Measuring and predicting viral gene mutation patterns

Abstract

The present invention discloses a method for measuring and predicting the mutation pattern of a virus (e.g., influenza virus) by identifying effective mutations and effective mutation periods in the amino acid sequence of the virus. During the effective mutation period, the mutation enables the virus to evade human immunity. Based on the analysis of the existing virus composition and the infection rate of the population, the method can measure the mutation activity of the virus gene ("g-metric") and optimize one or more parameters that represent the activity of the virus gene. The present invention can be used to predict the future gene activity of the virus, mutation, screening of virus vaccine strains and / or prediction of infectious disease outbreaks.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of U.S. Provisional Application No. 62 / 687,645, filed on June 20, 2018, the disclosure of which is incorporated by reference in its entirety.

[0003] background

[0004] The present invention relates generally to the genetic epidemiology of viral infectious diseases (eg, influenza), and more particularly to the measurement and prediction of viral gene (or amino acid) mutation patterns of viruses that cause infectious diseases.

[0005] Influenza, also known as "the flu," is a contagious respiratory illness that has plagued humans for centuries. When it was discovered that influenza was caused by a virus (the influenza virus, or flu virus), the hope was to produce an effective vaccine. After years of research, influenza vaccines are now widely used. However, influenza viruses quickly mutate into new strains, and a vaccine that is effective against one strain may not be effective against other (mutated) strains. Therefore, the "recipe" of influenza virus strains used in preparing influenza vaccines is regularly modified based on predictions of future effective strains, and governments encourage individuals to get a new influenza vaccine every year to help their immune systems keep up with mutated influenza viruses.

[0006] Currently, the annual plan for influenza vaccine production and distribution includes the need to decide which influenza virus strains to protect against in the next round of vaccination. Currently, this decision is based on studies of influenza virus samples from around the world, known antigenic sites (e.g., specific amino acids in the viral sequence), and lessons learned from experience about viral mutation patterns. The goal is to predict which influenza virus strains will be effective against the human immune system (i.e., produce disease) in the next approximately 18 months to 2 years. The influenza vaccine is developed based on this prediction.

[0007] Forecasts are not always accurate, so the effectiveness of the flu vaccine varies widely from year to year. This makes individuals less willing to get the flu vaccine, undermining the "herd immunity" effect that is achieved when most people are immunized against an infectious agent.

[0008] Therefore, it will be particularly important to improve the technology for predicting viral mutations, and particularly for predicting which mutations will be effective against the human immune system over a time frame of at least two years in the future.

[0009] Overview

[0010] Certain embodiments of the present invention relate to techniques for measuring and predicting viral mutation patterns based on viral sequences (e.g., amino acid sequences) and population prevalence levels. The prediction is based on identifying "effective mutations," mutations (variations in amino acid sequences or nucleic acid sequences) with evolutionary advantages that help the virus evade human immunity, as opposed to "unimportant mutations" that have no (or negligible) effect on the ability of the virus to survive and reproduce. The prediction is also based on the assumption that human immunity will eventually learn to recognize and prevent effective mutations (with or without the help of vaccines). This means that effective mutations have an "effective mutation period," which is the time period during which mutations enable the virus to evade human immunity. Using the techniques described herein, identifying effective mutations and determining effective mutation periods can more accurately predict which strains (i.e., which mutations) of a given virus will be prevalent in future time periods. This prediction can achieve a variety of practical purposes, including: (1) helping to select viral strains for vaccine production; (2) providing real-time information about the possible efficacy of a given version of a vaccine; and / or (3) predicting viral activity (e.g., the incidence of infectious diseases caused by the virus).

[0011] Some illustrative techniques used herein rely on longitudinal cohort analysis of influenza virus composition (amino acid sequence) and infection rate to calculate a measure of gene mutation activity of influenza virus, referred to herein as "g-metric". The following will describe in more detail how g-metric simulates gene activity from at least two aspects. The first is whether a single mutation should be considered important. Assuming that more adaptable mutations will spread widely after emergence, and unimportant mutations will not spread, the prevalence of a single residue will lead to a higher g-metric. The second aspect of gene activity is reflected in the number of genes that mutate at the same time, and the g-metric can capture potential antigenic shifts with multiple residue substitutions at the same time; at a given prevalence, a higher number of effective mutations will increase the g-metric. Therefore, the g-metric reflects the adaptability of the mutation and the number of effective mutations at the same time. In addition, if a certain site has more than one effective mutation period during the study period, the g-metric will cover the subsequent effective mutation period. Calculating the g-metric also includes optimizing parameters that further characterize the activity of influenza virus genes, such as the advantage threshold (the minimum prevalence required for a residue to be considered an effective mutation) and the extended effective period (representing the effective mutation after gaining an advantage to remain effective against human immunity). The g-metric and / or related parameters can be used to predict future gene activity of influenza viruses, which can help select virus strains for the next round of influenza vaccines and / or predict influenza outbreaks. Similar techniques can be applied to other viruses and related infectious diseases.

[0012] The following detailed description, together with the accompanying drawings, provides a better understanding of the nature and advantages of the claimed invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figures 1A-1C A simplified example of a coding sequence construction according to an embodiment of the present invention is shown. Figure 1A Four exemplary amino acid sequences observed over time are shown. Figure 1B It is shown that according to embodiments of the present invention, tag sequences within a study period can be defined. Figure 1C Shows the corresponding Figure 1A The amino acid sequence and Figure 1B The coding sequence of the tag sequence.

[0015] Figure 1D It shows that according to the embodiment of the present invention, Figure 1C The prevalence vector calculated for the coding sequence.

[0016] Figure 2 A simplified example of identifying effective mutations and effective mutation periods from a prevalence vector according to an embodiment of the present invention is shown.

[0017] Figure 3 and Figure 4 is a graph showing the correlation of the g-measure with the variation in influenza infection observed in a population. Figure 3 Data obtained from observations of influenza virus activity in Hong Kong, China from 1996 to 2015 are shown. Figure 4 Data obtained from observations of influenza virus activity in New York from 2003 to 2016 are shown.

[0018] Figure 5 A flow chart is shown of a method for measuring and predicting influenza virus activity according to an embodiment of the present invention.

[0019] Details

[0020] The techniques for modeling viral activity described in the present invention rely on longitudinal cohort analysis of viral composition (amino acid sequence) and infection rate to calculate a measure of the genetic mutation activity of the virus, referred to herein as the "g-measure". The analysis is performed in a "study period" divided into a set of time periods of equal duration. In some embodiments, each time period may be one year; other embodiments may define shorter time periods (e.g., three months, one month, one week) or longer time periods (e.g., two years, five years, etc.). For illustrative purposes, influenza or "flu" viruses are mentioned; however, the techniques described can be applied to other viruses.

[0021] For a given time period t, n t For each sample i in time period t, determine the amino acid sequence of the virus Where the index j represents a specific position within the amino acid sequence, and x is an identifier for a specific amino acid. The amino acid sequence of a given sample of influenza virus can be determined using conventional or other techniques, and the specific sequencing technique is not critical to understanding the present invention. Typically, n t Number of amino acid sequences An instance of is identified.

[0022] It is assumed that the virus can mutate during the study period, and different samples of influenza virus collected in the same time period can have different mutations. To facilitate the analysis of mutations, it is helpful to define a "tag sequence" within the study period, which can be used to represent each sample in a unified format. For k = 1, ..., K, the tag sequence can be an amino acid sequence {a k}, where K is defined as:

[0023]

[0024] where J is the total amino acid sequence length of the virus, and q j is the number of unique amino acids observed at position j during the entire study period. k} is composed of all unique amino acids observed at each position j in the amino acid sequence. Tag sequences enable mutations to be evaluated without establishing a reference sequence (which is routinely done); therefore, tag sequences are not a comparison of sequences, but rather provide a tool to capture the dynamics of every possible residue.

[0025] Given a label sequence {a k}, each observed amino acid sequence Can be represented as a coding sequence The coding sequence can be a sequence of K indicators (e.g., a number of bits), with one indicator corresponding to each position k in the tag sequence; if the corresponding amino acid at position j is present in sample i, the indicator at the kth position can be set to a first value (e.g., 1), and if not, it can be set to a second value (e.g., 0).

[0026] Figures 1A-1C The coding sequence according to the embodiment of the present invention is shown A simplified example of the construction of . Figure 1AFour exemplary amino acid sequences 101, 102, 103, 104 observed during a time period t (e.g., one year) are shown; amino acids are represented by single-letter codes using the standard IUPAC single-letter coding scheme. It can be seen that in the observed sequences 101-104, the first position (j=1) has an amino acid N or K; the second position (j=2) has an amino acid S; the third position (j=3) has an amino acid E or K; the fourth position (j=4) has an amino acid N; and the fifth position (j=5) has an amino acid A or T.

[0027] In this example, it is assumed that the amino acid sequence is also observed in other time periods during the study period (e.g., multiple years), and that other amino acids are observed to occur at some positions in at least one of those time periods. Specifically, it is assumed that the following observations are made: for position j=1, amino acid V, I, N, or K is observed; for position j=2, amino acid S is observed; for position j=3, amino acid E or K is observed; for position j=4, amino acid N or D is observed; and for position j=5, amino acid A or T is observed. Figure 1B It is shown that according to an embodiment of the present invention, a tag sequence 120 within a study period can be defined. In this example, the number of bits of the tag sequence 120 is sorted so that the first four tag sequence positions correspond to the amino acid observed at j=1, the next tag sequence position corresponds to the amino acid observed at j=2, and so on. When multiple bits of the tag sequence correspond to the same site in the amino acid sequence, the bits can be sorted based on the time period of the first observation. Other sorting can be used if necessary.

[0028] Figure 1C Coding sequences 131, 132, 133, 134 corresponding to amino acid sequences 101, 102, 103, 104, respectively, are shown. Coding sequences 131-134 provide the same information as the original amino acid sequences 101-104, but in a format that is analyzed computationally as described below. It should be understood that the amino acid sequence of influenza virus is much longer than the amino acid sequence in this simplified example, and the number of sequence samples obtained in a period of time can be much greater than the four examples shown. It should also be understood that Figures 1A-1C The specific sequences in are for illustrative purposes only and may or may not correspond to existing viruses.

[0029] Given a set of n corresponding to samples i observed during time period t t Coding sequence The prevalence vector in time period t Can be defined as:

[0030]

[0031] Prevalence vector p t Each component of can be understood as the prevalence of a specific amino acid at a specific position in the amino acid sequence. Figure 1D It shows that according to equation (2) Figure 1C The prevalence vector p of the coding sequence is calculated t .

[0032] To identify effective mutations, i.e. mutations that provide an evolutionary advantage against human immunity, the prevalence vector p can be analyzed over the entire time period of the study period. t The prevalence of label position k can be obtained from time period t 0 From zero to the following time period t 0 +1, etc. to identify mutations. It is assumed that effective mutations will increase prevalence and eventually reach at least a threshold prevalence, which is referred to herein as the “dominance threshold” and denoted as θ. For ease of analysis, if there is a time t during the study period 0 and time t θ , so that

[0033]

[0034] Then the position a of the label sequence k A mutation at is defined as valid. The value of the advantage threshold θ can be determined empirically as described below.

[0035] It is also useful to define an effective mutation period (EMP, denoted herein by ω), which represents the length of time that an effective mutation retains its evolutionary advantage. This period includes the transition time t θ -t 0 (i.e., from the time a mutation first appears to the time it reaches the dominance threshold.) The EMP also includes an “extended effective mutation period” denoted as h, which corresponds to the length of time a mutation retains its evolutionary advantage after reaching dominance. Thus, for a given mutation at site k, the total EMP is defined as:

[0036] ω k (θ,h)={t 0 <t≤t θ +h|θ,h,k}. (4)

[0037] The set of effective mutations during time period t (referred to as W in this paper) t It can be expressed as:

[0038]

[0039] The optimal values ​​of θ and h can be determined empirically using the fitting procedure described below. In principle, the tag sequence {a kDifferent positions k in the} have their own specific values ​​of θ and h; however, in practice, it is sometimes not feasible to collect enough data to determine the fit of each position, so it can be assumed that all mutations share the same values ​​of θ and h. In a specific example, θ = 0.8 and h = 2.

[0040] Figure 2 A simplified example of using a prevalence vector to identify effective mutations and EMPs according to an embodiment of the present invention is shown. Assuming that Figure 1B The tag sequence {a k}, and assuming Figure 1D The prevalence vector p is the prevalence vector for time period t=1. The figure also shows the prevalence vector p for time periods t=2 to t=7. t ; These vectors can be determined in the manner described above. For ease of illustration, assume θ=0.8 and h=2. For each effective mutation (i.e., a mutation that satisfies the conditions of equation (2)), the prevalence value within the transition time is shown in light gray, the prevalence value within the extended effective mutation period is shown in black, and the total EMP is outlined with a thick black line. It should be noted that although the values ​​of θ and h are assumed to be independent of the site, the total EMP may vary due to differences in transition times. In this analysis, mutations at sites k=6 and k=8 were not identified as effective mutations, even though they did meet the advantage threshold for at least some time periods, because the transition from zero prevalence to non-zero prevalence occurred before t=1.

[0041] After identifying effective mutations and EMPs, a metric that reflects the activity of gene mutations (referred to herein as the “g-metric”) can be calculated. Specifically, for each time period t, the indicator vector m of the K component t is defined as:

[0042]

[0043] where ω(θ,h) is defined according to equation (4). The g-metric can be defined as:

[0044]

[0045] Figure 2 The g calculated according to equation (7) for each time period is shown. t g-metric vector g = [g t ] indicates the trend of mutation activity in different time periods.

[0046] The g-metric can be understood as a function of the prevalence of all effective mutations in a given time period (e.g., the sum). It models two relevant aspects of gene activity. The first is whether a mutation should be considered important. Assuming that more adaptive mutations will spread widely after emergence, while insignificant mutations will not spread, the prevalence of a single residue will lead to a higher g-metric. The second aspect is the number of simultaneous mutations, which captures potential antigenic shifts with multiple residue substitutions at the same time; at a given prevalence, a higher number of effective mutations will increase the g-metric. Therefore, the g-metric reflects the adaptability of the mutation and the number of simultaneous effective mutations. In addition, if a site has more than one effective mutation period during the study period, the g-metric will cover all effective mutation periods. The g-metric can be used for various purposes, including: (1) predictive epidemiology; (2) selecting virus strains for the next round of influenza vaccine based on effective mutations and EMP; (3) evaluating currently available influenza vaccine strains based on comparing current effective mutations with vaccine strains.

[0047] As described above, the g-metric depends on two parameters: the advantage threshold θ and the extended effective mutation period h. In some embodiments, the values ​​of these parameters can be empirically determined based on population-level epidemic variables, such as the seroprevalence of subtypes, the number of cases of viral infection diagnosed over a period of time, or the hospitalization rate of viral infection over that period of time. It is expected that the temporal changes in the g-metric should be correlated with the temporal changes in the population-level epidemic variables, because the spread of new effective mutations will lead to more infections in the population.

[0048] Thus, in some embodiments of the invention, the following fitting procedure can be used to determine the values ​​of θ and h. A population-level prevalence variable (e.g., the number of diagnosed cases or the number of hospitalizations) is defined as a vector f = [f t ], where the index t represents any time period within the study period. A function S(f,g) is chosen that measures the quality of the match between vectors g and f. For example, S can be the p-value of the goodness-of-fit statistic for a generalized linear model, where f is the response variable and g is the predictor variable. In this case, smaller values ​​of S indicate a better match between the response and the prediction. The optimal values ​​of θ and h can be defined as the values ​​that minimize S Right now:

[0049]

[0050] Where H = {0, 1, 2, ..} and θ = [0.5, 1].

[0051] By way of explanation, Figure 3 and Figure 4 A plot showing the correlation of the g-measure with the variation in influenza infection observed in a population. Figure 3Data obtained from observations of influenza virus activity in Hong Kong, China from 1996 to 2015 are shown. The diamond data points connected by dotted lines correspond to the number of influenza A cases diagnosed each year. The circle data points connected by solid lines represent the number of cases predicted using the g-metric calculated as described above. Similarly, Figure 4 Data obtained from observations of influenza virus activity in New York from 2003 to 2016 are shown. The diamond data points connected by dashed lines show the percentage of influenza cases attributed to the H3 strain of the virus in a given year. The circle data points connected by solid lines represent the number of such cases predicted using the g-metric calculated as described above. Figure 3 and Figure 4 As can be seen in , the g-measure with optimal values ​​of θ and h can model the variation in influenza incidence in the population.

[0052] The g-metric as described herein can be used to make predictions about future influenza virus activity. In some embodiments, the future incidence of influenza can be predicted. For example, if the fitting function S(f, g) is the p-value of the goodness-of-fit statistic of the Poisson regression model, the following fitting model can be obtained from the existing data:

[0053]

[0054] where X is an environmental covariate associated with the epidemic (e.g., temperature and humidity), and T is a time variable; the coefficients are determined by fitting to When the sample size is sufficient, more complex fitting functions such as system dynamics models can also be used.

[0055] When the virus sequence samples of time period t+1 are available, we can use p t+1 and Calculate the g-measure. When sequence samples are not available (e.g., when t+1 corresponds to a time period in the future), one can use the conditional prevalence distribution in the existing data as To estimate p t+1 ; The estimate of the prevalence at time period t+1 is:

[0056]

[0057] Where E represents the conditional prevalence distribution Determine the expected value. It can be obtained from p in the above way t+1 Conduct m t+1 and g t+1 The predicted prevalence level is given by:

[0058]

[0059] In some embodiments, a prediction can be made for the next dominant influenza subtype. For example, a g-metric can be obtained for each subtype, and the one with the highest The subtype is the predicted dominant subtype for the next time period. In general, the change in the g-metric, which is a function of the mutation prevalence, can be used to predict the next dominant subtype and future influenza trends.

[0060] In some embodiments, prediction of effective mutations can also be performed. Equation (5) defines the effective mutation W in time period t t A collection of. t Equation (10) starts with W t+1 prediction, and the advantage threshold It can be used to identify mutations that are likely to become dominant within time period t+1. Extended effective mutation period Can be used to identify W t The effective mutation in , which may lose its effectiveness in time period t+1. t+1 The collection of can be used for vaccine antigen design. For example, for vaccines using genetic engineering, W t+1 Amino acids that need to be included in the vaccine can be identified.

[0061] In some embodiments, representative viral sequences for time period t can be defined For example, for each amino acid position j, the amino acid with the highest prevalence at that position can be defined as the representative amino acid. For illustration, please refer to Figure 1B The tag sequence and Figure 1D For site j=1, amino acid K has the highest prevalence (p=0.75); for site j=2, amino acid S has the highest prevalence (p=1); for site j=3, amino acids E and K have the same prevalence (p=0.5), so either can be chosen; for site j=4, amino acid N has the highest prevalence (p=1); and for site j=5, amino acid T has the highest prevalence (p=0.75). More generally, as described above, the tag sequence {a k} includes the number of amino acids corresponding to each position in the amino acid sequence q j In this case, the representative viral sequence Each amino acid will be:

[0062]

[0063] where r0 is the value of exponent r that produces:

[0064]

[0065] Among them, for sequence site j, the range (r L ,r U ) is defined as follows:

[0066]

[0067] r U =r L +q j (14b)

[0068] Representative virus sequences is a probabilistic summary of all effective mutant viruses included at time t. Comparing representative viral sequences with strains included in currently available influenza vaccines allows an assessment of the likely effectiveness of the vaccine. For example, representative viral sequences can be calculated and the distance between strains included in currently available influenza vaccines. To achieve this goal, the distance between sequences can be defined according to conventional sequence similarity metrics, such as the p-distance of amino acids or the Hamming distance. The smaller the distance, the better the match (and the vaccine may be more effective in protecting patients from influenza infection).

[0069] In some embodiments, representative viral sequences for a future time period In the same way, the forward-looking prevalence vector defined in equation (10) can be used to predict. In the case of influenza vaccine prepared with existing wild-type viruses, representative viral sequences can be identified and predicted. The best candidate virus strain for the next round of vaccine is selected by the existing wild-type virus with the closest distance. As described above, the distance can be defined according to conventional sequence similarity metrics, such as the p-distance of amino acids. When the predicted effective mutation of the representative viral sequence is not found in the wild-type strain, genetic engineering techniques can be applied to the wild-type sequence to make it exactly the same as the predicted sequence or as similar as possible.

[0070] The analysis methods described herein can be applied to sequence and epidemiological data for a specific region, global data, or a combination of regional and global data. Predictions of candidate vaccine viruses can be specific to a specific region (e.g., country, continent, or hemisphere) or made for global use.

[0071] The analysis method described herein can be applied to any or all gene segments of influenza virus. Since each gene may have different θ and h parameters, when the sample size is large enough, multiple g-metrics of different genes can be fitted simultaneously (global estimation), or the θ and h parameters of important genes (e.g., hemagglutinin and neuraminidase-the most common mutation segments) can be estimated first, and then the θ and h parameters of the remaining gene segments can be conditionally estimated (local optimization).

[0072] The analysis method described in this article can be applied to any influenza virus subtype, such as H3N2, pandemic H1N1, B / Yamagata, B / Victoria. The same method can also be applied to other known viruses that cause infectious diseases, such as A-EV71 virus (the cause of hand, foot and mouth disease), rhinovirus (the cause of the common cold) or emerging pathogens that can cause epidemics or pandemics.

[0073] Sequencing data used in analyses of the kind described herein can be obtained using any available sequencing technology, including but not limited to first generation sequencing (Sanger), next generation sequencing (Illumina platform), or third generation sequencing (PacBio platform or Nanopore platform).

[0074] The analytical methods described herein can be used in computer-implemented methods for predicting influenza virus activity. Figure 5 A flow chart of a process 500 for measuring and predicting influenza virus activity according to an embodiment of the present invention is shown. Figure 5 The process may be implemented using a conventionally designed computer system. Inputs to the process may include real-world data collected during the study period, including data on the incidence or rate of reported cases of influenza and sequence data of influenza viruses observed during the study period.

[0075] At block 502, a study period is defined. The study period can be as long as desired, such as 10 years, 15 years, 20 years, etc. The study period can be divided into multiple time periods of equal length (e.g., one year, three months, etc.). The selection of the study period and the length of each time period can be based on the availability of data that can be used to determine the prevalence of a particular mutation in influenza virus.

[0076] At box 504, a population-level epidemic variable is obtained for each time period. As described above, this can be a variable representing the number of times or frequency of influenza virus infection in a population. Depending on which data sources are available, population-level epidemic variables can be based on the number of reported influenza diagnosis cases and / or the number of reported influenza hospitalization cases. Such data can be obtained from public health records many years ago. In addition, sampling substitutions from prospective longitudinal groups can also be used, and process 500 can be implemented for any combination of data obtained retrospectively and / or from ongoing sampling.

[0077] At frame 506, obtain the amino acid sequence of the influenza virus sample in each time period. For example, influenza virus samples can be regularly collected and sequenced. Samples can be collected from infected patients, from environmental surfaces or in any other way. The amino acid sequence of influenza virus samples can be measured using conventional techniques. Note that the acquisition and sequencing of influenza viruses have become routine practices in at least some places in the world, allowing the use of previously acquired and currently acquired and recorded data to implement process 500.

[0078] At block 508, the coding sequence of each sample of influenza virus for all time periods is determined. As described above, the coding sequence can be determined by first generating a signature sequence representing each amino acid observed at each sequence position throughout the study period, and the coding sequence of a particular sample can be determined based on which observed amino acid is present in each sequence position for that particular sample.

[0079] At block 510, for each time period, a prevalence vector is determined from the coded sequences associated with the time period. The prevalence vector may be calculated as described above.

[0080] At block 512, one or more effective mutations may be identified based on the prevalence vectors of all time periods within the study period, and for each effective mutation, its effective mutation period may be identified. As described above, the identification of effective mutations may be based on whether the mutation first appears after the first time period and whether the mutation reaches a predominance threshold θ. The effective mutation period may be identified as the time from the first appearance of the mutation to reaching the predominance threshold plus the extended effective mutation period h.

[0081] At box 514, the g-metric is optimized based on the one or more effective mutations identified at box 512 and the population-level prevalence variables obtained at box 504. For example, as described above, a similarity function S(f, g) can be defined so that a smaller S represents a closer match between f (representing a vector of observed population-level prevalence variables) and g. The vector g-metric can be calculated using different combinations of values ​​for θ and h, and for each g(θ, h), the value of S can be determined. By iterating over different combinations of values ​​for θ and h, the value that minimizes S can be determined.

[0082] At box 516, future influenza virus activity (i.e., activity during at least one "future" time period t+1 after the last time period of the study period) is predicted. The prediction can be made based on the g-metric and / or pattern observed in the prevalence vector. The above-mentioned prediction method can be used. For example, equations (10) and (11) can be used to predict future epidemic levels. Future effective mutations can be predicted using the definition of effective mutations at equations (10) and equations (5). Future representative viral sequences can be predicted using equations (10) and (12)-(14b). The vaccine matching score can be calculated based on the distance between the current representative viral sequence (described above) and the viral strains included in the vaccine.

[0083] The predictions made at block 516 can be reported to medical professionals for various uses. Examples include: preparing for an expected increase in influenza viruses (including issuing public health advisories, producing additional drugs for treating influenza patients, etc.); selecting influenza strains (wild-type or genetically engineered sequences) to be included in influenza vaccines; and / or evaluating the likely effectiveness of currently available influenza vaccines.

[0084] Although the present invention has been described with reference to specific embodiments, changes and modifications will be possible to those skilled in the art. All of the processes described above are illustrative and may be modified. Processing operations described as separate blocks may be combined, the order of operations may be modified to the extent logic permits, the processing operations described above may be changed or omitted, and additional processing operations not specifically described may be added. Specific definitions and data formats may be modified as needed.

[0085] The study period can be as long or as short as desired, depending on the availability of data. In some embodiments, viral samples and population-level data can be localized to specific regions (e.g., countries, states or regions, cities), allowing modeling of geographic variations in viral activity.

[0086] Furthermore, while the above embodiments relate specifically to influenza virus, one skilled in the art will appreciate that the same analytical methods can be applied to other viruses associated with other infectious diseases and that the present invention is not limited to any particular virus.

[0087] The data analysis and computing operations described in the present invention can be implemented in conventionally designed computer systems such as desktop computers, laptop computers, tablet computers, mobile devices (e.g., smart phones), etc. Computing clusters and / or cloud-based computing systems can be used to increase computing power. Such systems include one or more processors that execute program code (e.g., a general-purpose microprocessor that can be used as a central processing unit (CPU) and / or a dedicated processor such as a graphics processing unit (GPU), which can provide enhanced parallel processing capabilities); memory and other storage devices that store program code and data; user input devices (e.g., keyboards, pointing devices such as mice or touchpads, microphones); user output devices (e.g., display devices, speakers, printers); combined input / output devices (e.g., touch screen displays); signal input / output ports; network communication interfaces (e.g., wired network interfaces, such as Ethernet interfaces and / or wireless network communication interfaces, such as Wi-Fi); and the like. Computer programs incorporating various features of the present invention can be encoded and stored on various computer-readable storage media; suitable media include disks or tapes, optical storage media such as compact disks (CDs) or DVDs (digital versatile disks), flash memory, and other non-temporary media. (It should be understood that "storage" of data is distinct from propagation of data using a transitory medium such as a carrier wave.) Computer-readable media encoded with the program code may be packaged with a compatible computer system or other electronic device, or the program code may be provided separately from the electronic device (e.g., via Internet download or as a separately packaged computer-readable storage medium). Input data and / or output data may be provided in a secure form, for example using blockchain or other encryption technology.

[0088] Therefore, although the invention has been described with respect to particular embodiments, it should be understood that the invention is intended to cover all modifications and equivalents within the scope of the appended claims.

Claims

1. A computer-implemented method for simulating viral activity, the method comprising: For each of a plurality of time periods within the study period, determining a quantitative measure of genetic activity of the virus, a "g-metric," wherein the g-metric models a combination of the prevalence of effective mutations and the number of effective mutations occurring simultaneously, wherein determining the g-metric comprises: For each time period t during the study period, a number of samples n of the virus are obtained. t Amino acid sequence data Based on the amino acid sequence data, determining a signature sequence formed by connecting all unique amino acids observed at each position of the amino acid sequence; Determining a coding sequence for each sample of the virus based on the amino acid sequence data, wherein the coding sequence is a binary sequence having a component corresponding to a tag sequence and indicating the presence or absence of a specific amino acid at each site of the amino acid sequence; For each time period, determine the prevalence vector p t , the prevalence vector includes multiple components corresponding to multiple sequence sites k Each of the components having a value indicating that the virus sample comprises a portion of a specific amino acid at a specific sequence position in the time period, wherein a prevalence vector for each time period is determined based on the coding sequences of all samples in the time period; Based on the prevalence vectors for all time periods, one or more effective mutations are identified, where an effective mutation is Residues that are zero in at least one time period and equal or exceed the dominance threshold in a subsequent time period; For each effective mutation, identifying an effective mutation period, the effective mutation period including the transition time from the first appearance of the effective mutation to reaching the advantage threshold, and further including an extended effective mutation period corresponding to the expected length of time that the effective mutation maintains evolutionary advantage; and The g-metric for each time period t is calculated based on the effective mutations identified in that time period, where the g-metric for a particular time period is defined as the prevalence component of effective mutations corresponding to the effective mutation period including time period t the sum of; and The g-metric and the prevalence vector components corresponding to one or more individual mutations are used to predict the activity of the virus in future time periods following the study period.

2. The method of claim 1, wherein the g-metric is based on data from a geographic region and the prediction of the activity of the virus is for the geographic region, wherein the geographic region is a specific country, continent, hemisphere or the world.

3. The method of claim 1, wherein The dominance threshold and the extended effective mutation period are determined by optimizing the fit between the g-measure and a population-level epidemic variable, where the population-level epidemic variable refers to the disease caused by infection with the virus during the study period.

4. The method of claim 1, wherein using the g-metric and the prevalence vector components corresponding to one or more individual mutations to predict the activity of the virus in a future time period after the study period comprises: predicting a future prevalence of the one or more individual mutations in a future time period after the study period based on a component of the prevalence vector corresponding to the one or more individual mutations and a conditional distribution of prevalence vectors derived from all time periods within the study period, wherein the conditional distribution relates the prevalence of a particular mutation in one time period to the prevalence of the particular mutation in a next time period; predicting a value of the g-metric for the future time period based on the predicted future prevalence of the one or more individual mutations; as well as A future value of a population-level prevalence variable of infection caused by the virus is predicted based at least in part on the predicted value of the g-metric.

5. The method of claim 1, wherein using the g-metric and the prevalence vector components corresponding to one or more individual mutations to predict the activity of the virus in a future time period after the study period comprises: predicting a future prevalence of the one or more individual mutations in a future time period after the study period based on components of the prevalence vector corresponding to the one or more individual mutations and a conditional distribution of prevalence vectors derived from time periods within the study period, wherein the conditional distribution relates the prevalence of a particular mutation in one time period to the prevalence of the particular mutation in a next time period; as well as At least one of the one or more mutations is predicted to become dominant in the future time period based on the predicted future prevalence of the one or more individual mutations.

6. The method of claim 5, further comprising: Selecting amino acids to be included in the vaccine, wherein the selecting comprises predicting at least one mutation of the one or more mutations to become dominant in the future time period.

7. The method of claim 1, wherein using the g-metric and the prevalence vector components corresponding to one or more individual mutations to predict the activity of the virus in a future time period after the study period comprises: predicting a future prevalence of the one or more individual mutations in a future time period after the study period based on components of the prevalence vector corresponding to the one or more individual mutations and a conditional distribution of prevalence vectors derived from time periods within the study period, the conditional distribution relating the prevalence of a particular mutation in one time period to the prevalence of the particular mutation in a next time period; as well as Representative viral sequences are defined based on the predicted future prevalence of the one or more individual mutations in future time periods following the study period.

8. The method of claim 7, wherein using the g-metric and the prevalence vector components corresponding to one or more individual mutations to predict the activity of the virus in a future time period after the study period further comprises: The viral gene segments of future representative strains are predicted based on the prevalence vector components corresponding to the one or more individual mutations.

9. The method of claim 7, further comprising: An existing viral strain is screened as the viral strain to be included in the vaccine that is closer than any other existing viral strain to a representative viral sequence in a subsequent time period.

10. The method of claim 1, further comprising: Based on the prevalence vector of the current time period, define the representative virus sequence of the current time period; determining a distance measure between the representative viral sequence and one or more viral strains included in the vaccine; as well as A likely efficacy of the vaccine is determined based at least in part on the distance measure.

11. A system comprising: Memory for storing data; and A processor coupled to the memory and configured to implement the method of any one of claims 1-10.

12. A computer-readable storage medium having program code instructions stored thereon, wherein when the program code instructions are executed by a processor of a computer system, the processor is caused to implement the method of any one of claims 1 to 10.

Images