Method for constructing segmented virus gene reassortant host adaptation prediction model

Abstract

The application belongs to the field of intelligent medical treatment, and particularly relates to a construction method of a segmented virus gene reassortment host adaptability prediction model. The method comprises the following steps: acquiring a virus data set and a host label; filtering the codon importance of different genes in the virus in the virus data set based on a codon importance threshold value of the different genes to obtain key codons of the different genes, combining data representation of the key codons of the different genes to obtain a simulation sequence of the virus data set; and inputting the simulation sequence of the virus data set and the host label into a neural network for iterative training to obtain the virus gene reassortment host adaptability prediction model. The simulation reassortment of the virus is realized based on the key codon region, and the human adaptability of the reassorted sequence is predicted based on the virus reassortment adaptability prediction model. The application can identify the reassortment result that is greater threat to human from a plurality of reassortment possibilities in time, so as to facilitate early targeted research and prevention of the reassortment result that is greater threat to human.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent medical treatment, and more particularly, to a method, device, medium and program product for constructing a segmented virus gene reassortment host adaptability prediction model. BACKGROUND

[0002] The genome of an influenza virus is segmented, and when different subtypes of virus strains recombine genetic material within the same host cell, offspring viruses with new antigenicity, transmission force and pathogenicity can be produced, which is referred to as gene reassortment. Gene reassortment is not only one of the important mechanisms of evolution and variation of influenza viruses, but also a key factor for inter-person or cross-species transmission. Therefore, studying influenza virus reassortment helps to understand the transmission and variation mechanisms of viruses, and can provide important scientific basis for vaccine development, antiviral drug development, epidemic prediction and prevention and control strategy formulation.

[0003] On the one hand, in traditional biological research, co-infection and genetic engineering techniques can be used for virus reassortment, but often require high experimental skills and equipment conditions, and new virus strains with higher pathogenicity or drug resistance can be produced during virus reassortment, posing a threat to laboratory biosafety. On the other hand, artificial intelligence (AI) methods such as machine learning (ML) and deep learning (DL) have achieved remarkable results in the field of biology in recent years, and have also shown amazing results in predicting virus host specificity and adaptability. Babayan et al. based on the genotypes of virus genetic code and encoded amino acids, etc. can predict the intermediate host, Li et al. established a machine learning model that can accurately predict the adaptability of the host of influenza A virus (IAV) based on the dinucleotide (DNT) characteristics of the gene, and revealed the host specificity of coronavirus genes based on the DNT cluster characteristics of the gene. However, the above AI methods can only predict the host adaptability of a single gene segment, and do not involve virus reassortment prediction.

[0004] Meanwhile, virological evidence highlights the polygenicity of IAV adaptation and adaptive reassortment. Hemagglutinin (HA) and neuraminidase (NA), two envelope glycoproteins that underlie viral serotypes, mediate viral binding and release. To date, only H1N1, H2N2, and H3N2 influenza viruses can bind to human-type influenza virus receptors and cause influenza pandemics through their HA specificity. Adaptive mutations in HA only cause fluctuations in receptor affinity, rather than a shift in avian-to-human binding specificity. However, compatibility and constraints between RNA polymerase-related genes are more complex and are crucial for the adaptation and adaptive reassortment of IAV. RNA polymerase-related genes such as PB2, PB1, PA, and NP are responsible for the proliferation and packaging of influenza viruses and are prone to reassortment between influenza viruses, so the present application studies polymerase-related genes. SUMMARY

[0005] In view of the above problems, the present application provides a method for constructing a segmented virus gene reassortment host adaptation prediction model, which uses key codon regions to simulate reassortment and realizes prediction of the adaptation of new virus simulation hosts, thereby facilitating early monitoring and response to high-risk human infection variations.

[0006] The first aspect of the present application first discloses a method for ordering the importance of virus genes, comprising:

[0007] S1: obtaining a test data set representing a virus gene and a host label, each gene data in the test data set being a data representation of each codon of the gene;

[0008] S2: predicting the host adaptation of the virus gene based on the test data set to obtain a predicted host label, and comparing the predicted host label with the host label to obtain a first prediction accuracy of the test set;

[0009] S3: using a sliding window method to mask the data representation of the codons in the first window to obtain a first masked test data set, predicting the host adaptation of the virus gene based on the first masked test data set to obtain a first window masked predicted host label, and comparing the first window masked predicted host label with the host label to obtain a first window masked prediction accuracy;

[0010] S4: repeating step S3 after sequentially masking the data representation of the codons in the first to nth windows by the sliding window to obtain a second prediction accuracy, the second prediction accuracy comprising the prediction accuracy of the first to nth windows, wherein n is the number of codons of the virus gene;

[0011] S5: obtaining the region importance of the first to nth windows based on the difference between the second prediction accuracy and the first prediction accuracy;

[0012] S6: obtaining codon importance by averaging the region importance of the window containing each codon.

[0013] Further, the data of the test set is connected at the beginning and end according to the gene codon, and then a sliding window mask is performed, wherein the end refers to the data representation of the last codon;

[0014] Optionally, the method for data representation of the codon is: connecting the gene sequence of the single gene at the beginning and end, and then using a sliding window method to traverse the gene sequence, the frequency of 64 different codons in the window forming a window codon frequency vector, and the window codon frequency vector representing the central codon of the window, and with the sliding of the window, each codon of the gene sequence is represented as the window codon frequency vector when the codon is the central codon.

[0015] Optionally, the gene data is an encoding feature matrix composed of each window codon frequency vector obtained by sliding window traversal;

[0016] Optionally, the viral gene host adaptability prediction is performed using a viral gene host adaptability prediction model, and the viral gene host adaptability prediction model is trained from a training data set of viral genes;

[0017] Optionally, S3 further comprises: obtaining the Bayesian posterior accuracy of the first window of the mask based on the mask first window prediction accuracy and the probability of each window of the mask; obtaining a third prediction accuracy by repeating the S3 step, the third prediction accuracy comprising the Bayesian posterior accuracy of the first window to the n-th window of the mask, and obtaining the Bayesian region importance of the first to the n-th window based on the difference between the third prediction accuracy and the first prediction accuracy; obtaining the Bayesian posterior codon importance by averaging the Bayesian region importance of each codon;

[0018] Optionally, the optimal codon importance is obtained by weighted average based on the codon importance and the Bayesian posterior codon importance.

[0019] The second aspect of the present application discloses a method for constructing a segmented viral gene reassortant host adaptability prediction model, comprising:

[0020] Obtaining a viral data set and a host label;

[0021] Filtering the codon importance based on the codon importance threshold of different genes for different genes of the viruses in the viral data set to obtain key codons of different genes, and combining the data representation of the key codons of different genes to obtain the reassortment result of the viral data set;

[0022] The recombination results of the virus dataset and the host label are input into the neural network for iterative training to obtain a virus gene recombination host adaptability prediction model.

[0023] Further, the method comprises:

[0024] The combination of the data-based representations of the key codons of different genes is performed by merging the data-based representations of the key codons of the genes in accordance with the order of the genes in the virus;

[0025] Optionally, the codon frequency vectors of the key codons of different genes are merged in accordance with the order of the genes in the virus to obtain a simulated sequence, which is a matrix.

[0026] The third aspect of the present application discloses a method for predicting the simulated recombination adaptability of a virus genome, comprising:

[0027] Obtaining a virus dataset, which is a data-based representation set of a virus;

[0028] For different genes of the virus in the virus dataset, the key codon regions of the different genes are obtained by filtering the codon importance based on the codon importance threshold of the different genes, and the key codons in the key codon regions of 1, 2 or 3 genes are randomly replaced to obtain a simulated recombination dataset of the virus dataset.

[0029] The simulated recombination results of the virus dataset are input into the virus gene recombination host adaptability prediction model to obtain the host adaptability corresponding to the simulated recombination results.

[0030] Further, the method for randomly replacing the key codons in the key codon regions is:

[0031] According to the codon importance of different genes, the data-based representations of the key codons of different genes constituting the virus are respectively intercepted, and the data-based representations of the key codons of 1, 2 or 3 corresponding genes from another strain of virus are used for random replacement to obtain the data-based representation of the recombination simulated sequence.

[0032] Optionally, the different viruses are viruses of different subtypes.

[0033] Optionally, any several key codon short fragments in the gene fragments of the influenza virus from humans are randomly replaced by the corresponding key codon short fragments of the corresponding genes of the avian virus subtype to obtain the simulated recombination results.

[0034] The fourth aspect of the present application discloses a codon importance ordering system for virus genes, comprising:

[0035] The first obtaining module is configured to obtain a test data set representing viral genes and a host label, each gene data in the test data set being a data representation of each codon of a gene;

[0036] The first prediction module is configured to predict host adaptability of the viral genes based on the test data set to obtain a predicted host label, and compare the predicted host label with the host label to obtain a first prediction accuracy of the test set;

[0037] The sliding mask prediction module is configured to mask data representations of codons in a first window of the test data set using a sliding window method to obtain a first masked test data set, predict host adaptability of the viral genes based on the first masked test data set to obtain a first window masked predicted host label, and compare the first window masked predicted host label with the host label to obtain a first window masked prediction accuracy. The sliding window masks data representations of codons in a first window to a nth window in turn, and the above-mentioned sliding mask prediction is repeated to obtain a second prediction accuracy, the second prediction accuracy including prediction accuracies of the first window to the nth window, where n is a number of codons of the viral genes;

[0038] The region importance calculation module is configured to obtain region importance of the first window to the nth window based on a difference between the second prediction accuracy and the first prediction accuracy;

[0039] The codon importance calculation module is configured to obtain codon importance by averaging region importance of a window containing each codon.

[0040] The present application (the fifth aspect) discloses a system for constructing a segmented viral gene reassortant host adaptability prediction model, comprising:

[0041] The obtaining module is configured to obtain a viral data set and a host label;

[0042] The reassortment module is configured to filter codon importance based on a codon importance threshold of different genes of the viral data set to obtain key codons of the different genes, and combine data representations of the key codons of the different genes to obtain a reassortment result of the viral data set;

[0043] The training module is configured to input the reassortment result of the viral data set and the host label into a neural network for iterative training to obtain a viral gene reassortant host adaptability prediction model.

[0044] The present application (the sixth aspect) also discloses a segmented viral gene reassortant host adaptability prediction system, comprising:

[0045] The third obtaining module is configured to obtain a viral data set, the viral data set being a data representation set of a virus;

[0046] The simulation recombination module is configured to filter the codon importance of different genes in the virus dataset based on a codon importance threshold, obtain key codon regions of the different genes after the filtering, and obtain simulation recombination results of the virus dataset by randomly replacing key codons in the key codon regions.

[0047] The third prediction module is configured to input the simulation recombination results of the virus dataset into a virus gene recombination host adaptability prediction model to obtain host adaptability corresponding to the simulation recombination results.

[0048] The seventh aspect of the present application discloses a computer device, the device comprising: a memory and a processor; the memory is configured to store program instructions; the processor is configured to invoke the program instructions, when the program instructions are executed, to perform the steps of the above-mentioned method.

[0049] The eighth aspect of the present application discloses a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the above-mentioned method.

[0050] The ninth aspect of the present application discloses a computer program product, comprising a computer program, which is executed by a processor to implement the steps of the above-mentioned method.

[0051] The present application has the following beneficial effects:

[0052] (1) The present application determines the key regions of codons by sorting the codon importance, and trains a recombination model based on the key codons to obtain a model that can predict the human adaptability of the virus after recombination.

[0053] (2) The present application realizes simulation recombination of the virus based on the key codon regions, and predicts the human adaptability of the recombined sequence based on the virus recombination adaptability prediction model. The present application can identify the recombination results that pose a greater threat to humans from a plurality of recombination possibilities in a timely manner, so as to facilitate early targeted research and prevention of the recombination results that pose a greater threat to humans. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0055] Figure 1 is a method flowchart provided by the second aspect of the embodiments of the present application;

[0056] Figure 2is a schematic diagram of a program product provided by the fifth aspect of the embodiment of the present application;

[0057] Figure 3 is a schematic diagram of a computer device provided by the embodiment of the present application;

[0058] Figure 4 is a schematic diagram of the architecture of an exemplary computer device provided by the embodiment of the present application;

[0059] Figure 5 is a schematic diagram of a storage medium provided by the embodiment of the present application;

[0060] Figure 6 is a Codon2Vec sliding window representation flowchart provided by the embodiment of the present application;

[0061] Figure 7 is a Codon2Vec preprocessing and representation evaluation flowchart provided by the embodiment of the present application;

[0062] Figure 8 is a ResNet residual network framework diagram provided by the embodiment of the present application;

[0063] Figure 9 is a codon importance calculation flowchart provided by the embodiment of the present application;

[0064] Figure 10 is a schematic diagram of a Bayesian model for calculating codon importance provided by the embodiment of the present application;

[0065] Figure 11 is a schematic diagram of an ablation experiment flowchart provided by the embodiment of the present application;

[0066] Figure 12 is a simulation of reassortant data generation schematic diagram provided by the embodiment of the present application;

[0067] Figure 13 is a reassortment adaptability model construction and prediction flowchart provided by the embodiment of the present application. DETAILED DESCRIPTION

[0068] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application.

[0069] In some of the flowcharts described in the description, claims, and drawings of the present application and above, a plurality of operations are included in a specific order, but it should be clearly understood that these operations can be performed in the order they appear in this paper or in parallel, the serial numbers of the operations such as S101, S102, etc. are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these flowcharts can include more or fewer operations, and these operations can be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this paper are used to distinguish different messages, devices, modules, etc. and do not represent the order of precedence. Also, "first" and "second" are different types.

[0070] The technical solutions in the embodiments of the present application will be described clearly and completely in the description of the embodiments of the present application combined with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0071] Figure 1 It is a kind of construction method flowchart of prediction model based on evaluation virus gene reassortment host adaptability provided by the embodiment of the present application, specifically, the method comprises the following steps:

[0072] S101: obtain virus data set and host label;

[0073] S102: for different genes of viruses in the virus data set, filter the codon importance based on the codon importance threshold value of different genes to obtain the key codons of different genes, and combine the data representation of the key codons of different genes to obtain the simulation sequence of the virus data set;

[0074] S103: input the simulation sequence of the virus data set and the host label into the neural network for iterative training to obtain the virus gene reassortment host adaptability prediction model.

[0075] The gene fragment of the virus refers to the nucleic acid sequence region in the virus genome with specific function or structural characteristics, which usually corresponds to the structural protein (such as capsid protein, envelope protein) or non-structural protein (such as replicase, regulatory protein) of the virus. Different viruses may contain multiple functionally independent gene fragments according to their genome characteristics (such as segmented or non-segmented, single-stranded or double-stranded, RNA or DNA).

[0076] Some examples of viral gene fragments:

[0077] NP (Nucleoprotein) gene segment of Avian Influenza Virus (e.g. H5N1): Encodes the nucleoprotein responsible for packaging viral RNA into ribonucleoprotein complexes (RNPs), which are essential for viral genome replication and transcription. This gene is highly conserved and commonly used for influenza virus typing and molecular diagnostics (e.g. RT-PCR detection).

[0078] HA (Hemagglutinin) and NA (Neuraminidase) gene segments of Influenza Virus: HA is responsible for host cell receptor binding and membrane fusion, while NA assists in virus release. Both are key determinants for influenza virus subtype classification (e.g. H1N1, H5N1).

[0079] gag, pol, env gene segments of HIV: Encode viral capsid proteins, reverse transcriptase / integrase, and envelope glycoproteins, respectively. They are critical targets for antiviral drugs (e.g. reverse transcriptase inhibitors) and vaccine development.

[0080] S (Spike Protein) gene segment of SARS-CoV-2: Encodes the viral spike protein that mediates ACE2 receptor binding in host cells. It is a central target for COVID-19 vaccine development (e.g. mRNA vaccines).

[0081] The study of gene segments provides a molecular basis for virus tracing (e.g. analyzing the host origin of Avian Influenza through NP gene), developing diagnostic reagents (e.g. antigen detection based on HA / NA), designing vaccines (e.g. screening S protein antigen epitopes), and developing antiviral drugs.

[0082] In some embodiments, the research aims to provide a method for constructing a reassortment host adaptation prediction model of IAV (Influenza A Virus), and using the model to predict the human adaptation of simulated reassortment sequences of Avian Influenza Virus (AIV).

[0083] Firstly, the Codon2Vec representation method is used, which is based on sliding window codon frequency statistics and transformer encoding layer pre-training for encoding representation;

[0084] Then, based on the ResNet residual network, single gene adaptation prediction models of PB2, PB1, PA, and NP four segments are established, and the key codon regions in each gene are calculated through ablation experiments;

[0085] Finally, according to the calculated key codon regions, simulated reassortment of Avian Influenza Virus sequences and human influenza virus sequences is realized, and a reassortment host adaptation prediction model of IAV is constructed based on ResNet, which is finally applied to the reassortment host adaptation prediction of simulated reassortment sequences.

[0086] First, the eight segments of the influenza virus genome, HA, NA, M, NS, NP, PB1, PB2, and PA, respectively encode 10 proteins: hemagglutinin (HA), neuraminidase (NA), matrix proteins M2 and M1, non-structural proteins NS1 and NS2, nucleocapsid, and three polymerases. The NP protein encoded by segment 5, together with the polymerase trimer PB1, PB2, and PA, and viral RNA, forms the nucleoprotein of the influenza virus (vRNP). The RNA of the influenza virus is transcribed and replicated in the nucleus of the cell, and one of the key processes is the nucleocytoplasmic shuttling of the nucleoprotein. During the early stage of infection when the influenza virus invades the cell, the nucleoprotein of the influenza virus rapidly enters the nucleus to start transcription, and as a major component of the vRNP complex, NP mediates the nuclear import of the vRNP complex through its nuclear localization signals (NLSs). The NP protein is translated from the viral mRNA in the cytoplasm and enters the nucleus, facilitating the replication and transcription of the viral genome. In summary, the NP protein of the influenza virus is an important viral protein that determines the host range and pathogenicity of the influenza virus.

[0087] At present, the present application is mainly directed to the study of polymerase-related genes, and the single-gene host adaptability and reassortment host adaptability prediction models trained are also for these four genes, but the research method in the present application can be extended to other gene segments or even other viruses, i.e., other genes or viruses can be characterized using the method, and the corresponding adaptability prediction model can be trained. For other gene segments of IAV, they can be generated together with the polymerase-related genes for simulated reassortment, and an AI model suitable for more gene simulated reassortment adaptability prediction can be trained.

[0088] The processing flow and specific running steps of the present application are as follows:

[0089] I. Codon2Vec characterization method

[0090] 1.1 (step 1): input the cleaned csv file containing IAV gene sequences:

[0091] The file contains multiple lines, each line including an IAV virus sequence and its corresponding host label;

[0092] Influenza A Virus (IAV) is the general term for the influenza A virus, which belongs to the Orthomyxoviridae family. Its sub-type classification is based on two glycoprotein antigens on the surface of the virus: Hemagglutinin (HA) and Neuraminidase (NA). There are currently 18 subtypes of HA (H1-H18) and 11 subtypes of NA (N1-N11), and different combinations of HA and NA form various subtypes of IAV (e.g. H1N1, H3N2, etc.).

[0093] Therefore, the IAV gene sequence file contains the gene sequences of different subtypes of IAV.

[0094] The cleaned IAV gene sequence file contains the gene sequences of N virus fragments and their corresponding host labels.

[0095] N = 8, including fragments HA, NA, M, NS, NP, PB1, PB2, PA;

[0096] Since different fragments need to be predicted separately, different fragments are processed separately,

[0097] First, for the first fragment, such as the NP fragment, use the sliding window method to traverse each row of the NP fragment gene sequence, with a window size (WS1) of 128. The codon frequency in the window is used to encode the central codon in the current window. Since there are 64 types of codons, each central codon is represented as a one-dimensional vector with 64 numerical features.

[0098] As the window slides, each codon in the sequence is encoded and represented as a two-dimensional matrix with codon number x 64, as shown in Figure 6 .

[0099] Figure 6 The gene sequence of a certain row is represented as: ATG1 GAA2 AGA3 ATA4 AAA5 GAA6 CTA7 GAT8 … TAGn

[0100] The size of the sliding window is S, and when S contains 128 codons, S = 384;

[0101] Based on the current window, the frequency of 64-bit codons is calculated, and the frequency of each codon in the current window is calculated as follows:

[0102]

[0103] When the window size is 128 codons, S = 384, and the above formula corresponds to:

[0104]

[0105] condon j represents 64 codons;

[0106] Thus, the vector representation of the codon frequency is obtained as:

[0107]

[0108] represents the codon frequency vector within the window, with dimension 64*1; thus, the central codon is encoded into a 1x64 vector,

[0109] For example, the codons contained in the 64th window are ATG1 GAA2 AGA3 ATA4 AAA5 GAA6 CTA7 GAT8... GUC 63 AUA 64 AUC 65 ... UUC 128 ;

[0110] The central codon in the current window is AUA 64 ,

[0111] After the above operation, with the sliding of the window, each codon in the sequence is encoded and characterized, i.e. the central codon is characterized by the codon frequency vector, and thus a two-dimensional matrix of the number of codons contained in the gene sequence x 64 is obtained, i.e. assuming that the gene sequence of an NP fragment contains n codons, the dimension of the encoded feature matrix is n x 64, as shown in Figure 6 .

[0112] Since we want to encode each codon, for the central codon of the window being ATG1, the sequence of the window is obtained by splicing the codons at the end of the sequence, and thus the codon frequency vector is calculated, and when the central codon is the last codon, the sequence of the window is calculated by splicing the codons in the first half of the sequence.

[0113] The determination of the sliding window is shown in the following pseudo code:

[0114] Table 1 Logic of encoding codons in sliding window

[0115]

[0116]

[0117] Due to the input size requirement of the subsequent ResNet framework, zero vector padding is performed at the end of the feature matrix. For example, the feature matrix size of the PB2 gene is 760x64, while the input parameter of the subsequent ResNet model framework is 128x128x3, so the size of the feature matrix of the PB2 gene is adjusted to 768x64, that is, an 8x64 zero vector is added at the end of the original feature matrix. Finally, the representation matrix of the same gene of multiple different subtypes of virus sequences is stored in an npy file.

[0118]

[0119] wherein, represents the codon frequency vector when the 1st codon in the sequence is the central codon; ORF represents the gene fragment sequence, and A(ORF) represents the encoded encoding feature matrix.

[0120] In some embodiments, in order to effectively encode, that is, to reduce the 0 value in the codon frequency vector, it is recommended that the window width be greater than 64.

[0121] In some embodiments, the window width of the sliding window is 128 codons.

[0122] In some embodiments, the window width of the sliding window is 100 codons.

[0123] In some embodiments, the prediction network model used replaces the RestNet network with any of the following: DenseNet, ResNeXt, Inception-ResNet, MobileNetV2, EfficientNet, Wide Residual Networks (WRN), and when applied, the dimension of the transformation matrix is changed based on the input requirements of the network, or the window width of the sliding window is changed.

[0124] 1.2 (step 2): Input the npy file obtained in step 1, and use the transformer encoding layer to pre-train it to obtain the optimized feature encoding (the size is consistent with step 1), and store it in another npy file, wherein the Transformer pre-training performs an unsupervised training to perform a self-encoding task to reconstruct the input data to obtain the optimized feature encoding, Figure 7 A'(ORF) shown.

[0125] 1.3 (step 3): input the two npy files obtained in step 1 and step 2, use umap dimension reduction and hierarchical clustering in unsupervised learning to evaluate the representation effect of the representation matrix in the two npy files (i.e. evaluate the representation effect of transformer encoding layer pre-training), and use the external evaluation index of clustering, adjusted rand index (ARI) and normalized mutual information (NMI) to evaluate the clustering effect, and the final evaluation result is the dimension reduction visualization result of the two representation matrices and the clustering effect score.

[0126] According to the evaluation result, the optimal npy file of the training set corresponding encoding matrix is stored, that is, the optimal encoding feature matrix (if the transformer encoding layer pre-training effect is not good on this segment, the original representation matrix is used for representation, that is, the npy file obtained in step 1; otherwise, the representation matrix after transformer encoding layer pre-training is used, that is, the npy file obtained in step 2). The processing flow of step 2-3 is shown in Figure 7

[0127] In some embodiments, only the clustering evaluation index is used to select the optimal encoding feature matrix.

[0128] In some embodiments, only the dimension reduction visualization result is used to select the optimal encoding feature matrix.

[0129] In some embodiments, the optimal encoding feature matrix is selected by integrating multiple evaluation results, and the selection process is shown in Figure 7 According to different indexes, the value of c is updated, and A(ORF) or A'(ORF) is selected as the optimal encoding feature matrix according to the value of c.

[0130] In some embodiments, the encoding feature matrix obtained based on the sliding window is used to perform the following subsequent steps.

[0131] In some embodiments, the encoding feature matrix after using the Transformer auto-encoding is used to perform the following subsequent steps.

[0132] In some embodiments, the optimal encoding feature matrix is used to perform the following subsequent steps.

[0133] Unless otherwise specified, the following encoding feature matrix A(ORF) is any one of the following: encoding feature matrix, Transformer output encoding feature matrix, and optimal encoding feature matrix.

[0134] II. Establishing a ResNet single-fragment adaptive prediction model ​

[0135] 2.1 (Step 4): Training of the single-gene host adaptation prediction model using ResNet residual network framework (as shown in FIG. 2.1) is performed. In the training, the label used is the original host label of the sequence. If the host is avian, the model label is “0”; if the host is human, the model label is “1”. The trained model (i.e., the single-segment host adaptation prediction model) is saved and can be used for predicting the host adaptation of IAV with unknown host. Figure 8

[0136] That is, steps 1-4 are repeated for different segments to obtain single-segment adaptation prediction models for different segments.

[0137] In some embodiments, steps 1-4 are independently performed for 4 gene sequences (PB2, PB1, PA and NP), and finally 4 single-gene adaptation prediction models are obtained.

[0138] In some embodiments, steps 1-4 are independently performed for 8 gene sequences, and finally 8 single-gene adaptation prediction models are obtained.

[0139] III. Ablation experiment / Bayesian model calculates the order of importance of codons in each gene segment

[0140] 3.1 (Step 5): The trained single-segment host adaptation prediction model is used to predict the encoded feature matrix of the test set (the same processing is performed on the genes of the test set to obtain the encoded feature matrix of the test set), and the current prediction accuracy (acc1) is obtained. The prediction accuracy is used to observe the influence of the mask in the ablation experiment (Step 6).

[0141] Step 6: In order to evaluate the importance of each codon in the gene segment, a sliding window mask (window size = 80) is performed on the encoded feature matrix of the test set along the row direction of the matrix. That is, the encoded sequences of the corresponding 80 codons in the sliding window are covered, i.e., the codon frequency vectors representing the codons in the sliding window region are masked as zero vectors. Since the codon frequency vectors in the window are zero vectors, the codon mask is shown in FIG. 6A, or the encoded feature matrix after masking is shown in FIG. 6B, where some vectors are represented as zero vectors. When performing the sliding window mask on the encoded feature matrix, the judgment of the corresponding mask region is shown in Table 2. Figure 10 Figure 11

[0142] Table 2: Judgment logic of the sliding window mask encoded feature matrix (pseudo code)

[0143] ​​​

[0144]

[0145] The codon feature vector in the first window of the mask is a zero vector, and the encoding feature matrix A(IRF) of the first window of the mask is obtained mask1 ,

[0146] For the second window, A(ORF) is obtained mask2 ,

[0147] For the nth window, A(ORF) is obtained maskn ,

[0148] Finally, for a gene fragment containing n codons, n encoding feature matrices after masking are obtained, and the optimized feature encoding is obtained after inputting the n encoding feature matrices after masking into the pre-trained Transformer representation matrix. After obtaining the optimal feature matrix in the manner of step 3, input the trained single gene / fragment host adaptability model for prediction to obtain the accuracy (acc2, acc2 is a list) corresponding to each mask feature matrix. Compare acc1 with acc2, and the difference between the two can reflect the importance of the codon vector in the current mask region. If the difference between acc2 and acc1 is large, it means that the 80 codons in this part of the vector are masked and have a greater impact on the accuracy, that is, the importance of the codon vector representation is high.

[0149] Then slide the window. With the sliding of the window, according to the difference between acc2 and acc1 obtained by predicting after masking the corresponding region vector, the importance of each block region vector is calculated in turn. Since the importance obtained at present is for the vector in a whole region of the mask, the importance of the codon frequency vector in the region is averaged according to the window size (80) to obtain the importance of each vector (Vector importance, VI). Then, according to the sliding window size (128) in the representation process of step 1, the importance of each vector is averaged to obtain the importance of each codon in the sequence (Codon importance, CI) - CI1. The specific details are as shown in Figure 9 , Table 3:

[0150] Table 3 Codon importance calculation process

[0151]

[0152]

[0153]

[0154] Wherein acc nAccuracy of the n-th sliding window position, VI n Importance of the n-th vector, CI n Importance of the n-th codon.

[0155] 3.2 (Step 7:) Calculate the importance of each codon (CI2) using the Bayesian model:

[0156] As shown in the following equation, the encoded feature matrix is input into the single-segment host adaptability prediction model to obtain the accuracy acc1: Figure 10

[0157] After the encoded feature matrix is masked by the sliding window, the encoded feature matrix of the i-th window is obtained in turn, and the single-segment host adaptability prediction model is input to obtain the accuracy acc2:

[0158] Wherein, acc2 is a list, and the elements in the list respectively represent the accuracy obtained when the i-th window is masked:

[0159] acc2 = [acc21, acc22, …, acc2 i n-1, acc2 n ]

[0160] Wherein, acc2 i n-1 represents the accuracy of the single-segment host adaptability prediction model input by the encoded feature matrix of the i-th window; the Bayesian accuracy acc3 is obtained by substituting acc2 into the Bayesian model, that is, P(B i |A)

[0161]

[0162] Wherein, P(B i |A) represents the probability of event B i occurring based on condition A, wherein event A represents the prediction accuracy; B i represents the event of the encoded feature matrix of the i-th window;

[0163] P(B i ) represents the probability of the encoded feature matrix of the i-th window appearing, in our scenario, since we use a sliding window mask with a step size of 1 codon, n is the number of codons;

[0164] A represents the prediction accuracy, A|B i represents the prediction accuracy obtained when the i-th window is masked, P(A|B i ) represents the probability of the feature matrix A under the condition of event B i , which is acc2 i here.​

[0165] Therefore, the posterior probability of acc2 (post-acc2), i.e., acc3, is expressed as:

[0166]

[0167] acc2 i The accuracy is obtained by inputting the encoded feature matrix of the i-th window of the mask into the single-fragment host adaptive prediction model.

[0168] acc3 i This represents the posterior accuracy of the single-fragment host adaptive prediction model after inputting the encoded feature matrix of the i-th window of the mask.

[0169] acc3 = [acc31, acc32, ..., acc3] i ,…,acc3 n ]

[0170] The posterior importance of the mask window is obtained based on the difference between acc3 and acc1: CI2 is calculated by replacing acc-list in Table 3 with acc3-acc1 and then calculating the codon importance based on the codon importance calculation process shown in Table 3.

[0171] Step 8: Average the codon importance calculated in Steps 6 and 7, as shown in Formula 1, to obtain the final importance of each codon, as follows: Figure 10 As shown, normalize it and save it to a CSV file.

[0172]

[0173] The ablation experiment of the four gene fragments requires running steps 5-8 independently to obtain the codon importance results on the four gene fragments.

[0174] In some embodiments, the codon importance is obtained using only the CI calculated in step 6.

[0175] In some embodiments, codon importance is obtained using Bayesian-corrected CI2.

[0176] In some embodiments, codon importance is obtained by combining the mean or weighted sum of the codon importance (CI1) from step 6 and the Bayesian-corrected CI2.

[0177] IV. Key Codon Combinations

[0178] Step 9: Based on the evaluation results of Step 3, input the optimal coding feature matrix of the 4-segment training set. Based on the CSV file of the codon importance results obtained in Step 8, extract and merge the key codons that are important for adaptation in the 4 segments.

[0179] That is, according to the calculated codon importance in the four genes, the key codons found in the literature are compared, and the codon importance threshold is selected under the premise of containing as many key codons as possible found in the literature, the vector representation of the codons greater than the importance threshold is extracted from the feature matrix, and the key codon representations of the four genes are obtained in turn according to their order in the segment. Finally, the four segments are merged in the order of PB2, PB1, PA, and NP. Different importance thresholds can be selected as needed, but after the threshold is selected, there is only one possible result for the splicing result.

[0180] According to the size requirements of the subsequent model, the above-mentioned merged feature matrix is filled in size (step 1), and the encoding matrix of the key codons of the four genes in the feature training set is obtained, such as Figure 12 The upper part shows that: by screening the key codons of the four genes (PB2, PB1, PA, and NP), the depth of green in the figure represents different codon importance thresholds. Taking the most stringent codon importance threshold as an example, the simulated ORF' represents the simulation sequence, and it can be seen that the simulation sequence is obtained by concatenating the key codon regions of PB2, PB1, PA, and NP.

[0181] Based on the feature and the ResNet residual network framework in step 4, a reassortment adaptation prediction model is trained (here, the self-catenation form of the four gene segment features of the same strain virus in the training set is used, and the label is the same as the original strain), as shown in Figure 13 , which is used to predict the reassortment host adaptation of avian influenza viruses after 2020.

[0182] Step 10: Since the present application is to predict the reassortment host adaptation of the reassortment sequence that does not exist in nature, it is necessary to generate the feature representation of all possible simulated reassortment sequences of avian influenza according to the existing avian influenza virus sequences.

[0183] In some embodiments, the four-segment encoding feature matrix of the prediction set is input, the key codon representation of the four segments is extracted according to the codon importance result csv file obtained in step 8, and the key codon encoding matrix of the four sequence segments of the selected human influenza virus model strain is randomly replaced, and then the feature representation of the reassortment simulation sequence is generated, as shown in Figure 12 , and saved to the npy file.

[0184] In some embodiments, Figure 12Figure 2 shows a schematic diagram of simulated reassortment, where green is the IAV gene sequence detected in avian or swine, sky blue is the gene sequence from human H1N1, and dark blue is the gene sequence of H3N2 with human host;

[0185] PB2' is a short fragment of key codons in the entire PB2 gene segment of the virus strain replaced by the calculated key codons in the PB2 gene;

[0186] PB1' is a short fragment of key codons in the entire PB1 gene segment of the virus strain replaced by the calculated key codons in the PB1 gene;

[0187] PA' is a short fragment of key codons in the entire PA gene segment of the virus strain replaced by the calculated key codons in the PA gene;

[0188] NP' is a short fragment of key codons in the entire NP gene segment of the virus strain replaced by the calculated key codons in the NP gene;

[0189] After random replacement, the gene sequence of simulated reassortment is obtained.

[0190] In some embodiments, the possibility of virus reassortment is represented as:

[0191]

[0192] wherein R n represents the number of reassortment possibilities, 4 represents 4 segments that are reassorted, represents that 1 gene of the human model strain virus is replaced, represents that 2 genes of the human model strain virus are replaced, represents that 3 genes of the human model strain virus are replaced.

[0193] For different genes of the viruses in the virus data set, the key codon regions of different genes are obtained by filtering the codon importance based on the codon importance threshold of different genes, and the simulated reassortment results of the virus data set are obtained by randomly replacing the key codons in the key codon regions. The method for randomly replacing the key codons in the key codon regions is:

[0194] According to the codon importance of different genes, the data characterization of the key codons of different genes constituting the virus is intercepted respectively, and the data characterization of the key codons of the corresponding genes from different viruses is randomly replaced to obtain the data characterization of the simulated recombination sequence; here, the corresponding genes refer to: according to the calculated codon importance of each gene, the key codon characterization on the corresponding gene is intercepted, for example: PB2 gene may have 1, 2, 3 codon importance, and the 1, 2, 3 codon characterizations are intercepted, while the PB1 gene may have 4, 5, 6 codon importance, and the 4, 5, 6 codon characterizations are intercepted, and the important codons of 1, 2 or 3 genes are randomly replaced, that is, the 1, 2, 3 codon characterizations on the PB2 gene of virus A are used to replace the 1, 2, 3 codon characterizations on the PB2 gene of virus B, which is the random replacement of the key codons of 1 gene.

[0195] In some embodiments, the step of simulating recombination is: after randomly replacing 1, 2 or 3 key codon short fragments from 1, 2 or 3 genes of four gene fragments (PB2, PB1, PA, NP) of human H1N1 or H3N2 model strain virus with the corresponding key codon short fragments of PB2, PB1, PA, NP of avian virus subtypes, a simulated recombination gene sequence is obtained.

[0196] In some embodiments, the key codon regions of PB2, PB1, PA, NP are selected, and the simulated recombination results are obtained after randomly combining the screened key codons.

[0197] The above simulated recombination is based on the research of the four gene fragments in this study. When the research involves eight genes, any 1-7 key codon short fragments from eight gene fragments (HA, NA, M, NS, NP, PB1, PB2, PA) of human H1N1 or H3N2 model strain virus are randomly replaced with the corresponding key codon short fragments of the corresponding gene fragments of avian virus subtypes to obtain a simulated recombination gene sequence.

[0198] Step 11: input the npy file obtained in step 10, use the recombination adaptability prediction model trained in step 9 to perform simulated recombination host adaptability prediction, and obtain the simulated recombination adaptability prediction result, that is, the predicted host label (adapted to humans or adapted to birds) of the simulated recombination sequence. Figure 13 ) Analysis of the prediction results of each simulated recombination can determine which fragments of the avian influenza virus have a higher risk of recombination and adaptation to humans, and which subtypes have a higher risk of simulated recombination and adaptation to humans, providing direction guidance for prevention and control of influenza A virus. At the same time, this method can be applied to the recombination adaptability prediction problem of other viruses, or extended to the recombination monitoring of more gene fragments.

[0199] In the reassortment adaptability prediction problem of some other viruses, if the fragments contained are different from those of IAV, after collecting data, the above method is used to predict the adaptability of the fragments, determine the codon importance ranking of the fragments, screen out the key codons, train the reassortment adaptability prediction model of the virus, and simulate the adaptability of different reassortment results after reassortment, so as to carry out targeted prevention.

[0200] Figure 3 This is a schematic diagram of a computer device provided in an embodiment of the present invention, such as... Figure 3 As shown, the device 2000 may include: one or more processors 2010 and one or more memories 2020; wherein the memories store computer-readable code that, when run by the one or more processors, can perform the methods described above.

[0201] The processor in this embodiment can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, operations, and logic block diagrams disclosed in this embodiment. The general-purpose processor can be a microprocessor or any conventional processor, and can be based on an x86 or ARM architecture.

[0202] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0203] For example, the method or apparatus according to embodiments of this disclosure can also be used by means of Figure 4 The architecture of the computing device 3000 shown is used for implementation. For example... Figure 4As shown, the computing device 3000 can include a bus 3010, one or more CPUs 3020, a read only memory (ROM) 3030, a random access memory (RAM) 3040, a communication port connected to a network 3050, an input / output component 3060, a hard disk 3070, etc. The storage devices in the computing device 3000, such as the ROM 3030 or the hard disk 3070, can store various data or files used in processing and / or communication of the method provided by the present disclosure and the program instructions executed by the CPU. The computing device 3000 can also include a user interface 3080. Of course, Figure 4 The architecture shown is exemplary only, and in implementing different devices, components shown can be omitted, Figure 4 one or more components of the computing device shown.

[0204] The embodiments of the present disclosure also provide a computer readable storage medium, such as Figure 5 As shown, a schematic diagram of a storage medium 4000 provided by the embodiments of the present disclosure is shown, and the computer readable instructions 4010 are stored on the computer storage medium 4020. When the computer readable instructions 4010 are executed by a processor, the method according to the embodiments of the present disclosure described with reference to the above figures can be executed. The computer readable storage medium in the embodiments of the present disclosure can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. The non-volatile memory can be a read only memory (ROM), a programmable read only memory (PROM), an erasable programmable read only memory (EPROM), an electrically erasable programmable read only memory (EEPROM) or a flash memory. The volatile memory can be a random access memory (RAM) used as an external cache. By way of example, and not limitation, many forms of RAM can be used, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct memory bus random access memory (DRAM). It should be noted that the memory of the method described herein is intended to include, but not be limited to, these and any other suitable types of memory. It should be noted that the memory of the method described herein is intended to include, but not be limited to, these and any other suitable types of memory.

[0205] The embodiments of the present disclosure also provide a computer program product or computer program, which, when executed by a processor, implements the steps of the above method, such as Figure 2 As shown, the computer program product or computer program includes:

[0206] The acquisition module 201 acquires a virus dataset and a host label.

[0207] The recombination module 202: for different genes of a virus in the virus dataset, based on the codon importance threshold of the different genes, filters the codon importance to obtain key codons of the different genes, and combines data representation of the key codons of the different genes to obtain a recombination result of the virus dataset.

[0208] The training module 203: input the recombination result of the virus dataset and the host label into a neural network to iteratively train to obtain a virus gene recombination host adaptability prediction model.

[0209] It should be noted that the flowcharts and block diagrams in the drawings illustrate the possible implementation architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, which contains one or more executable instructions for implementing the specified logic function. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different order from that noted in the drawings. For example, two blocks represented in succession can actually be executed substantially in parallel, and sometimes they can be executed in reverse order, depending on the function involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0210] Generally, various example embodiments of the present disclosure can be implemented in hardware or special-purpose circuitry, software, firmware, logic, or any combination thereof. Certain aspects can be implemented in hardware, while other aspects can be implemented in firmware or software which can be executed by a controller, microprocessor or other computing device. When aspects of the disclosure are illustrated or described as a block diagram, flowchart, or using some other pictorial representation, it is to be understood that the blocks, apparatus, systems, techniques or methods described herein can be implemented in hardware, software, firmware, special-purpose circuitry, general-purpose hardware or controller or other computing device, or some combination thereof.

[0211] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0212] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the embodiments of the device described above are merely schematic; for example, the division of the units is only a logical function division; there can be another division manner for the actual implementation; for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, or the among different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electric, mechanical or other forms.

[0213] The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0214] In addition, each functional unit in the various embodiments of the present application can be integrated in one processing unit, or each unit can exist physically as a separate unit, or two or more units can be integrated in one unit. The integrated unit can be implemented in the form of hardware, or in the form of a software functional unit.

[0215] The example embodiments of the present disclosure described in detail above are merely illustrative, rather than restrictive. Those skilled in the art should understand that various modifications and combinations can be made to the embodiments or features thereof without departing from the principles and spirits of the present disclosure, and such modifications should fall within the scope of the present disclosure.

Claims

1. A method for codon importance sequencing of a viral gene, characterized in that, The method comprises: S1: obtaining a test data set representing viral genes and a host label, each gene data in the test data set is a data representation of each codon of the gene; the data of the test data set is masked by a sliding window after being connected at the beginning and end, wherein the end refers to the data representation of the last codon; the method of data representation of the codon is: after connecting the gene sequence of a single gene at the beginning and end, the gene sequence is traversed using a sliding window method, the frequency of 64 different codons in the window constitutes a window codon frequency vector, and the window codon frequency vector represents the central codon of the window; with the sliding of the window, each codon of the gene sequence is represented as the window codon frequency vector when the codon is the central codon; S2: predicting the host label of the test data set by a viral gene host adaptability prediction model to obtain a predicted host label, and comparing the predicted host label with the host label to obtain a first prediction accuracy of the test set; the viral gene host adaptability prediction model is obtained by training a training data set of viral genes; the label used for training is the original host label of the sequence, and different labels are assigned according to different host types; S3: calculating the second prediction accuracy corresponding to each masking feature matrix after sequentially masking the data representation of the codons in the first to nth window by a sliding window, wherein the second prediction accuracy comprises a list of the masking first window prediction accuracy to the masking nth window prediction accuracy, wherein n is the number of codons of the viral gene; wherein the calculation method of the masking first window prediction accuracy comprises: obtaining a first masked test data set by using a sliding window method to mask the data representation of the codons in the first window of the test data set, obtaining a masking first window predicted host label based on the first masked test data set, and obtaining the masking first window prediction accuracy by comparing the masking first window predicted host label with the host label; the calculation method of the masking nth window prediction accuracy is the same as that of the first window prediction accuracy, but the nth window is the window after window sliding; S4: obtaining the region importance of the first to nth window based on the difference between the second prediction accuracy and the first prediction accuracy; S5: obtaining the importance of each vector by averaging the region importance of the window containing each codon; the importance of each vector is averaged to obtain the codon importance of each codon according to the size of the sliding window in the data representation.

2. The method of codon importance ranking of viral genes according to claim 1, wherein, The gene data is an encoding feature matrix composed of window codon frequency vectors obtained by sliding window traversal.

3. The method of codon importance ranking of viral genes according to claim 1, wherein, S3 further comprises: obtaining the Bayesian posterior accuracy of the first mask window based on the mask first window prediction accuracy and the probability of each window of the mask; repeating the steps of S3 to obtain a third prediction accuracy, the third prediction accuracy comprising the Bayesian posterior accuracy of the first to the n-th window of the mask, and obtaining the Bayesian region importance of the first to the n-th window based on the difference between the third prediction accuracy and the first prediction accuracy; and obtaining the Bayesian posterior codon importance by averaging the Bayesian region importance of each codon.

4. The method of codon importance ranking of viral genes according to claim 3, wherein, The optimal codon importance is obtained based on the weighted average of the codon importance and the Bayesian posterior codon importance.

5. A method for constructing a segmented viral gene reassortment host adaptability prediction model, characterized in that, The method comprises: obtaining a virus dataset and a host label; filtering the codon importance of different genes of viruses in the virus dataset based on a codon importance threshold of the different genes to obtain key codons of the different genes, and combining the data-based representations of the key codons of the different genes to obtain a reassortment result of the virus dataset; wherein the codon importance is obtained based on the method of any one of claims 1-4; inputting the reassortment result of the virus dataset and the host label into a neural network for iterative training to obtain a virus gene reassortment host adaptability prediction model.

6. A method for constructing a segmented virus gene reassortant host adaptation prediction model, comprising the steps of: The method comprises: obtaining a virus dataset and a host label; filtering the codon importance of different genes of viruses in the virus dataset based on a codon importance threshold of the different genes to obtain key codons of the different genes, and combining the data-based representations of the key codons of the different genes to obtain a simulated sequence of the virus dataset; wherein the codon importance is obtained based on the method of any one of claims 1-4; inputting the simulated sequence of the virus dataset and the host label into a neural network for iterative training to obtain a virus gene reassortment host adaptability prediction model.

7. The method of claim 6, wherein the method further comprises: determining the number of segments of the segmented virus; and determining the number of genes in each segment of the segmented virus. The method comprises: The way of combining the data-based representations of the key codons of the different genes is to merge the data-based representations of the key codons of the genes in accordance with the order of the genes in the virus.

8. The method of claim 6, wherein the method further comprises: determining the number of segments of the segmented virus; and determining the number of genes in each segment of the segmented virus. The simulated sequence is obtained by merging the codon frequency vectors of the key codons of the different genes in accordance with the order of the genes in the virus to obtain a simulated sequence, and the simulated sequence is a matrix.

9. A segmented viral genome simulation reassortment fitness prediction method, characterized by, The method comprises: obtaining a virus dataset, the virus dataset being a set of data-based representations of viruses; filtering the codon importance of different genes of viruses in the virus dataset based on a codon importance threshold of the different genes to obtain key codon regions of the different genes, and randomly replacing the key codons in the key codon regions of the K genes to obtain a simulated reassortment dataset of the virus dataset; inputting the simulated reassortment dataset of the virus dataset into the virus gene reassortment host adaptability prediction model based on any one of claims 5-8 to obtain the host adaptability corresponding to the simulated reassortment result.

10. The segmented viral genome simulation reassortant fitness prediction method of claim 9, wherein, The method of randomly replacing the key codons in the key codon regions is: According to the codon importance of different genes, the data characterization of the key codons of K different genes constituting the virus is intercepted respectively, and the data characterization of the key codons of K corresponding genes from different viruses is randomly replaced to obtain the data characterization of the reassortant simulation sequence.

11. The segmented viral genome simulation reassortment fitness prediction method of claim 10, wherein, The different viruses are viruses of different subtypes.

12. The segmented viral genome simulation reassortant fitness prediction method of claim 10, wherein, The key codon short fragments in K gene fragments of the influenza virus from human are randomly replaced by the corresponding key codon short fragments of the corresponding genes of the avian virus subtype to obtain a simulation reassortment result.

13. The segmented viral genome simulation reassortant fitness prediction method of claim 12, wherein, K is in the range of 1-7.

14. A computer device, comprising: The device comprises a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to realize the steps of the method of any one of claims 1-13.

15. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and the computer program is executed by a processor to realize the steps of the method of any one of claims 1-13.

16. A computer program product comprising a computer program, characterized in that, The computer program is executed by a processor to realize the steps of the method of any one of claims 1-13.

Images