Data prediction method and device for influenza virus antigen relationship
By performing feature extraction and prediction model training on the sample strain data of influenza virus, antigen relationship prediction is optimized, and the problem of inaccuracy and sensitivity of antigen relationship prediction in the prior art is solved, and the adaptability of vaccines and influenza viruses are improved. Mutational response ability.
Patent Information
- Application Number
- CN202411573695.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-06
- Publication Date
- 2024-12-06
AI Technical Summary
Existing hemagglutination inhibition (HI) tests have long cycles, limited sensitivity and accuracy when predicting antigen mutations of influenza viruses, and are particularly unsatisfactory in dealing with new strains, resulting in analysis of similarity between vaccine strains and epidemic strains. Difficulties affect the effectiveness of the vaccine.
By reading the strain data of the sample strain pair, determine the virus characteristic data and neutralization rate, train and adjust using the prediction model, and optimize the model based on the deviation between the predicted antigen relationship and the actual antigen relationship to improve the accuracy of antigen relationship prediction .
More accurate predictions of strain antigen relationships have been achieved, ensuring the adaptability and effectiveness of the vaccine, and the ability to adjust the vaccine more quickly to cope with influenza virus mutations.
Smart Images

Figure CN119092141A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of machine learning, and more particularly to a method and device for predicting data on influenza virus antigen relationships. Background Art
[0002] Influenza is a disease caused by influenza viruses. Flu vaccination is the most effective way to prevent influenza. Its prevention principle is to stimulate the body to produce specific immune responses to influenza virus antigens, thereby reducing the spread and severity of the disease. The process of developing influenza vaccines relies on the antigenic relationship between vaccine strains and epidemic strains. In other words, in the process of developing influenza vaccines, the influenza vaccine for epidemic strains is determined by analyzing whether the vaccine strains (strains that have appeared in history) are similar to epidemic strains (the latest emerging and popular strains). However, due to the high variability of influenza viruses, it is very difficult to recommend vaccine strains that are similar to current and future epidemic strains.
[0003] Currently, the hemagglutination inhibition (HI) test is generally used to determine the antigenic variation of influenza viruses, thereby deciding whether the vaccine strain needs to be replaced. Among them, viruses with hemagglutinin (HA) can agglutinate human or animal red blood cells, which is called hemagglutination. The hemagglutination phenomenon can be inhibited by corresponding antibodies, which is called HI test. The HI test is a technology for measuring antigens or antibodies.
[0004] However, the HI test cycle is long, and its sensitivity and accuracy are limited. Especially when dealing with new strains, the effect of predicting the antigenic relationship between two strains is not ideal. Therefore, when the antigenic relationship prediction model is trained based on the test data obtained from the HI test, the antigenic relationship output by the trained antigenic relationship prediction model is not accurate enough. In summary, the antigenic relationship between strains analyzed by the HI test is not accurate, and it is difficult to make rapid adjustments to ensure the effectiveness of the vaccine when dealing with seasonal influenza virus antigenic drift.
[0005] Based on this, how to accurately analyze the antigenic relationship between virus strains to lay the foundation for subsequent vaccine development is an urgent problem to be solved. Summary of the invention
[0006] The embodiments of this specification provide a data prediction method and device for influenza virus antigen relationship to partially solve the problems existing in the above-mentioned prior art.
[0007] The embodiments of this specification adopt the following technical solutions: This specification provides a data prediction method for influenza virus antigen relationship, including: Read the strain data of each sample strain pair, and for each sample strain pair, there are two different strains in the sample strain pair; For each sample strain pair, determine the virus characteristic data of the sample strain pair according to the strain data corresponding to the sample strain pair; For each sample strain pair, determine the neutralization rate of the sample strain pair, and determine the actual antigenic relationship of the sample strain pair based on the neutralization rate and the determined target evaluation threshold, the neutralization rate is used to characterize the relative size of the neutralization ability of the serum corresponding to one strain in the sample strain pair and the neutralization ability of the serum on the other strain in the sample strain pair, the actual antigenic relationship is used to characterize whether the two corresponding strains in the sample strain pair are similar, and input the virus feature data of each sample strain pair into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model; The loss value is determined based on the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, and the prediction model is trained based on the loss value to predict the antigenic relationship through the trained prediction model, and the loss value is positively correlated with the deviation.
[0008] Optionally, for each sample strain pair, determining the virus characteristic data of the sample strain pair according to the strain data corresponding to the sample strain pair specifically includes: For each sample strain pair, the basic physicochemical characteristics and structural correlation characteristics of the sample strain pair are determined based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the basic physicochemical characteristics are used to characterize the charge, polarity, hydrogen bonding index, volume, hydrophobicity, accessible surface area and van der Waals volume of the sample strain pair, and the structural correlation characteristics are used to characterize the structural correlation matrix of the sample strain pair, and the elements in the structural correlation matrix are used to characterize the similarity coefficients based on the folding angle between different amino acid residues on the protein main chain of the sample strain pair; For each sample strain pair, the virus characteristic data of the sample strain pair is determined based on the basic physicochemical characteristics and the structural correlation characteristics.
[0009] Optionally, before determining the virus feature data of each sample strain pair according to the basic physicochemical characteristics and the structural correlation characteristics, the method further includes: For each sample strain pair, the amino acid coding feature of the sample strain pair is determined according to the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the amino acid coding feature is used to characterize the coding information obtained after encoding the hemagglutinin amino acid sequences of the two corresponding strains in the sample strain pair based on a preset coding rule; For each sample strain pair, the virus characteristic data of the sample strain pair is determined based on the basic physicochemical characteristics and the structural correlation characteristics, specifically including: For each sample strain pair, the virus characteristic data of the sample strain pair is determined based on the basic physicochemical characteristics, the structural correlation characteristics and the amino acid coding characteristics.
[0010] Optionally, determining a target evaluation threshold includes: For each sample strain pair, the genetic distance of the sample strain pair is determined based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the genetic distance is used to characterize the difference between the hemagglutinin amino acid sequences of the two corresponding strains in the sample strain pair; For each sample strain pair, determine the first auxiliary antigen relationship of the sample strain pair according to the genetic distance, and obtain each candidate evaluation threshold, and for each candidate evaluation threshold, determine the control antigen relationship of the sample strain pair assuming that the candidate evaluation threshold is used to determine the antigen relationship according to the candidate evaluation threshold and the neutralization rate of the sample strain pair; For each candidate evaluation threshold, determine a first evaluation score corresponding to the candidate evaluation threshold according to the control antigen relationship of each sample strain pair and the first auxiliary antigen relationship of each sample strain pair; A target evaluation threshold is determined according to the first evaluation scores corresponding to each candidate evaluation threshold.
[0011] Optionally, for each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the first auxiliary antigen relationship of each sample strain pair, determining a first evaluation score corresponding to the candidate evaluation threshold specifically includes: For each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the first auxiliary antigen relationship of each sample strain pair, determine the first accuracy rate and the first balance value corresponding to the candidate evaluation threshold, wherein the first accuracy rate is determined according to the proportion of the first auxiliary antigen relationship in the control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship, and the first balance value is determined according to the number of the first auxiliary antigen relationship in each control antigen relationship of each sample strain pair and the number of the first auxiliary antigen relationship in each control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship; For each candidate evaluation threshold, a first evaluation score corresponding to the candidate evaluation threshold is determined according to a first accuracy rate and a first balance value corresponding to the candidate evaluation threshold.
[0012] Optionally, before determining the target evaluation threshold according to the first evaluation scores corresponding to the candidate evaluation thresholds, the method further includes: Deploy preset auxiliary prediction models; For each sample strain pair, extract the sequence features of the sample strain pair based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair; For each auxiliary prediction model, the sequence feature is input into the auxiliary prediction model to obtain the second auxiliary antigen relationship of the sample strain pair output by the auxiliary prediction model, and for each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the second auxiliary antigen relationship of each sample strain pair output by the auxiliary prediction model, the second evaluation score of the candidate evaluation threshold for the auxiliary prediction model is determined; Determine the target evaluation threshold according to the first evaluation score corresponding to each candidate evaluation threshold, specifically including: For each candidate evaluation threshold, determine an average evaluation score corresponding to the candidate evaluation threshold according to the first evaluation score corresponding to the candidate evaluation threshold and each second evaluation score corresponding to the candidate evaluation threshold determined based on each auxiliary prediction model; The target evaluation threshold is determined according to the average evaluation scores corresponding to each candidate evaluation threshold.
[0013] Optionally, for each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the second auxiliary antigen relationship of each sample strain pair output by the auxiliary prediction model, determining the second evaluation score of the candidate evaluation threshold for the auxiliary prediction model specifically includes: For each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the second auxiliary antigen relationship of each sample strain pair output by the auxiliary prediction model, determine the second accuracy rate of the candidate evaluation threshold for the auxiliary prediction model and the second balance value for the auxiliary prediction model, wherein the second accuracy rate is determined according to the proportion of the second auxiliary antigen relationship output by the auxiliary prediction model in the control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship, and the second balance value is determined according to the number of the second auxiliary antigen relationship output by the auxiliary prediction model in each control antigen relationship of each sample strain pair and the number of the second auxiliary antigen relationship output by the auxiliary prediction model in each control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship; For each candidate evaluation threshold, a second evaluation score of the candidate evaluation threshold for the auxiliary prediction model is determined according to a second accuracy rate of the candidate evaluation threshold for the auxiliary prediction model and a second balance value for the auxiliary prediction model.
[0014] This specification provides a data prediction method for influenza virus antigen relationship, comprising: Determine the viral characteristic data of the target strain pair; Inputting the virus feature data into a pre-trained prediction model to predict the antigenic relationship of the target strain pair, wherein the prediction model is trained by the model training method in the above-mentioned data prediction method for influenza virus antigenic relationship, and the antigenic relationship is used to characterize whether the two corresponding strains in the target strain pair are similar; Business processing is performed according to the antigen relationship.
[0015] This specification provides a data prediction device for influenza virus antigen relationship, including: A reading module is used to read the strain data of each sample strain pair. For each sample strain pair, there are two different strains in the sample strain pair. A first determination module is used to determine, for each sample strain pair, virus characteristic data of the sample strain pair according to the strain data corresponding to the sample strain pair; A module is obtained, which is used to determine the neutralization rate of each sample strain pair, and determine the actual antigenic relationship of the sample strain pair according to the neutralization rate and the determined target evaluation threshold, wherein the neutralization rate is used to characterize the relative size of the neutralization ability of the serum corresponding to one strain in the sample strain pair and the neutralization ability of the serum to the other strain in the sample strain pair, and the actual antigenic relationship is used to characterize whether the two corresponding strains in the sample strain pair are similar, and the virus feature data of each sample strain pair is input into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model; The training module is used to determine the loss value according to the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, and train the prediction model according to the loss value to predict the antigenic relationship through the trained prediction model, and the loss value is positively correlated with the deviation.
[0016] The present specification provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned data prediction method for influenza virus antigen relationship is implemented.
[0017] At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects: In the data prediction method for influenza virus antigenic relationship provided in this specification, first, the strain data of each sample strain pair is read, and for each sample strain pair, there are two different strains in the sample strain pair. For each sample strain pair, the virus characteristic data of the sample strain pair is determined according to the strain data corresponding to the sample strain pair. For each sample strain pair, the neutralization rate of the sample strain pair is determined, and the actual antigenic relationship of the sample strain pair is determined according to the neutralization rate and the determined target evaluation threshold. At the same time, the virus characteristic data of each sample strain pair is input into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model. Then, according to the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, the loss value is determined, and the prediction model is trained to predict the antigenic relationship through the trained prediction model.
[0018] It can be seen from the above method that in the data prediction method for influenza virus antigenic relationships provided in this specification, for each sample strain pair, the actual antigenic relationship of the sample strain pair is determined by the neutralization rate of the sample strain pair and the determined target evaluation threshold. The actual antigenic relationship determined in this way is more accurate. Furthermore, by training the prediction model with a more accurate actual antigenic relationship, a prediction model with better prediction effect can be obtained, and the predicted antigenic relationship is also more accurate. In this way, a more accurate antigenic relationship between the two strains can be obtained, and based on the antigenic relationship between the two strains, an influenza vaccine that is more suitable for the current epidemic strains can be selected. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] The drawings described herein are used to provide a further understanding of this specification and constitute a part of this specification. The illustrative embodiments and descriptions of this specification are used to explain this specification and do not constitute an improper limitation on this specification. In the drawings: Figure 1 A schematic diagram of a flow chart of a data prediction method for influenza virus antigen relationship provided in this specification; Figure 2 A schematic diagram of a flow chart of a data prediction method for influenza virus antigen relationship provided in this specification; Figure 3 A schematic diagram of a data prediction device for influenza virus antigen relationship provided in this specification; Figure 4 A schematic diagram of a data prediction device for influenza virus antigen relationship provided in this specification; Figure 5 A method corresponding to the Figure 1 or Figure 2 Schematic diagram of electronic equipment. DETAILED DESCRIPTION
[0020] In order to make the purpose, technical solutions and advantages of this specification more clear, the technical solutions of this specification will be clearly and completely described below in combination with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.
[0021] Influenza is a disease caused by influenza virus, which belongs to the family Orthomyxoviridae and is an enveloped RNA virus. Its genome consists of eight segments of negative-strand RNA, encoding a variety of important proteins including hemagglutinin and neuraminidase. Due to the high variability of hemagglutinin and neuraminidase, influenza virus can evade recognition by the immune system, leading to persistent influenza epidemics. The most effective way to deal with influenza is to get vaccinated against influenza, which works by stimulating the body to produce a specific immune response to influenza virus antigens, thereby reducing the spread and severity of the disease.
[0022] In the process of developing influenza vaccines, the influenza vaccines that respond to the epidemic strains are determined by analyzing the similarities between vaccine strains (strains that have appeared in history) and epidemic strains (strains that have recently appeared and spread). However, due to the high variability of influenza viruses, it is very difficult to determine vaccine strains that are similar to current and future epidemic strains.
[0023] At present, the hemagglutination inhibition (HI) test is generally used to determine the antigenic variation of influenza viruses, so as to decide whether it is necessary to replace the influenza vaccine corresponding to the epidemic strain. Among them, viruses with hemagglutinin (HA) can agglutinate human or animal red blood cells, which is called hemagglutination. The hemagglutination phenomenon can be inhibited by corresponding antibodies, which is called HI test. HI test is a technology for measuring antigens or antibodies. However, the HI test cycle is long, and the sensitivity and accuracy are limited. Especially when dealing with new strains, the effect of predicting the antigenic relationship between two strains is not ideal.
[0024] Based on this, the present specification provides a data prediction method for influenza virus antigenic relationships. First, the strain data of each sample strain pair is read. For each sample strain pair, there are two different strains in the sample strain pair. For each sample strain pair, the virus characteristic data of the sample strain pair is determined according to the strain data corresponding to the sample strain pair. For each sample strain pair, the neutralization rate of the sample strain pair is determined, and the actual antigenic relationship of the sample strain pair is determined based on the neutralization rate and the determined target evaluation threshold. At the same time, the virus characteristic data of each sample strain pair is input into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model. Then, according to the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, the loss value is determined, and the prediction model is trained to predict the antigenic relationship through the trained prediction model.
[0025] It can be seen from the above method that the method of this specification can determine the actual antigenic relationship of each sample strain pair based on the neutralization rate of each sample strain pair and the determined target evaluation threshold, and train the prediction model based on the actual antigenic relationship to obtain a prediction model with more accurate output results, and the predicted antigenic relationship is also more accurate. In this way, a more accurate antigenic relationship between the two strains can be obtained, and based on the antigenic relationship between the two strains, a flu vaccine that is more suitable for the current epidemic strains can be selected.
[0026] The technical solutions provided by the embodiments of this specification are described in detail below in conjunction with the accompanying drawings.
[0027] Figure 1 A schematic diagram of a data prediction method for influenza virus antigen relationship provided in this specification, comprising the following steps: S101: Read the strain data of each sample strain pair. For each sample strain pair, there are two different strains in the sample strain pair.
[0028] In this specification, the execution subject for implementing a data prediction method for influenza virus antigen relationship can be a terminal device such as a laptop computer, a tablet computer, or of course, a server. For the sake of ease of description, this specification only takes the server as an execution subject as an example to illustrate a data prediction method for influenza virus antigen relationship provided in this specification.
[0029] The server reads the strain data of each sample strain pair, and for each sample strain pair, there are two different strains in the sample strain pair. The strain data is the data of two strains, for example, it can be the hemagglutinin amino acid sequence data of the two strains. Hemagglutinin refers to hemagglutinin, which is columnar and can bind to receptors on the surface of red blood cells of animals such as humans, birds, pigs, and guinea pigs to cause coagulation. The hemagglutinin amino acid sequence is the basis for its key role in viral infection and immune response.
[0030] It should be noted that the two different strains corresponding to the sample strain pair can be strains for influenza virus or strains for other viruses, such as strains for hepatitis virus (including type A, type B, type C, etc.), herpes virus, human immunodeficiency virus and other viruses.
[0031] S102: For each sample strain pair, determine the virus characteristic data of the sample strain pair according to the strain data corresponding to the sample strain pair.
[0032] For each sample strain pair, the server determines the virus characteristic data of the sample strain pair based on the strain data corresponding to the sample strain pair through a preset script.
[0033] Specifically, for each sample strain pair, the server determines the basic physicochemical characteristics of the sample strain pair based on the hemagglutinin amino acid sequence data of the corresponding two strains in the sample strain pair, where the basic physicochemical characteristics are used to characterize the charge, polarity, hydrogen bonding index, volume, hydrophobicity, accessible surface area and van der Waals volume of the sample strain pair.
[0034] It should be noted that the charge here refers to the charge state of the sample strain pair, which may vary depending on the charge distribution of proteins, nucleic acids and other biological molecules on the surface of the two corresponding strains in the sample strain pair; polarity is related to the charge distribution of the two corresponding strains in the sample strain pair; the hydrogen bond index reflects the number and strength of hydrogen bonds in the two corresponding strains in the sample strain pair, among which the strength of hydrogen bonds is a standard for measuring the strength of hydrogen bond interaction, which reflects the ability of hydrogen bonds to exist stably between or within molecules, and the ability to resist external interference (such as temperature, pressure, solvent, etc.); volume is related to the size and shape of the two corresponding strains in the sample strain pair; hydrophobicity refers to the measure of the ability of the two corresponding strains in the sample strain pair to interact with water molecules; accessible surface area refers to the surface area of the two corresponding strains in the sample strain pair exposed to the solvent (such as water). Van der Waals volume refers to the spatial volume occupied by atoms or molecular groups of the two corresponding strains in the sample strain pair.
[0035] Furthermore, for each sample strain pair, the server determines the virus characteristic data of the sample strain pair based on the basic physical and chemical characteristics of the sample strain pair.
[0036] In order to further obtain more accurate virus feature data for each sample strain pair, the server can also combine the basic physical and chemical characteristics of each sample strain pair and the structural correlation characteristics of each sample strain pair to determine the virus feature data of each sample strain pair. Among them, the structural correlation characteristics are used to characterize the structural correlation matrix of the sample strain pair, and the elements in the structural correlation matrix are used to characterize the similarity coefficients based on the folding angle between different amino acid residues on the protein main chain of the sample strain pair. For example, at this time there is a 2*2 structural correlation matrix, which can be expressed as:
[0037] Among them, each value (a, b, c, d) in the above 2*2 matrix is an element, and for each value, the value is used to characterize the similarity coefficient based on the folding angle between different amino acid residues on the protein main chain of the sample strain pair.
[0038] Then, for each sample strain pair, the server determines the virus characteristic data of the sample strain pair based on the basic physicochemical characteristics of the sample strain pair and the structural correlation characteristics of the sample strain pair.
[0039] On this basis, the server can further improve the virus feature data of each sample strain pair to obtain more accurate virus feature data for each sample strain pair. Specifically, for each sample strain pair, the server determines the amino acid coding feature of the sample strain pair based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the amino acid coding feature is used to characterize the coding information obtained after encoding the hemagglutinin amino acid sequences of the two corresponding strains in the sample strain pair based on the preset coding rules, and the preset coding rules here can be to use 21 consecutive decimal values from 0 to 1 to encode different types of amino acids (a total of 20 types) and amino acid deletions (indicated by "-") in the two corresponding strains in each sample strain pair.
[0040] For each sample strain pair, the server determines the virus characteristic data of the sample strain pair based on the basic physicochemical characteristics, structural correlation characteristics and amino acid coding characteristics of the sample strain pair.
[0041] It should be noted that the above-mentioned basic physicochemical characteristics, structure-related characteristics, and amino acid coding characteristics can be used to determine the virus characteristic data individually, in pairs, or all three characteristics can be used together to determine the virus characteristic data, and no specific limitation is made here.
[0042] S103: For each sample strain pair, determine the neutralization rate of the sample strain pair, and determine the actual antigenic relationship of the sample strain pair based on the neutralization rate and the determined target evaluation threshold, and input the virus feature data of each sample strain pair into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model.
[0043] For each sample strain pair, the server determines the neutralization rate of the sample strain pair, which is used to characterize the relative size of the neutralization ability of the serum corresponding to one strain in the sample strain pair to the strain and the neutralization ability of the serum corresponding to the strain to the other strain in the sample strain pair. Here, the serum corresponding to the strain actually refers to the serum induced by the strain. The serum induced by the strain means that when an organism (such as a human or an animal) is infected with the strain, its immune system will recognize and fight against the strain, thereby producing a specific immune response. In this process, the organism will produce antibodies against the virus strain, and these antibodies are present in the serum of the organism.
[0044] The neutralization rate of the above sample strain can be expressed by the formula:
[0045] Among them, i refers to one strain in the sample strain pair, j refers to the other strain in the sample strain pair, It represents the relative size of the neutralizing ability of the serum corresponding to strain i against strain i and the neutralizing ability of the serum corresponding to strain i against strain j. It represents the titration value of the serum corresponding to strain i against the pseudovirus i of strain i, that is, under specific experimental conditions, the result obtained when measuring the neutralization ability of the serum corresponding to strain i against the pseudovirus i of strain i. The titration value here reflects the concentration or titer of the neutralizing antibodies in the serum corresponding to strain i (in the field of immunology, titer usually refers to the ability or concentration of antibody solution to inhibit the growth of pathogens), that is, how much pseudovirus i can be neutralized by the serum corresponding to strain i. It represents the titration value of the serum corresponding to strain i against the pseudovirus j of strain j, that is, under specific experimental conditions, the result obtained when measuring the neutralization ability of the serum corresponding to strain i against the pseudovirus j of strain j. The titration value here reflects the concentration or titer of the neutralizing antibodies in the serum corresponding to strain i, that is, how much pseudovirus j can be neutralized by the serum corresponding to strain i.
[0046] It should be noted that , The value of is determined by the pseudovirus-based neutralizing antibody assay (PN), which uses pseudoviruses to simulate the infection process of the virus and evaluates the titer of neutralizing antibodies by detecting the blocking effect of neutralizing antibodies on pseudovirus infection. The specific steps are: select an appropriate retrovirus or lentivirus vector system, integrate the envelope protein gene of the target virus into the vector, and construct a pseudovirus with exogenous envelope protein. These pseudoviruses can simulate the infection process of the target virus, but due to the lack of self-replication ability, they have high biosafety. Then, neutralizing antibodies are extracted from the sample to be tested (such as serum, plasma, etc.), which may be produced by natural infection or vaccination. Further, the pseudovirus is mixed with neutralizing antibodies of different dilutions and then inoculated on susceptible cells. The infection of pseudoviruses on susceptible cells can be judged by observing the susceptible cell pathological effect, fluorescent signal or chemiluminescent signal. Finally, the titer of neutralizing antibodies is determined based on the degree of inhibition of infection. The higher the titer of neutralizing antibodies, the stronger their ability to inhibit pseudovirus infection and the greater their protective effect on the body. Usually, the neutralizing antibody dilution required to inhibit 50% pseudovirus infection is defined as the neutralizing antibody titer.
[0047] For determination value, firstly, select an appropriate retrovirus or lentivirus vector system, integrate the envelope protein gene of virus i into the vector, construct a pseudovirus i with exogenous envelope protein, extract neutralizing antibodies from the serum induced by virus i, mix the pseudovirus i with neutralizing antibodies of different dilutions, and then inoculate them on susceptible cells, judge the infection of pseudovirus i on susceptible cells by observing susceptible cytopathic effect, fluorescent signal or chemiluminescent signal, and determine the titer of neutralizing antibody according to the degree of inhibition of infection, that is, determine the titration value of serum corresponding to strain i against pseudovirus i of strain i according to the degree of inhibition of infection of pseudovirus i.
[0048] For determination value, firstly, select an appropriate retrovirus or lentivirus vector system, integrate the envelope protein gene of virus j into the vector, construct a pseudovirus j with exogenous envelope protein, extract neutralizing antibodies from the serum induced by virus i, mix the pseudovirus j with neutralizing antibodies of different dilutions, and then inoculate them on susceptible cells, judge the infection of pseudovirus j on susceptible cells by observing susceptible cytopathic effect, fluorescent signal or chemiluminescent signal, and determine the titer of neutralizing antibody according to the degree of inhibition of infection, that is, determine the titration value of pseudovirus j of strain j by serum corresponding to strain i according to the degree of inhibition of infection of pseudovirus j.
[0049] Furthermore, for each sample strain pair, the server determines the actual antigenic relationship of the sample strain pair based on the neutralization rate of the sample strain pair and the determined target evaluation threshold, and the actual antigenic relationship is used to characterize whether the two corresponding strains in the sample strain pair are similar. In this specification, if the neutralization rate of the sample strain pair is not greater than the determined target evaluation threshold, the two corresponding strains in the sample strain pair are considered to be similar, and the actual antigenic relationship of the sample strain pair can be recorded as "1"; if the neutralization rate of the sample strain pair is greater than the determined target evaluation threshold, the two corresponding strains in the sample strain pair are considered to be dissimilar, and the actual antigenic relationship of the sample strain pair can be recorded as "0".
[0050] In this specification, the target evaluation threshold may be determined by the server based on the genetic distance of each sample strain pair. Specifically, for each sample strain pair, the server determines the genetic distance of the sample strain pair based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the genetic distance is used to characterize the difference between the hemagglutinin amino acid sequences of the two corresponding strains in the sample strain pair. When determining the genetic distance of the sample strain pair, the server may first use a multiple sequence alignment algorithm (such as ClustalW, MAFFT, etc.) to align the hemagglutinin amino acid sequences of the two strains to obtain the number of different amino acids between the hemagglutinin amino acid sequences of the two strains after alignment.
[0051] Then, for each sample strain pair, the server determines the first auxiliary antigen relationship of the sample strain pair based on the genetic distance of the sample strain pair, and the first auxiliary antigen relationship is used to characterize whether the two corresponding strains in the sample strain pair are similar. Specifically, the server can set a threshold value, and when the genetic distance is not less than the threshold value, the server considers that the two corresponding strains in the sample strain pair are not similar, and when the genetic distance is less than the threshold value, the server considers that the two corresponding strains in the sample strain pair are similar.
[0052] Next, the server obtains each candidate evaluation threshold (which can be set according to the neutralization rate of each sample strain pair. For example, if the neutralization rate of each sample strain pair is distributed between 0-50, then each candidate evaluation threshold is an integer between 0-50, that is, there are 51 candidate evaluation thresholds in total), and for each candidate evaluation threshold, based on the candidate evaluation threshold and the neutralization rate of the sample strain pair, determines the control antigen relationship of the sample strain pair assuming that the candidate evaluation threshold is used to determine the antigen relationship.
[0053] For example, when the candidate evaluation threshold is 5, if the neutralization rate of the sample strain pair is not greater than 5, it is considered that the two corresponding strains in the sample strain pair are similar, and the control antigen relationship of the sample strain pair can be recorded as "1"; if the neutralization rate of the sample strain pair is greater than 5, it is considered that the two corresponding strains in the sample strain pair are not similar, and the control antigen relationship of the sample strain pair can be recorded as "0".
[0054] For each candidate evaluation threshold, the server determines the first evaluation score corresponding to the candidate evaluation threshold based on the control antigen relationship of each sample strain pair and the first auxiliary antigen relationship of each sample strain pair. In other words, the first evaluation score is used to characterize the proportion of sample strain pairs whose control antigen relationship is consistent with the first auxiliary antigen relationship in each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship. For example, there are 100 sample strain pairs, of which 60 sample strain pairs have a control antigen relationship that is consistent with the first auxiliary antigen relationship, then the first evaluation score corresponding to the candidate evaluation threshold is 60%. Therefore, this evaluation score actually also reflects the accuracy of using the candidate evaluation threshold to determine the antigen relationship.
[0055] Next, the server determines the target evaluation threshold according to the first evaluation scores corresponding to the candidate evaluation thresholds. In other words, the server uses the candidate evaluation threshold corresponding to the first evaluation score with the highest value as the target evaluation threshold.
[0056] Of course, in order to obtain a more accurate target evaluation threshold, the first evaluation score in this specification may include two parts: a first accuracy rate and a first balance value. The first accuracy rate is determined based on the proportion of the first auxiliary antigen relationship in the control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship, and the first balance value is determined based on the number of first auxiliary antigen relationships in each control antigen relationship of each sample strain pair and the number of non-first auxiliary antigen relationships in each control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship.
[0057] Specifically, the first accuracy can be expressed by referring to the following formula:
[0058] The number of correctly predicted sample strain pairs refers to the number of sample strain pairs that have the first auxiliary antigen relationship in the control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship.
[0059] The first balance value can be expressed by referring to the following formula: ; ; ; Among them, for each sample strain pair, the sample strain pair corresponds to a control antigen relationship and a first auxiliary antigen relationship. Assuming that the candidate evaluation threshold is used to determine the antigen relationship, a true positive refers to a sample strain pair whose first auxiliary antigen relationship is a similar relationship and whose control antigen relationship is also a similar relationship, and a true positive TP is the number of sample strain pairs with true positives; a false positive refers to a sample strain pair whose first auxiliary antigen relationship is a dissimilar relationship but whose control antigen relationship is a similar relationship, and a false positive FP is the number of sample strain pairs with false positives; a false negative refers to a sample strain pair whose first auxiliary antigen relationship is a similar relationship but whose control antigen relationship is a dissimilar relationship, and a false negative FN is the number of sample strain pairs with false negatives.
[0060] Then, the server determines a first evaluation score according to the first accuracy rate and the first balance value. For example, the first evaluation score is obtained by multiplying the first accuracy rate and the first balance value; or for another example, the first evaluation score is obtained by adding the first accuracy rate and the first balance value.
[0061] In addition, in order to obtain a more accurate target evaluation threshold, the server in this specification can also determine the second auxiliary antigen relationship of each sample strain pair through the preset auxiliary prediction models, obtain the second evaluation score of each candidate evaluation threshold for each auxiliary prediction model, and determine the target evaluation threshold based on the first evaluation score and the second evaluation score.
[0062] Specifically, the server deploys preset auxiliary prediction models, and for each sample strain pair, extracts the sequence features of the sample strain pair based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair.
[0063] It should be noted that the sequence feature extraction process here may be different from the virus feature data extraction process mentioned in the above process. This is because, for different auxiliary prediction models, the server determines the sequence features of each sample strain pair in different processes, so that the determined sequence features of each sample strain pair are more suitable for different auxiliary prediction models, so that the auxiliary prediction model outputs more accurate results.
[0064] For example, the auxiliary prediction model can be a two-dimensional convolutional neural network model for predicting influenza virus antigen variants, in which the server determines the sequence features of the sample strain pair based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the sequence features are used to characterize the amino acid coding information and protein three-dimensional structure information of the sample strain pair. The server inputs the sequence features of the sample strain pair into the auxiliary prediction model so that the auxiliary prediction model outputs the antigenic relationship of the sample strain pair; For another example, the auxiliary prediction model can be a general computational model of influenza A virus antigen variants based on conservative antigenic structures, in which the server determines the sequence features of the sample strain pair based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the sequence features are used to characterize the amino acid coding information, secondary structure information, and conserved region information of the sample strain pair. The secondary structure information here is used to characterize the spatial structure of the local main chain in the protein molecule of the sample strain pair, mainly including structural elements such as α-helix, β-fold, β-turn, and random curl, while the conserved region information is used to characterize the amino acid sequence region of the sample strain pair that remains relatively stable during the evolution of the virus. The server inputs the sequence features of the sample strain pair into the auxiliary prediction model so that the auxiliary prediction model outputs the antigenic relationship of the sample strain pair.
[0065] Furthermore, for each auxiliary prediction model, the server inputs the sequence features into the auxiliary prediction model to obtain the second auxiliary antigen relationship of the sample strain pair output by the auxiliary prediction model. The second auxiliary antigen relationship is used to characterize whether the corresponding two strains in the sample strain pair output by the auxiliary prediction model are similar.
[0066] For each auxiliary prediction model, it is necessary to determine the second evaluation score of each candidate evaluation threshold for the auxiliary prediction model. Specifically, for each candidate evaluation threshold, the server determines the second evaluation score of the candidate evaluation threshold for the auxiliary prediction model based on the control antigen relationship of each sample strain pair and the second auxiliary antigen relationship of each sample strain pair output by the auxiliary prediction model. In other words, the second evaluation score is used to characterize the proportion of sample strain pairs whose control antigen relationship in each sample strain pair is consistent with the second auxiliary antigen relationship output by the auxiliary prediction model, assuming that the candidate evaluation threshold is used to determine the antigen relationship. For example, there are 100 sample strain pairs, of which 75 sample strain pairs have a control antigen relationship that is consistent with the second auxiliary antigen relationship output by the auxiliary prediction model, then the second evaluation score of the candidate evaluation threshold for the auxiliary prediction model is 75%. Therefore, the second evaluation score can actually also reflect the accuracy of using the candidate evaluation threshold to determine the antigen relationship.
[0067] Next, for each candidate evaluation threshold, the average evaluation score corresponding to the candidate evaluation threshold is determined based on the first evaluation score corresponding to the candidate evaluation threshold and the second evaluation scores corresponding to the candidate evaluation threshold determined based on the auxiliary prediction models. For example, if there are six auxiliary prediction models, for each candidate evaluation threshold, the candidate evaluation threshold corresponds to one first evaluation score and six second evaluation scores, and the server determines the average of the seven values to obtain the average evaluation score corresponding to the candidate evaluation threshold.
[0068] Then, the server uses the candidate evaluation threshold corresponding to the average evaluation score with the highest value as the target evaluation threshold based on the average evaluation scores corresponding to the candidate evaluation thresholds.
[0069] Of course, in order to obtain a more accurate target evaluation threshold, the second evaluation score in this specification may also include two parts: a second accuracy rate and a second balance value. Among them, the second accuracy rate is determined based on the proportion of the second auxiliary antigen relationship output by the auxiliary prediction model in the control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship, and the second balance value is determined based on the number of second auxiliary antigen relationships output by the auxiliary prediction model in each control antigen relationship of each sample strain pair and the number of second auxiliary antigen relationships not output by the auxiliary prediction model in each control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship.
[0070] Specifically, the second accuracy can be expressed by referring to the following formula:
[0071] Among them, for each auxiliary prediction model, the number of sample strain pairs predicted correctly refers to the number of sample strain pairs in which the second auxiliary antigen relationship output by the auxiliary prediction model appears in the control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship.
[0072] The second balance value can be expressed by referring to the following formula: ; ; ; Among them, for each auxiliary prediction model, each sample strain pair corresponds to a control antigen relationship and a second auxiliary antigen relationship. Assuming that the candidate evaluation threshold is used to determine the antigen relationship, a true positive refers to a sample strain pair whose second auxiliary antigen relationship is a similar relationship and whose control antigen relationship is also a similar relationship. The true positive TP is the number of sample strain pairs with true positives; a false positive refers to a sample strain pair whose second auxiliary antigen relationship is a dissimilar relationship but whose control antigen relationship is a similar relationship. The false positive FP is the number of sample strain pairs with false positives; a false negative refers to a sample strain pair whose second auxiliary antigen relationship is a similar relationship but whose control antigen relationship is a dissimilar relationship. The false negative FN is the number of sample strain pairs with false negatives.
[0073] Then, the server determines a second evaluation score according to the second accuracy rate and the second balance value. For example, the second accuracy rate is multiplied by the second balance value to obtain the second evaluation score; for another example, the second accuracy rate is added to the second balance value to obtain the second evaluation score.
[0074] While determining the actual antigenic relationship of each sample strain pair, the server inputs the virus feature data of each sample strain pair into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model.
[0075] It should be noted that the above process describes the determination of the first balance value or the second balance value based on the precision and recall rate. In this specification, assuming that the candidate evaluation threshold is used to determine the antigen relationship, the server actually determines the first balance value based on the number of first auxiliary antigen relationships in each control antigen relationship of each sample strain pair and the number of non-first auxiliary antigen relationships in each control antigen relationship of each sample strain pair. Therefore, the first balance value can be determined by the precision rate or the recall rate alone. For example, the server can directly use the precision rate as the first balance value or directly use the recall rate as the first balance value, and can also obtain the first evaluation score, but the first balance value determined by the precision rate and the recall rate is relatively more accurate, thereby making the first evaluation score determined by the first balance value better. The same is true for the determination of the second balance value. The second balance value can be determined by the precision rate or the recall rate alone. For example, the server can directly use the precision rate as the second balance value or directly use the recall rate as the second balance value.
[0076] S104: Determine a loss value based on the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, and train the prediction model based on the loss value to perform antigenic relationship prediction through the trained prediction model.
[0077] The server determines the loss value based on the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, and trains the prediction model based on the loss value so that the predicted antigenic relationship output by the trained prediction model is consistent with the actual antigenic relationship, wherein the loss value is positively correlated with the deviation. Next, the server can predict the antigenic relationship of the strain pair using the trained prediction model.
[0078] Among them, the prediction model can be a model built in combination with a random forest algorithm. Random forest is an integrated learning method that improves the prediction performance and robustness of the model by constructing multiple decision trees and combining their prediction results.
[0079] Specifically, the server constructs the original training set based on the virus feature data of each sample strain pair and the corresponding actual antigen relationship. The server can randomly extract samples with replacement from the original training set to construct multiple training subsets, each of which will be used to train a decision tree in the random forest, which together form the random forest model. Before training the random forest model, the server also needs to set some key parameters, such as the number of decision trees, maximum depth, minimum number of sample splits, etc. During the training process, the random forest will use the prediction results of multiple decision trees to vote or average to obtain the final prediction result.
[0080] In this specification, for each sample strain pair, the actual antigenic relationship of the sample strain pair is determined by the neutralization rate of the sample strain pair and the determined target evaluation threshold. The actual antigenic relationship obtained in this way is more accurate, that is, it can clearly distinguish whether the corresponding two strains in the sample strain pair are similar. Based on the more accurate actual antigenic relationship, the prediction model is trained, and the obtained prediction model is also more accurate, and the antigenic relationship predicted by the prediction model is also more accurate.
[0081] Moreover, when determining the virus characteristic data of each sample strain pair, this instruction manual can determine it based on the basic physicochemical characteristics, structural correlation characteristics and amino acid coding characteristics of each sample strain pair. The virus characteristic data determined in this way is more accurate and can better reflect the characteristics of each sample strain pair.
[0082] In addition, in the process of determining the target evaluation threshold, an average evaluation threshold can be determined according to the first evaluation score and the second evaluation score to obtain a more accurate target evaluation threshold. The first evaluation score and the second evaluation score can also include an accuracy rate and a balance value to avoid a single determination method resulting in the first evaluation score and the second evaluation score being inaccurate.
[0083] According to the above-mentioned data prediction method for influenza virus antigen relationship, a data prediction method for influenza virus antigen relationship can also be obtained, and the specific process is shown in the figure below.
[0084] Figure 2 A schematic diagram of a data prediction method for influenza virus antigen relationship provided in this specification, comprising the following steps: S201: Determine the virus characteristic data of the target strain pair.
[0085] In this specification, the execution subject for implementing a data prediction method for influenza virus antigen relationship can be a terminal device such as a laptop computer, a tablet computer, or of course, a server. For the sake of ease of description, this specification only takes the server as an execution subject as an example to illustrate a data prediction method for influenza virus antigen relationship provided in this specification.
[0086] The server determines the basic physicochemical characteristics, structural correlation characteristics and amino acid coding characteristics of the target strain pair according to the hemagglutinin amino acid sequence data of the two corresponding strains in the target strain pair through a preset script. The two corresponding strains in the target strain pair here can be strains for influenza virus or strains for other viruses, such as strains for hepatitis virus (including type A, type B, type C, etc.), herpes virus, human immunodeficiency virus and other viruses. Then the server determines the virus feature data of the target strain pair according to the basic physicochemical characteristics, structural correlation characteristics and amino acid coding characteristics of the target strain pair. The specific process has been described in detail in S102 and will not be repeated here.
[0087] S202: Inputting the virus feature data into a pre-trained prediction model to predict the antigenic relationship of the target strain pair, wherein the prediction model is trained by the model training method in the above-mentioned data prediction method for influenza virus antigenic relationship.
[0088] The server inputs the virus feature data of the target strain pair into a pre-trained prediction model to predict the antigenic relationship of the target strain pair, wherein the prediction model is trained using the model training method in the above-mentioned data prediction method for influenza virus antigenic relationship, and the antigenic relationship is used to characterize whether the two corresponding strains in the target strain pair are similar.
[0089] S203: Execute business processing according to the antigen relationship.
[0090] The server performs business processing according to the antigenic relationship of the target strain pair. The business processing performed here can be that if the two corresponding strains in the target strain pair come from different regions and different times, the server can build an evolutionary tree of the virus by comparing the antigenic relationship of strains in different regions and different time points, and then display information that can represent the history of virus transmission and mutation trends to the user; it can also be that if the two corresponding strains in the target strain pair are vaccine strains and epidemic strains for influenza virus, respectively, the server can determine the antigenic relationship between the vaccine strain and the epidemic strain according to the antigenic relationship of the target strain pair, and obtain evaluation information. The evaluation information here is used to evaluate the matching and protective efficacy of the influenza vaccine corresponding to the vaccine strain to the epidemic strain, so as to generate prompt information based on the evaluation information to prompt the user whether the influenza vaccine needs to be replaced or a new influenza vaccine needs to be developed.
[0091] In this manual, through the pre-trained prediction model, a more accurate antigenic relationship of the target strain pair can be output, and based on the more accurate antigenic relationship, the mutation trend of the influenza virus can be better predicted, and the user can be promptly prompted to replace the influenza vaccine or develop a new influenza vaccine.
[0092] Furthermore, when determining the virus characteristic data of a target strain pair, the method provided in this specification can be determined based on the basic physicochemical characteristics, structural correlation characteristics, and amino acid coding characteristics of the target strain pair. The virus characteristic data determined in this way is more accurate and can better reflect the characteristics of the target strain pair.
[0093] The above are one or more implementations of the data prediction method for influenza virus antigen relationship of this specification. Based on the same idea, this specification also provides a corresponding data prediction device for influenza virus antigen relationship, such as Figure 3 , Figure 4 shown.
[0094] Figure 3 A schematic diagram of a data prediction device for influenza virus antigen relationship provided in this specification, comprising: A reading module 301 is used to read the strain data of each sample strain pair. For each sample strain pair, there are two different strains in the sample strain pair. A first determination module 302 is used to determine, for each sample strain pair, virus characteristic data of the sample strain pair according to the strain data corresponding to the sample strain pair; Obtaining module 303, for each sample strain pair, determining the neutralization rate of the sample strain pair, and determining the actual antigenic relationship of the sample strain pair according to the neutralization rate and the determined target evaluation threshold, wherein the neutralization rate is used to characterize the relative magnitude of the neutralization ability of the serum corresponding to one strain in the sample strain pair to the strain and the neutralization ability of the serum to the other strain in the sample strain pair, and the actual antigenic relationship is used to characterize whether the two corresponding strains in the sample strain pair are similar, and inputting the virus feature data of each sample strain pair into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model; The training module 304 is used to determine the loss value according to the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, and train the prediction model according to the loss value to predict the antigenic relationship through the trained prediction model, and the loss value is positively correlated with the deviation.
[0095] Figure 4 A schematic diagram of a data prediction device for influenza virus antigen relationship provided in this specification, comprising: The second determination module 401 is used to determine the virus characteristic data of the target strain pair; An input module 402 is used to input the virus feature data into a pre-trained prediction model to predict the antigenic relationship of the target strain pair, wherein the prediction model is trained by the model training method in the above-mentioned data prediction method for influenza virus antigenic relationship, and the antigenic relationship is used to characterize whether the two corresponding strains in the target strain pair are similar; The execution module 403 is used to execute business processing according to the antigen relationship.
[0096] This specification also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 or Figure 2 A data prediction method for influenza virus antigenic relationships is provided.
[0097] This manual also provides Figure 5 The one shown corresponds to Figure 1 or Figure 2 A schematic diagram of the electronic device. Figure 5 As mentioned above, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 or Figure 2The data prediction method for influenza virus antigen relationship. Of course, in addition to the software implementation, this specification does not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.
[0098] For the improvement of a technology, it can be clearly distinguished whether it is a hardware improvement (for example, improvement of the circuit structure of diodes, transistors, switches, etc.) or a software improvement (improvement of the method flow). However, with the development of technology, many improvements of the method flow today can be regarded as direct improvements of the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented with a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming themselves, without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one kind of HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.
[0099] The controller may be implemented in any suitable manner, for example, the controller may take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, and the memory controller may also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that, in addition to implementing the controller in a purely computer-readable program code manner, the controller may be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller may be considered as a hardware component, and the devices for implementing various functions included therein may also be considered as structures within the hardware component. Or even, the devices for implementing various functions may be considered as both software modules for implementing the method and structures within the hardware component.
[0100] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.
[0101] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.
[0102] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0103] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0104] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0106] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0107] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0108] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0109] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0110] It should be understood by those skilled in the art that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0111] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.
[0112] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0113] The above description is only an embodiment of the present specification and is not intended to limit the present specification. For those skilled in the art, the present specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification shall be included in the scope of the claims of the present specification.
Claims
1. A data prediction method for influenza virus antigen relationship, characterized in that: include: Read the strain data of each sample strain pair, and for each sample strain pair, there are two different strains in the sample strain pair; For each sample strain pair, determine the virus characteristic data of the sample strain pair according to the strain data corresponding to the sample strain pair; For each sample strain pair, determine the neutralization rate of the sample strain pair, and determine the actual antigenic relationship of the sample strain pair based on the neutralization rate and the determined target evaluation threshold, the neutralization rate is used to characterize the relative size of the neutralization ability of the serum corresponding to one strain in the sample strain pair and the neutralization ability of the serum on the other strain in the sample strain pair, the actual antigenic relationship is used to characterize whether the two corresponding strains in the sample strain pair are similar, and input the virus feature data of each sample strain pair into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model; The loss value is determined based on the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, and the prediction model is trained based on the loss value to predict the antigenic relationship through the trained prediction model, and the loss value is positively correlated with the deviation.
2. The method according to claim 1, characterized in that For each sample strain pair, the virus characteristic data of the sample strain pair is determined according to the strain data corresponding to the sample strain pair, specifically including: For each sample strain pair, the basic physicochemical characteristics and structural correlation characteristics of the sample strain pair are determined based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the basic physicochemical characteristics are used to characterize the charge, polarity, hydrogen bonding index, volume, hydrophobicity, accessible surface area and van der Waals volume of the sample strain pair, and the structural correlation characteristics are used to characterize the structural correlation matrix of the sample strain pair, and the elements in the structural correlation matrix are used to characterize the similarity coefficients based on the folding angle between different amino acid residues on the protein main chain of the sample strain pair; For each sample strain pair, the virus characteristic data of the sample strain pair is determined based on the basic physicochemical characteristics and the structural correlation characteristics.
3. The method according to claim 2, characterized in that Before determining the virus characteristic data of each sample strain pair according to the basic physicochemical characteristics and the structural correlation characteristics, the method further includes: For each sample strain pair, the amino acid coding feature of the sample strain pair is determined according to the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the amino acid coding feature is used to characterize the coding information obtained after encoding the hemagglutinin amino acid sequences of the two corresponding strains in the sample strain pair based on a preset coding rule; For each sample strain pair, the virus characteristic data of the sample strain pair is determined based on the basic physicochemical characteristics and the structural correlation characteristics, specifically including: For each sample strain pair, the virus characteristic data of the sample strain pair is determined based on the basic physicochemical characteristics, the structural correlation characteristics and the amino acid coding characteristics.
4. The method according to claim 1, characterized in that Determine target assessment thresholds, including: For each sample strain pair, the genetic distance of the sample strain pair is determined based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair, wherein the genetic distance is used to characterize the difference between the hemagglutinin amino acid sequences of the two corresponding strains in the sample strain pair; For each sample strain pair, determine the first auxiliary antigen relationship of the sample strain pair according to the genetic distance, and obtain each candidate evaluation threshold, and for each candidate evaluation threshold, determine the control antigen relationship of the sample strain pair assuming that the candidate evaluation threshold is used to determine the antigen relationship according to the candidate evaluation threshold and the neutralization rate of the sample strain pair; For each candidate evaluation threshold, determine a first evaluation score corresponding to the candidate evaluation threshold according to the control antigen relationship of each sample strain pair and the first auxiliary antigen relationship of each sample strain pair; A target evaluation threshold is determined according to the first evaluation scores corresponding to each candidate evaluation threshold.
5. The method according to claim 4, characterized in that For each candidate evaluation threshold, the first evaluation score corresponding to the candidate evaluation threshold is determined according to the control antigen relationship of each sample strain pair and the first auxiliary antigen relationship of each sample strain pair, specifically including: For each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the first auxiliary antigen relationship of each sample strain pair, determine the first accuracy rate and the first balance value corresponding to the candidate evaluation threshold, wherein the first accuracy rate is determined according to the proportion of the first auxiliary antigen relationship in the control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship, and the first balance value is determined according to the number of the first auxiliary antigen relationship in each control antigen relationship of each sample strain pair and the number of the first auxiliary antigen relationship in each control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship; For each candidate evaluation threshold, a first evaluation score corresponding to the candidate evaluation threshold is determined according to a first accuracy rate and a first balance value corresponding to the candidate evaluation threshold.
6. The method according to claim 4, characterized in that Before determining the target evaluation threshold according to the first evaluation scores corresponding to the candidate evaluation thresholds, the method further includes: Deploy preset auxiliary prediction models; For each sample strain pair, extract the sequence features of the sample strain pair based on the hemagglutinin amino acid sequence data of the two corresponding strains in the sample strain pair; For each auxiliary prediction model, the sequence feature is input into the auxiliary prediction model to obtain the second auxiliary antigen relationship of the sample strain pair output by the auxiliary prediction model, and for each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the second auxiliary antigen relationship of each sample strain pair output by the auxiliary prediction model, the second evaluation score of the candidate evaluation threshold for the auxiliary prediction model is determined; Determine the target evaluation threshold according to the first evaluation score corresponding to each candidate evaluation threshold, specifically including: For each candidate evaluation threshold, determine an average evaluation score corresponding to the candidate evaluation threshold according to the first evaluation score corresponding to the candidate evaluation threshold and each second evaluation score corresponding to the candidate evaluation threshold determined based on each auxiliary prediction model; The target evaluation threshold is determined according to the average evaluation scores corresponding to each candidate evaluation threshold.
7. The method according to claim 6, characterized in that For each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the second auxiliary antigen relationship of each sample strain pair output by the auxiliary prediction model, the second evaluation score of the candidate evaluation threshold for the auxiliary prediction model is determined, specifically including: For each candidate evaluation threshold, according to the control antigen relationship of each sample strain pair and the second auxiliary antigen relationship of each sample strain pair output by the auxiliary prediction model, determine the second accuracy rate of the candidate evaluation threshold for the auxiliary prediction model and the second balance value for the auxiliary prediction model, wherein the second accuracy rate is determined according to the proportion of the second auxiliary antigen relationship output by the auxiliary prediction model in the control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship, and the second balance value is determined according to the number of the second auxiliary antigen relationship output by the auxiliary prediction model in each control antigen relationship of each sample strain pair and the number of the second auxiliary antigen relationship output by the auxiliary prediction model in each control antigen relationship of each sample strain pair, assuming that the candidate evaluation threshold is used to determine the antigen relationship; For each candidate evaluation threshold, a second evaluation score of the candidate evaluation threshold for the auxiliary prediction model is determined according to a second accuracy rate of the candidate evaluation threshold for the auxiliary prediction model and a second balance value for the auxiliary prediction model.
8. A data prediction method for influenza virus antigenic relationships, characterized in that: include: Determine the viral characteristic data of the target strain pair; Inputting the virus feature data into a pre-trained prediction model to predict the antigenic relationship of the target strain pair, wherein the prediction model is trained by the model training method in the method according to any one of claims 1 to 7, and the antigenic relationship is used to characterize whether the two corresponding strains in the target strain pair are similar; Business processing is performed according to the antigen relationship.
9. A data prediction device for influenza virus antigen relationship, characterized in that: include: A reading module is used to read the strain data of each sample strain pair. For each sample strain pair, there are two different strains in the sample strain pair. A first determination module is used to determine, for each sample strain pair, virus characteristic data of the sample strain pair according to the strain data corresponding to the sample strain pair; A module is obtained, which is used to determine the neutralization rate of each sample strain pair, and determine the actual antigenic relationship of the sample strain pair according to the neutralization rate and the determined target evaluation threshold, wherein the neutralization rate is used to characterize the relative size of the neutralization ability of the serum corresponding to one strain in the sample strain pair and the neutralization ability of the serum to the other strain in the sample strain pair, and the actual antigenic relationship is used to characterize whether the two corresponding strains in the sample strain pair are similar, and the virus feature data of each sample strain pair is input into the prediction model to be trained to obtain the predicted antigenic relationship for each sample strain pair output by the prediction model; The training module is used to determine the loss value according to the deviation between the predicted antigenic relationship of each sample strain pair and the actual antigenic relationship of each sample strain pair, and train the prediction model according to the loss value to predict the antigenic relationship through the trained prediction model, and the loss value is positively correlated with the deviation.
10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method described in any one of claims 1 to 8 is implemented.
Citation Information
Patent Citations
Method for predicting flu antigen through model and application thereof
CN101847179A
Method and device for rapidly identifying antigenicity of H9N2 subtype avian influenza strain
CN117912568A
Novel method and device for rapidly identifying antigenicity of coronavirus strain
CN118782148A
