix group based on the inner product of the antigen distance label vector included in any matrix group and the normalized feature vector corresponding to each protein sequence feature vector.

[0082] In a specific implementation, any protein sequence feature vector can be normalized to obtain a normalized feature vector corresponding to the protein sequence feature vector, so that each feature vector is within the same range, thereby improving the accuracy of subsequent calculation of the inner product.

[0083] Among them, the local key protein sequence feature vector is the protein sequence feature vector with the highest correlation with the antigen distance label vector in the protein sequence feature matrix, and is the most representative protein sequence feature vector for capturing antigen evolution information.

[0084] For any protein sequence feature vector, the inner product of the antigen distance label vector and the normalized feature vector corresponding to the protein sequence feature vector is calculated. The inner product calculation can be used to measure the correlation between the two vectors. Therefore, based on the calculated inner product, the protein sequence feature vector with the highest correlation with the antigen distance label vector can be determined from the various protein sequence feature vectors included in any matrix group. As the key protein sequence feature vectors, these key protein sequence feature vectors can capture the complex patterns of antigen evolution and are the most representative.

[0085] Further, in an exemplary embodiment, the local key protein sequence feature vectors in the protein sequence feature matrix included in any matrix group are determined according to the inner products of the normalized feature vectors corresponding to the antigen distance label vectors included in any matrix group and the protein sequence feature vectors, including: determining the inner product modulus square value corresponding to each protein sequence feature vector in any matrix group according to the modulus square of the inner product of the normalized feature vectors corresponding to the antigen distance label vectors included in any matrix group and the protein sequence feature vectors; determining the protein sequence feature vector corresponding to the largest inner product modulus square value as the local key protein sequence feature vector according to the inner product modulus square value corresponding to each protein sequence feature vector; determining the other protein sequence feature vectors in each protein sequence feature vector except the local key protein sequence feature vector as new protein sequence feature vectors, and returning the step of determining the inner product modulus square value corresponding to each protein sequence feature vector in any matrix group according to the modulus square of the inner product of the normalized feature vectors corresponding to the antigen distance label vectors included in any matrix group and the protein sequence feature vectors.

[0086] Assume there are n matrix groups (X k , Y k ), k=1,…,n. Among them, X k is m k The matrix with l rows and columns, X k is the protein sequence feature matrix; Y k The length is m k Column vector, Y k is the antigen distance label vector. For any matrix group X k Expressed as .

[0087] Furthermore, in an exemplary embodiment, before determining the modulus square value of the inner product corresponding to each protein sequence feature vector in any matrix group based on the modulus square of the inner product of the normalized feature vector corresponding to each protein sequence feature vector included in any matrix group, it also includes: when determining the first local key protein sequence feature vector, dividing any protein sequence feature vector in any matrix group by its corresponding norm to obtain a normalized feature vector corresponding to any protein sequence feature vector.

[0088] In the specific implementation, the protein sequence feature matrix X k Each column vector (protein sequence feature vector) in is divided by its corresponding Euclidean norm, that is, each protein sequence feature vector is unit orthogonalized to obtain the normalized feature vector corresponding to each protein sequence feature vector.

[0089] Next, we can establish the protein sequence feature matrix X based on the maximum selection principle. k Distance label vector Y with antigen k The model relationship between them, the process of establishing the influenza antigen prediction model is as follows:

[0090] According to the boundary zeroing property and the maximum selection principle, we have the following formula (1):

[0091] ; (1)

[0092] Among them, Y k is the antigen distance label vector, is any protein sequence feature vector in the protein sequence feature matrix, j = 1, 2, ..., l; is the norm of any protein sequence feature vector; is the normalized feature vector corresponding to any protein sequence feature vector. Any protein sequence feature vector is normalized, that is, divided by its norm to obtain a normalized feature vector; is the inner product of the normalized feature vector corresponding to the antigen distance label vector and any protein sequence feature vector; It is the square of the modulus of the inner product of the normalized feature vector corresponding to the antigen distance label vector and any protein sequence feature vector, that is, the square of the modulus of the inner product corresponding to any protein sequence feature vector.

[0093] According to the zero-approaching property of the boundary, the maximum value of the square of the inner product modulus is determined, that is, at X k Find the protein sequence feature vector corresponding to j=1, 2, ..., l in Y k The inner product modulus square value of the normalized feature vector corresponding to the protein sequence feature vector reaches the maximum. I1 can be used to represent the protein sequence feature vector corresponding to the maximum inner product modulus square value in X k The number of columns in X, that is, the first local key protein sequence feature vector in X k Therefore, the I1 site is one of the key sites, that is, one of the key sites related to antigenicity in the viral HA protein sequence. It is the protein sequence feature vector corresponding to the largest inner product modulus square value, that is, the first local key protein sequence feature vector determined.

[0094] The first local key protein sequence feature vector can be normalized to obtain the normalized feature vector corresponding to the first local key protein sequence feature vector .

[0095] The first local sub-weight corresponding to any matrix group is , which is equal to the inner product value of the first local key protein sequence feature vector and the antigen distance label vector.

[0096] Can Exclude the multiple protein sequence feature vectors corresponding to any matrix group, use the remaining other protein sequence feature vectors as new multiple protein sequence feature vectors, and determine the next local key protein sequence feature vector from the new multiple protein sequence feature vectors. Similarly, multiple local key protein sequence feature vectors can be determined, as well as multiple local sub-weights corresponding to any matrix group.

[0097] Further, in an exemplary embodiment, when determining the Nth local key protein sequence feature vector, the projection vector of any protein sequence feature vector on the last determined local key protein sequence feature vector is subtracted from any protein sequence feature vector in any matrix group to obtain a normalized feature vector corresponding to any protein sequence feature vector; N is an integer greater than 1; and the projection vector is equal to the product of the inner product value of any protein sequence feature vector and the last determined local key protein sequence feature vector and the last determined local key protein sequence feature vector.

[0098] In a specific implementation, the protein sequence feature vectors in the new multiple protein sequence feature vectors in any matrix pair are unit orthogonalized to obtain the normalized feature vectors corresponding to each protein sequence feature vector.

[0099] When the second local key protein sequence feature vector is determined, the normalized feature vector corresponding to any protein sequence feature vector can be expressed as: .visible. is any protein sequence feature vector among multiple new protein sequence feature vectors; for The projection vector on the last determined local key protein sequence feature vector (i.e., the first determined local key protein sequence feature vector) u1; Subtract the projection vector of any protein sequence feature vector on the last determined key protein sequence feature vector from any protein sequence feature vector, and then Divide by its corresponding norm to obtain the normalized feature vector corresponding to any protein sequence feature vector. Here, It is an orthogonal process, ensuring the key sites This pre-orthogonalization process can eliminate redundancy and noise in the data, allowing the influenza antigen prediction model to focus more on extracting key features and relationships. Using the idea of ​​orthogonal projection for feature selection ensures that the selected features have the maximum amount of information in the orthogonal space, further improving the prediction accuracy.

[0100] The maximum selection principle is then used to continue determining the second local key protein sequence feature vector.

[0101] As shown in formula (2):

[0102] ; (2)

[0103] Among them, the I2 site is the second key site selected. is the second key protein sequence feature vector identified.

[0104] The normalized feature vector corresponding to the second key protein sequence feature vector can be obtained The second local sub-weight corresponding to any matrix group is , which is equal to the inner product value of the second local key protein sequence feature vector and the antigen distance label vector. By analogy, multiple local sub-weights corresponding to any matrix group can be obtained.

[0105] And so on. ,u j is the normalized feature vector corresponding to the jth local key protein sequence feature vector.

[0106] In an exemplary embodiment, after the protein sequence feature vector corresponding to the largest inner product modulus square value is determined as the local key protein sequence feature vector, it also includes: judging whether the largest inner product modulus square value is less than or equal to the convergence threshold; if the largest inner product modulus square value is less than or equal to the convergence threshold, returning to the step of determining the other protein sequence feature vectors among the protein sequence feature vectors except the local key protein sequence feature vector as new protein sequence feature vectors, until the largest inner product modulus square value is greater than the convergence threshold.

[0107] Pick , such that there exists an integer satisfy: . For the maximum selection principle termination symbol, you can choose , so that , is the convergence threshold, that is, each time a local key protein sequence feature vector is determined, if the inner product modulus square value corresponding to the local key protein sequence feature vector is greater than When , no key sites are selected, that is, no local key protein sequence feature vectors are selected, and the above cycle process ends. In summary, the convergence level after adaptive maximum selection can be obtained .

[0108] By setting the convergence threshold, the influenza antigen prediction model has great adaptability and can accurately screen out features that are highly correlated with the target variable (antigen distance label vector Y), thereby avoiding unnecessary feature redundancy and effectively preventing overfitting. The adaptability in the above process is reflected in: , by setting the minimum convergence threshold This ensures that key points are selected at each step to avoid overfitting and to better characterize the relationship between features and target variables.

[0109] In summary, through the principle of maximum selection, key protein sequence feature vectors can be selected to effectively avoid the overfitting problem, and the adaptability can be improved through the convergence threshold, thereby more accurately predicting the antigen evolution trend.

[0110] Suppose that we determine from any of the n matrix groups local key protein sequence feature vectors, and local sub-weights, then the local sub-weight corresponding to any matrix group is , and the local key protein sequence feature vector corresponding to any matrix group is .

[0111] Traverse n matrix groups (X k , Y k), k=1,…,n, we can get the local weights composed of the local sub-weights corresponding to each matrix group: . And, the local key protein sequence feature vectors corresponding to each matrix group can be obtained: The union L of the local key protein sequence feature vectors corresponding to each matrix group can be expressed as: .

[0112] In an exemplary embodiment, local weights are determined based on local sub-weights corresponding to each matrix group, including: constructing a local key protein sequence feature matrix based on the local key protein sequence feature vectors in the protein sequence feature matrix included in any matrix group; determining a transformation matrix based on the inverse matrix of the local key protein sequence feature matrix and the normalized feature vector corresponding to the local key protein sequence feature vector; determining local weights based on the local sub-weights corresponding to each matrix group and the transformation matrix.

[0113] In the specific implementation, for any matrix group, the local sub-weight corresponding to the matrix group is , n=1, the local key protein sequence feature matrix constructed by the local key protein sequence feature vector included in the matrix group is , the normalized eigenvector U corresponding to each local key protein sequence eigenvector in the matrix group is (U1,…,U n ), the transformation matrix , get the local weight corresponding to the matrix group According to the above method, the local weights corresponding to each matrix group can be obtained , if there are n matrix groups, then there are also n local weights.

[0114] In summary, by introducing local weights and global weights and assigning different weights to each training sample and the overall task, we can achieve accurate evaluation and local optimization of feature importance, which not only enhances the generalization ability of the model, but also makes it more robust in the face of local changes. This brings new solutions to the field of influenza virus antigen evolution prediction and significantly improves the overall performance of the model.

[0115] It should be noted that the method for determining the global key protein sequence feature vector in the global protein sequence feature matrix can refer to the method for determining the local key protein sequence feature vector in the protein sequence feature matrix mentioned above.

[0116] In an exemplary embodiment, before determining the global weight according to the global key protein sequence feature vector in the global protein sequence feature matrix, the method further includes:

[0117] The antigen distance label vectors included in each matrix group are concatenated to obtain a global antigen distance label vector; the normalized feature vectors corresponding to each protein sequence feature vector included in the global protein sequence feature matrix are determined; and the global key protein sequence feature vector in the global protein sequence feature matrix is ​​determined according to the inner product of the global antigen distance label and the normalized feature vector corresponding to each protein sequence feature vector included in the global protein sequence feature matrix.

[0118] Further, in an exemplary embodiment, the global key protein sequence feature vector in the global protein sequence feature matrix is ​​determined according to the inner product of the normalized feature vector corresponding to each protein sequence feature vector included in the global protein sequence feature matrix and the global antigen distance label, including:

[0119] According to the modulus square of the inner product of the normalized feature vector corresponding to each protein sequence feature vector included in the global protein sequence feature matrix and the global antigen distance label vector, determine the modulus square of the inner product corresponding to each protein sequence feature vector in the global protein sequence feature matrix;

[0120] According to the inner product modulus square values ​​corresponding to each protein sequence feature vector in the global protein sequence feature matrix, the protein sequence feature vector corresponding to the largest inner product modulus square value is determined as the global key protein sequence feature vector;

[0121] The steps include determining a new global protein sequence feature matrix based on the protein sequence feature vectors other than the global key protein sequence feature vector in the protein sequence feature matrix, returning the modulus square of the inner product of the normalized feature vector corresponding to the global antigen distance label vector and the protein sequence feature vector included in the global protein sequence feature matrix, and determining the modulus square of the inner product corresponding to each protein sequence feature vector in the global protein sequence feature matrix.

[0122] It should be noted that the method for determining the global weight based on the global key protein sequence feature vector in the global protein sequence feature matrix can refer to the method for determining the local weight based on the local key protein sequence feature vector in the protein sequence feature matrix included in each matrix group.

[0123] In an exemplary embodiment, a global weight is determined based on a global key protein sequence feature vector in a global protein sequence feature matrix, including: determining a global sub-weight corresponding to the global protein sequence feature matrix based on an inner product value of the global key protein sequence feature vector in the global protein sequence feature matrix and an antigen distance label vector; constructing a global key protein sequence feature matrix based on the global key protein sequence feature vector in the global protein sequence feature matrix; determining a transformation matrix based on an inverse matrix of the global key protein sequence feature matrix and a normalized feature vector corresponding to the global key protein sequence feature vector; determining a global weight based on the global sub-weight corresponding to the global protein sequence feature matrix and the transformation matrix.

[0124] In the specific implementation, X in n sets of matrix pairs k Do column concatenation, and we have new (m1, ..., m n ) is a matrix with 1 rows and 1 columns, namely the global protein sequence feature matrix, denoted as The same method can be used to generate a new antigen distance label vector, that is, Y in n sets of matrix pairs k Concatenate and get the global antigen distance label direction, recorded as The method of determining the global weight according to the global key protein sequence feature vector in the global protein sequence feature matrix can refer to the method of determining the local weight according to the local key protein sequence feature vector in the protein sequence feature matrix included in each matrix group, and the specific details are not repeated here. The final global weight is w g .

[0125] In an exemplary embodiment, a preset influenza antigen prediction model is constructed based on the local weight and the global weight. The prediction function for predicting the antigen distance in the preset influenza antigen prediction model may be:

[0126] ;

[0127] in, is the antigen distance prediction vector output by the preset influenza antigen prediction model, X can be the feature matrix of the protein sequence to be detected, For custom parameters, is the global weight, As local weights, the above method can efficiently establish the model relationship between features and targets.

[0128] In summary, by adding two important weight parameters global weight and local weight The local weights assign different weights to sample data to accurately capture local features, and the global weights ensure the global stability of the influenza antigen prediction model. The combination of the two improves the adaptability and generalization ability of the influenza antigen prediction model, reduces the risk of overfitting, and enhances interpretability. It can also maintain excellent model performance when analyzing complex data to be tested (such as the feature matrix of the protein sequence to be tested).

[0129] After the preset influenza antigen prediction model is determined, the preset influenza antigen prediction model can be tested using a test sample set to obtain a test result of the preset influenza antigen prediction model, which can be used to evaluate the performance of the preset influenza antigen prediction model.

[0130] The test sample set is used to evaluate the performance of the influenza antigen prediction model on unseen data and provide a measure of the predictive ability of the influenza antigen prediction model in practical applications. The test results may include accuracy, recall, mean square error, etc. These test results can be used to evaluate the generalization ability, accuracy, robustness, overfitting or underfitting of the influenza antigen prediction model.

[0131] Assume that the protein sequence feature matrix included in the test sample set is , then according to the model prediction parameter w and the protein sequence feature matrix included in the test sample set , determine the antigen distance prediction vector as , so the test sample set includes the antigen distance label vector Distance prediction vector The difference between (test set error), and obtain multiple test results of the preset influenza antigen prediction model until Meet the threshold condition and average multiple test results.

[0132] The preset influenza antigen prediction model is used to output the antigen prediction results of the protein sequence feature matrix to be detected. Assume that the protein sequence feature matrix to be predicted is ,pass A predicted antigenic distance Y can be obtained t , and important amino acid positions (key positions) .

[0133] To facilitate the understanding of those skilled in the art, FIG3 exemplarily provides a schematic diagram of the performance evaluation of an influenza antigen prediction model. The preset influenza antigen prediction model is tested by a test sample set to obtain the test results of the preset influenza antigen prediction model, thereby evaluating the model performance of the influenza antigen prediction model, and comparing and analyzing the preset influenza antigen prediction model with three machine learning algorithms: Ridge regression, least absolute shrinkage and selection operator (LASSO), and elastic network regression (ENR). The correlation coefficient (CORR), root mean square error (RMSE), precision (Precision) and F1 score (F1-Score) are used to evaluate the model performance. CORR (correlation coefficient) quantifies the degree of linear correlation between the model prediction value and the actual observation value, which not only reveals the strength of the correlation between the predicted value and the true value, but also points out the directional relationship between them; RMSE (root mean square error) is used to evaluate the degree of difference between the predicted value and the actual value. The lower the RMSE, the closer the prediction result of the model is to the true value, indicating that the model has a stronger predictive ability. In the evaluation of specific tasks (such as identifying the similarity of two influenza virus antigens), the Precision metric is introduced. Precision focuses on the proportion of samples that are actually positive when the model identifies similar antigens, helping to understand the accuracy of the model in specific tasks. Finally, in order to more comprehensively evaluate the performance of the model in identifying antigens, the F1 value metric is used. F1-Score is the harmonic mean of Precision and Recall, which not only considers the accuracy of the model in identifying positive examples, but also the model's ability to find all positive examples.

[0134] As shown in Figure 3, compared with other machine learning models, the root mean square error RMSE value of the preset influenza antigen prediction model is the lowest, while the precision and F1 scores are above 0.9, and the correlation coefficient is 0.74. This shows that the influenza antigen prediction model has a better performance in predicting the antigenicity of influenza viruses than other machine learning models.

[0135] The above method introduces local weights and global weights, which can assign different weights to each matrix group through local weights, so as to capture the importance of local features in the data, so that the influenza antigen prediction model can more finely adjust its parameters to adapt to local changes in the data, thereby improving the prediction performance in specific situations. The global weight provides a global perspective for the entire influenza antigen prediction model. It ensures the consistency and stability of the model as a whole, thus helping to reduce the risk of overfitting. The combination of local weights and global weights not only improves the adaptability and generalization ability of the model, but also enables the model to capture the characteristics of the data more comprehensively by balancing the local and global perspectives.

[0136] FIG. 4 is another influenza antigen prediction model processing method according to an exemplary embodiment. As shown in FIG. 4 , the method is applied to the terminal 102 in FIG. 1 as an example for explanation, and includes the following steps:

[0137] Step S402: obtaining multiple protein sequences, generating similarity matrices corresponding to the multiple protein sequences through multiple sequence alignment, and determining the similarity matrix as a protein sequence feature matrix.

[0138] The similarity matrix includes the similarities between two protein sequences in multiple protein sequences.

[0139] Step S404, obtaining a hemagglutination inhibition response value matrix corresponding to a plurality of protein sequences.

[0140] The hemagglutination inhibition response value matrix includes response values ​​of multiple protein sequences interacting with multiple sera respectively.

[0141] Step S406, determining the antigen distance label vector corresponding to the protein sequence feature matrix according to the maximum value of the reaction value in the hemagglutination inhibition reaction value matrix and the ratio between the maximum value and each reaction value.

[0142] Step S408, obtaining a training sample set.

[0143] The training sample set includes multiple matrix groups, the matrix groups include a protein sequence feature matrix and an antigen distance label vector corresponding to the protein sequence feature matrix; the protein sequence feature matrix includes multiple protein sequence feature vectors.

[0144] Step S410, for any matrix group in each matrix group, determine the local sub-weight corresponding to any matrix group according to the inner product value of the local key protein sequence feature vector in the protein sequence feature matrix included in any matrix group and the antigen distance label vector.

[0145] Step S412, constructing a local key protein sequence feature matrix based on the local key protein sequence feature vectors in the protein sequence feature matrix included in any matrix group; determining a transformation matrix based on the inverse matrix of the local key protein sequence feature matrix and the normalized feature vectors corresponding to the local key protein sequence feature vectors; determining local weights based on the local sub-weights corresponding to each matrix group and the transformation matrix.

[0146] In an exemplary embodiment, before determining the local weights based on the local key protein sequence feature vectors in the protein sequence feature matrix included in each matrix group, it also includes: for any matrix group in each matrix group, determining the normalized feature vector corresponding to each protein sequence feature vector included in any matrix group; determining the local key protein sequence feature vector in the protein sequence feature matrix included in any matrix group based on the inner product of the antigen distance label vector included in any matrix group and the normalized feature vector corresponding to each protein sequence feature vector.

[0147] Further, in an exemplary embodiment, the local key protein sequence feature vectors in the protein sequence feature matrix included in any matrix group are determined according to the inner products of the normalized feature vectors corresponding to the antigen distance label vectors included in any matrix group and the protein sequence feature vectors, including: determining the inner product modulus square value corresponding to each protein sequence feature vector in any matrix group according to the modulus square of the inner product of the normalized feature vectors corresponding to the antigen distance label vectors included in any matrix group and the protein sequence feature vectors; determining the protein sequence feature vector corresponding to the largest inner product modulus square value as the local key protein sequence feature vector according to the inner product modulus square value corresponding to each protein sequence feature vector; determining the other protein sequence feature vectors in each protein sequence feature vector except the local key protein sequence feature vector as new protein sequence feature vectors, and returning the step of determining the inner product modulus square value corresponding to each protein sequence feature vector in any matrix group according to the modulus square of the inner product of the normalized feature vectors corresponding to the antigen distance label vectors included in any matrix group and the protein sequence feature vectors.

[0148] Step S414, concatenate the protein sequence feature matrices included in each matrix group to obtain a global protein sequence feature matrix, and determine the global sub-weight corresponding to the global protein sequence feature matrix according to the inner product value of the global key protein sequence feature vector and the antigen distance label vector in the global protein sequence feature matrix.

[0149] In an exemplary embodiment, before determining the global weight based on the global key protein sequence feature vector in the global protein sequence feature matrix, it also includes: concatenating the antigen distance label vectors included in each matrix group to obtain a global antigen distance label vector; determining the normalized feature vector corresponding to each protein sequence feature vector included in the global protein sequence feature matrix; determining the global key protein sequence feature vector in the global protein sequence feature matrix based on the inner product of the global antigen distance label and the normalized feature vector corresponding to each protein sequence feature vector included in the global protein sequence feature matrix.

[0150] Further, in an exemplary embodiment, the global key protein sequence feature vectors in the global protein sequence feature matrix are determined according to the inner products of the normalized feature vectors corresponding to the global antigen distance labels and the respective protein sequence feature vectors included in the global protein sequence feature matrix, including: determining the inner product modulus square value corresponding to each protein sequence feature vector in the global protein sequence feature matrix according to the modulus square of the inner product of the normalized feature vectors corresponding to the global antigen distance label vectors and the respective protein sequence feature vectors included in the global protein sequence feature matrix; determining the protein sequence feature vector corresponding to the largest inner product modulus square value as the global key protein sequence feature vector according to the inner product modulus square value corresponding to each protein sequence feature vector in the global protein sequence feature matrix; determining a new global protein sequence feature matrix according to the protein sequence feature vectors other than the global key protein sequence feature vector in the protein sequence feature matrix included in the global protein sequence feature matrix, and returning the step of determining the inner product modulus square value corresponding to each protein sequence feature vector in the global protein sequence feature matrix according to the modulus square of the inner product of the normalized feature vectors corresponding to the global antigen distance label vectors and the respective protein sequence feature vectors included in the global protein sequence feature matrix.

[0151] Step S416, constructing a global key protein sequence feature matrix based on the global key protein sequence feature vectors in the global protein sequence feature matrix; determining a transformation matrix based on the inverse matrix of the global key protein sequence feature matrix and the normalized feature vectors corresponding to the global key protein sequence feature vectors; determining a global weight based on the global sub-weights corresponding to the global protein sequence feature matrix and the transformation matrix.

[0152] Step S418, constructing a preset influenza antigen prediction model based on the local weights and the global weights.

[0153] The preset influenza antigen prediction model is used to output the antigen prediction results of the protein sequence feature matrix to be detected.

[0154] It should be noted that the specific limitations of the above steps can be found in the specific limitations of an influenza antigen prediction model processing method mentioned above, which will not be repeated here.

[0155] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0156] Based on the same inventive concept, the embodiment of the present application also provides an influenza antigen prediction model processing device for implementing the above-mentioned influenza antigen prediction model processing method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in the one or more influenza antigen prediction model processing device embodiments provided below can refer to the above limitations on the influenza antigen prediction model processing method, and will not be repeated here.

[0157] In an exemplary embodiment, as shown in FIG5 , a influenza antigen prediction model processing device is provided, comprising:

[0158] The acquisition module 510 is used to acquire a training sample set; the training sample set includes multiple matrix groups, the matrix groups include a protein sequence feature matrix and an antigen distance label vector corresponding to the protein sequence feature matrix; the protein sequence feature matrix includes multiple protein sequence feature vectors.

[0159] The weight determination module 520 is used to determine the local weight according to the local key protein sequence feature vectors in the protein sequence feature matrix included in each of the matrix groups.

[0160] The weight determination module 520 is further used to connect the protein sequence feature matrices included in each of the matrix groups in series to obtain a global protein sequence feature matrix, and determine the global weight according to the global key protein sequence feature vector in the global protein sequence feature matrix.

[0161] The model determination module 530 is used to construct a preset influenza antigen prediction model according to the local weights and the global weights; the preset influenza antigen prediction model is used to output the antigen prediction result of the protein sequence feature matrix to be detected.

[0162] In one embodiment, the weight determination module is also used to determine the local sub-weight corresponding to any matrix group in each of the matrix groups according to the inner product value of the local key protein sequence feature vector in the protein sequence feature matrix included in the any matrix group and the antigen distance label vector; and determine the local weight according to the local sub-weight corresponding to each of the matrix groups.

[0163] In one embodiment, the weight determination module is also used to construct a local key protein sequence feature matrix based on the local key protein sequence feature vectors in the protein sequence feature matrix included in any one of the matrix groups; determine the transformation matrix based on the inverse matrix of the local key protein sequence feature matrix and the normalized feature vectors corresponding to the local key protein sequence feature vectors; determine the local weights based on the local sub-weights corresponding to each of the matrix groups and the transformation matrix.

[0164] In one of the embodiments, the weight determination module is also used to determine, for any matrix group in each of the matrix groups, a normalized feature vector corresponding to each protein sequence feature vector included in the any matrix group; and to determine the local key protein sequence feature vector in the protein sequence feature matrix included in the any matrix group based on the inner product of the antigen distance label vector included in the any matrix group and the normalized feature vector corresponding to each protein sequence feature vector.

[0165] In one embodiment, the weight determination module is also used to determine the inner product modulus square value corresponding to each protein sequence feature vector in any matrix group according to the modulus square of the inner product of the normalized feature vector corresponding to each protein sequence feature vector included in any matrix group and the antigen distance label vector included in any matrix group; determine the protein sequence feature vector corresponding to the largest inner product modulus square value as the local key protein sequence feature vector according to the inner product modulus square value corresponding to each protein sequence feature vector; determine the other protein sequence feature vectors in the protein sequence feature vectors except the local key protein sequence feature vector as new protein sequence feature vectors, and return the step of determining the inner product modulus square value corresponding to each protein sequence feature vector in any matrix group according to the modulus square of the inner product of the normalized feature vector corresponding to each protein sequence feature vector included in any matrix group and the antigen distance label vector included in any matrix group.

[0166] In one embodiment, the acquisition module is also used to obtain multiple protein sequences, generate a similarity matrix corresponding to the multiple protein sequences through multiple sequence alignment, and determine the similarity matrix as a protein sequence feature matrix; the similarity matrix includes the similarity between any two protein sequences in the multiple protein sequences; obtain a hemagglutination inhibition reaction value matrix corresponding to the multiple protein sequences; the hemagglutination inhibition reaction value matrix includes the reaction values ​​of the multiple protein sequences interacting with multiple sera respectively; and determine the antigen distance label vector corresponding to the protein sequence feature matrix according to the maximum value of the reaction value in the hemagglutination inhibition reaction value matrix and the ratio between the maximum value and each of the reaction values.

[0167] Each module in the above influenza antigen prediction model processing device can be implemented in whole or in part by software, hardware, and a combination thereof. Each of the above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each of the above modules.

[0168] In an exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be shown in FIG6. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, the memory, and the input / output interface are connected via a system bus, and the communication interface, the display unit, and the input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, a method for processing an influenza antigen prediction model is implemented. The display unit of the computer device is used to form a visually visible picture, which may be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0169] Those skilled in the art will understand that the structure shown in FIG. 6 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0170] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above-mentioned influenza antigen prediction model processing method. The steps of the influenza antigen prediction model processing method here can be the steps of the influenza antigen prediction model processing method in each of the above-mentioned embodiments.

[0171] In one embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the processor executes the steps of the above-mentioned influenza antigen prediction model processing method. The steps of the influenza antigen prediction model processing method here can be the steps of the influenza antigen prediction model processing method in each of the above-mentioned embodiments.

[0172] In one embodiment, a computer program product is provided, including a computer program, which, when executed by a processor, causes the processor to perform the steps of the above-mentioned influenza antigen prediction model processing method. The steps of the influenza antigen prediction model processing method here can be the steps of the influenza antigen prediction model processing method in each of the above-mentioned embodiments.

[0173] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0174] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0175] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, 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 application.

[0176] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for processing an influenza antigen prediction model, characterized in that: The method comprises: Acquire a training sample set; the training sample set includes a plurality of matrix groups, the matrix groups include a protein sequence feature matrix and an antigen distance label vector corresponding to the protein sequence feature matrix; the protein sequence feature matrix includes a plurality of protein sequence feature vectors; Determining local weights according to local key protein sequence feature vectors in the protein sequence feature matrices included in each of the matrix groups; The protein sequence feature matrices included in each of the matrix groups are connected in series to obtain a global protein sequence feature matrix, and a global weight is determined according to a global key protein sequence feature vector in the global protein sequence feature matrix; A preset influenza antigen prediction model is constructed according to the local weights and the global weights; the preset influenza antigen prediction model is used to output the antigen prediction result of the protein sequence feature matrix to be detected.

2. The method according to claim 1, characterized in that Determining the local weight according to the local key protein sequence feature vectors in the protein sequence feature matrices included in each of the matrix groups includes: For any matrix group in each of the matrix groups, according to the inner product value of the local key protein sequence feature vector in the protein sequence feature matrix included in the any matrix group and the antigen distance label vector, determine the local sub-weight corresponding to the any matrix group; The local weight is determined according to the local sub-weights corresponding to the matrix groups.

3. The method according to claim 2, characterized in that The determining the local weight according to the local sub-weights corresponding to the matrix groups includes: Constructing a local key protein sequence feature matrix according to the local key protein sequence feature vectors in the protein sequence feature matrix included in any one of the matrix groups; Determine a conversion matrix according to the inverse matrix of the local key protein sequence feature matrix and the normalized feature vector corresponding to the local key protein sequence feature vector; The local weight is determined according to the local sub-weights corresponding to each of the matrix groups and the transformation matrix.

4. The method according to claim 1, characterized in that: Before determining the local weights according to the local key protein sequence feature vectors in the protein sequence feature matrices included in each of the matrix groups, the method further comprises: For any matrix group in each of the matrix groups, determining a normalized feature vector corresponding to each protein sequence feature vector included in the any matrix group; The local key protein sequence feature vectors in the protein sequence feature matrix included in any matrix group are determined according to the inner products of the antigen distance label vectors included in any matrix group and the normalized feature vectors corresponding to each protein sequence feature vector.

5. The method according to claim 4, characterized in that Determining the local key protein sequence feature vector in the protein sequence feature matrix included in any matrix group according to the inner product of the normalized feature vector corresponding to each protein sequence feature vector respectively included in the antigen distance label vector included in any matrix group comprises: Determine the inner product modulus square value corresponding to each protein sequence feature vector in any matrix group according to the modulus square of the inner product of the normalized feature vector corresponding to each protein sequence feature vector respectively included in any matrix group; According to the inner product modulus square values ​​corresponding to the respective protein sequence feature vectors, the protein sequence feature vector corresponding to the largest inner product modulus square value is determined as the local key protein sequence feature vector; The steps of determining the other protein sequence feature vectors among the protein sequence feature vectors except the local key protein sequence feature vector as new protein sequence feature vectors, returning the modulus square of the inner product of the normalized feature vector corresponding to each protein sequence feature vector respectively with the antigen distance label vector included in any matrix group, and determining the modulus square value of the inner product corresponding to each protein sequence feature vector in any matrix group.

6. The method according to claim 1, characterized in that Before acquiring the training sample set, the method further includes: Acquire multiple protein sequences, generate similarity matrices corresponding to the multiple protein sequences by multiple sequence alignment, and determine the similarity matrix as a protein sequence feature matrix; the similarity matrix includes similarities between any two protein sequences in the multiple protein sequences; Obtaining a hemagglutination inhibition reaction value matrix corresponding to the multiple protein sequences; the hemagglutination inhibition reaction value matrix includes reaction values ​​of the multiple protein sequences interacting with multiple sera respectively; The antigen distance label vector corresponding to the protein sequence feature matrix is ​​determined according to the maximum value of the reaction value in the hemagglutination inhibition reaction value matrix and the ratio between the maximum value and each of the reaction values.

7. An influenza antigen prediction model processing device, characterized in that: The device comprises: An acquisition module is used to acquire a training sample set; the training sample set includes a plurality of matrix groups, the matrix groups include a protein sequence feature matrix and an antigen distance label vector corresponding to the protein sequence feature matrix; the protein sequence feature matrix includes a plurality of protein sequence feature vectors; A weight determination module, used to determine a local weight according to a local key protein sequence feature vector in the protein sequence feature matrix included in each of the matrix groups; The weight determination module is further used to connect the protein sequence feature matrices included in each of the matrix groups in series to obtain a global protein sequence feature matrix, and determine the global weight according to the global key protein sequence feature vector in the global protein sequence feature matrix; The model determination module is used to construct a preset influenza antigen prediction model according to the local weights and the global weights; the preset influenza antigen prediction model is used to output the antigen prediction result of the protein sequence feature matrix to be detected.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

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

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images