Computer-implemented method, computer program product and system for simulating a cell culture process

Abstract

A computer-executed method for simulating a cell culture process is provided. The method includes: obtaining (S10) a measurable parameter value for at least one operational measurement of the cell culture process, the measurable parameter value being a value of a measurable parameter in a cell culture process model; using the obtained measurable parameter value, and employing Bayesian inference, estimating (S20) a value of an unmeasurable parameter in the model, wherein the model describes the cell culture process using a coupled ordinary differential equation including the measurable and unmeasurable parameters, wherein one or more of the unmeasurable parameters relate to lysed cells during the cell culture process; receiving (S30) one or more new measurable parameter values ​​related to the one or more operational conditions of the cell culture process; and simulating (S40) the cell culture process using: a model of the cell culture process; the estimated value of the unmeasurable parameter in the model; and the received one or more measurable parameter values.

Description

[0001] This application relates to computer-executed methods, computer program products, and systems for simulating and / or controlling cell culture processes. Background Technology

[0002] Cell growth during cell culture can be described using mathematical models. For example, Monod's growth kinetics can be used to describe the growth kinetics of cell culture. Models of the cell culture process can be used to simulate the process. In some cases, simulation results can be used to control the cell culture process to achieve the desired cell growth.

[0003] When modeling different types and / or stages of a cell culture process, separate models or separate sets of model parameters (e.g., model coefficients) can be constructed corresponding to the respective types and / or stages of the cell culture process. For example, a cell culture process under batch / feedback operation may require a different model than a cell culture process under continuous culture exchange operation. Furthermore, different stages in a fed-batch operation, such as the exponential growth phase and the stationary phase, may require different models or at least different sets of model parameters. Invention Overview

[0005] According to one aspect, the problem involves providing reliable models of cell culture processes applicable to different types and / or different stages of cell culture processes, thereby facilitating the control of cell culture processes.

[0006] This problem is solved by the features disclosed in the independent claim. Further exemplary embodiments are defined by the dependent claims.

[0007] According to one aspect, a computer-executed method for simulating a cell culture process is provided. The method includes:

[0008] Obtain a measurable parameter value for at least one operation of a cell culture process, wherein the measurable parameter value is a value of a measurable parameter in a cell culture process model, and one or more measurable parameters relate to one or more operating conditions of the cell culture process.

[0009] Using the obtained measurable parameter values, Bayesian inference is used to predict the values ​​of unmeasurable parameters in the model, wherein the model describes the cell culture process using a coupled ordinary differential equation that includes measurable and unmeasurable parameters, wherein one or more of the unmeasurable parameters relate to cell lysis during the cell culture process;

[0010] Receive one or more new measurable parameter values ​​related to one or more operating conditions of the cell culture process;

[0011] The following method is used to simulate the cell culture process:

[0012] A model of the cell culture process;

[0013] Predicted values ​​of unmeasurable parameters in the model; and

[0014] Receive one or more new measurable parameter values.

[0015] In this disclosure, the term "measurable parameter" can refer to a parameter whose value can be measured and / or quantified relative to the operation of a cell culture process. For example, the value of a measurable parameter can be measured using a suitable sensor during the operation of a cell culture process. Furthermore, for example, the value of a measurable parameter can be quantified based on calculations using values ​​measured using one or more sensors during the operation of a cell culture process.

[0016] Examples of measurable parameters in this disclosure may include, but are not limited to, fresh culture medium feed flow rate, fresh culture medium metabolite concentration, cell effluent flow rate, harvest flow rate, recirculation flow rate, bulk solution volume, viable cell density, viability (e.g., percentage of live cells), visible dead cell concentration, metabolite concentration, temperature, pH, lysed oxygen, etc.

[0017] Furthermore, in this disclosure, the term "unmeasurable parameter" can refer to a parameter whose value cannot be directly measured and / or quantified in a manner similar to that described above for measuring and / or quantifying the value of measurable parameters.

[0018] In the methods according to the foregoing aspects, the one or more unmeasurable parameters may include the concentration of lysed cells during cell culture and / or the concentration of biological materials that have a toxic effect on living cells. In this disclosure, "lysed cells" can be understood as dead cells that have decomposed and become part of a large fluid.

[0019] Other examples of unmeasurable parameters in this disclosure may include, but are not limited to, maximum growth rate, cell death rate in the absence of lysed cells, toxicity of lysed cells or biological material, substrate values ​​below which cell growth is inhibited, sensitivity to quadratic terms deviating from optimal conditions, and values ​​above which growth-inhibiting terms are inhibited.

[0020] In the various embodiments and examples described herein, the “operating conditions” of the cell culture process may include, but are not limited to, fresh culture medium feed flow rate, fresh culture medium composition, cell effluent flow rate, harvest flow rate, temperature, pH, dissolved oxygen, agitation rate, viable cell target (e.g., for perfusion), top gas (e.g., O2 and / or CO2), operating volume, etc. “Operating conditions” can be understood as directly measured or specified. For example, a setpoint can be specified as a target value for a parameter representing the operating conditions.

[0021] Furthermore, in the methods described above, the concentration of lysed cells can be tracked by modeling the breakdown of dead cells.

[0022] Furthermore, in the methods described above, the generation of toxic biomaterials can be modeled as a function of living cell density.

[0023] In some implementations and examples, modeling of dead cell decomposition can involve cell mortality, which is adjusted by the concentration of lysed cells and / or the concentration of toxic biomaterials.

[0024] In some implementations and instances, the coupled ordinary differential equations may include the following equations (4) and (5), which involve modeling the decomposition of dead cells:

[0025] and

[0026]

[0027] Where x l It is the concentration of lysed cells, k. l It is the rate of dead cells that transform into lysed cells, x d It is the concentration of dead cells, F h It is the harvest flow rate during cell culture, F. b V is the cell effluent flow rate during cell culture, and μ is the volume of the bulk solution during cell culture. d It is the cell death rate.

[0028] Cell death rate μ d It can be defined by the following equation (9):

[0029]

[0030] Where k d The mortality rate is in the absence of cell lysis. It is the value of the toxin, which can include x. l The concentration m of metabolites, or the value of biological materials. And k t It is a constant representing the increase in mortality from toxins.

[0031] Furthermore, methods according to any of the foregoing aspects may also include:

[0032] Receive information indicating the desired cell growth during cell culture; and

[0033] Based on the results of the simulated cell culture process, the optimal operating conditions for achieving the desired cell growth are determined.

[0034] In the various embodiments and examples described herein, "information indicating desired cell growth" may include, for example, one or more desired trajectories of one or more measurable parameters and / or one or more non-measurable parameters. More specifically, for example, "information indicating desired cell growth" may include one or more desired trajectories of one or more of the following parameters: total cell concentration, live cell concentration, dead cell concentration, lysed cell concentration, viability, concentration of the i-th metabolite, and value of biological material.

[0035] In some cases, specific optimal operating conditions can be provided and displayed to the user. The user can then operate one or more devices that perform the cell culture process, ensuring the process proceeds under these defined conditions. This display and user interaction provide improved and continuous human-computer interaction while operating the one or more devices.

[0036] Furthermore, the methods described above may also include:

[0037] Generate one or more control signals for controlling one or more devices to perform a cell culture process under determined optimal operating conditions; and

[0038] Output one or more control signals generated.

[0039] In the case of generating and outputting one or more control signals as described above, the method according to the above aspects can also be considered as a method for controlling the cell culture process.

[0040] In some implementations and examples, each of the “one or more control signals” may correspond to one of the determined optimal operating conditions. For example, where the determined optimal operating conditions include fresh culture medium flow rate, cell discharge flow rate, and / or harvest flow rate, the control signal corresponding to the flow rate may include a signal controlling a corresponding valve for achieving the determined optimal flow rate, said valve being included in one or more devices for performing the cell culture process.

[0041] Furthermore, for example, where the determined optimal operating conditions include a fresh culture medium composition, the corresponding control signals may include signals for controlling the preparation and / or application of the fresh culture medium to achieve the determined optimal fresh culture medium composition. In more specific instances, in some cases, the basal culture medium is mixed with one or more other culture media to produce an optimal feed medium (e.g., a fresh culture medium with an optimal composition). In this case, the control signals may include signals for controlling the mixing of the basal and other culture media, thereby including appropriate amounts of each culture medium to be mixed in the fresh culture medium to achieve the determined optimal fresh culture medium composition.

[0042] Furthermore, for example, where the determined optimal operating conditions include temperature, the corresponding control signals may include signals controlling heating and / or cooling devices to achieve the optimal temperature for the determined cell culture process. Furthermore, for example, where the determined optimal operating conditions include stirring rate, the corresponding control signals may include signals for controlling a motor to achieve the determined optimal stirring rate, said motor being included in one or more devices performing the cell culture process. Furthermore, for example, if the determined optimal operating conditions include pH, dissolved oxygen, live cell target (e.g., for perfusion), top gas (e.g., O2 and / or CO2), and / or operating volume, the corresponding control signals may include one or more signals to control corresponding valves and / or actuators included in one or more devices performing the cell culture process to achieve the corresponding determined optimal operating conditions.

[0043] Given that the determined optimal operating conditions include temperature, pH, dissolved oxygen, live cell target (e.g., for perfusion), agitation rate, top gas (e.g., O2 and CO2), and / or operating volume, in another specific instance, the corresponding control signal may include a target value (e.g., a setpoint) to be sent to the (local) controller of one or more devices performing the cell culture process. The (local) controller of the one or more devices performing the cell culture process may then control and / or actuate the corresponding devices (e.g., heating and / or cooling devices, motors, valves, and / or actuators) to achieve the target value (e.g., the setpoint) included in the corresponding control signal.

[0044] According to another aspect, a computer-executed method for identifying cell culture processes is provided. The method includes:

[0045] Obtain a measurable parameter value for at least one operation of a cell culture process, wherein the measurable parameter value is a value of a measurable parameter in a cell culture process model, and one or more measurable parameters relate to one or more operating conditions of the cell culture process.

[0046] Using the obtained values ​​of measurable parameters, Bayesian inference is used to estimate the values ​​of unmeasurable parameters in the model, wherein the model describes the cell culture process using a coupled ordinary differential equation that includes measurable and unmeasurable parameters, wherein one or more unmeasurable parameters relate to cell lysis during the cell culture process;

[0047] The estimated values ​​of unmeasurable parameters are stored in the storage medium.

[0048] According to another aspect, other computer-executed methods for simulating cell culture processes are provided. The methods include:

[0049] Receive values ​​of one or more measurable parameters related to one or more operating conditions of the cell culture process; and

[0050] The following methods are used to simulate the cell culture process:

[0051] A model of the cell culture process;

[0052] Predicted values ​​of unmeasurable parameters in the model; and

[0053] Received values ​​of one or more measurable parameters.

[0054] The cell culture process model described therein utilizes a coupled ordinary differential equation that includes measurable and non-measurable parameters to describe the cell culture process, and one or more measurable parameters relate to the one or more operating conditions of the cell culture process;

[0055] One or more of these unmeasurable parameters relate to cell lysis during cell culture; and

[0056] The estimated value of the unmeasurable parameter is estimated using Bayesian inference, based on the measurable parameter value measured for at least one operation in the cell culture process.

[0057] According to another aspect, a computer program product is provided. The computer program product includes computer-readable instructions that, when loaded and executed on a computer, cause the computer to perform the methods according to any of the foregoing aspects.

[0058] According to another aspect, a system for simulating cell culture processes is provided. The system includes:

[0059] A storage medium for storing a cell culture process model, the model describing the cell culture process using coupled ordinary differential equations including measurable and non-measurable parameters, wherein one or more measurable parameters relate to one or more operating conditions of the cell culture process, and one or more non-measurable parameters relate to cell lysis during the cell culture process; and

[0060] Processors configured to perform the following operations:

[0061] Obtain measurable parameter values ​​for at least one operation of the cell culture process, wherein the measurable parameter values ​​are the values ​​of measurable parameters in the model;

[0062] Using the obtained measurable parameter values, Bayesian inference is used to predict the values ​​of unmeasurable parameters in the model;

[0063] Receive one or more new measurable parameter values ​​related to one or more operating conditions during cell culture;

[0064] The following method is used to simulate the cell culture process:

[0065] A model of the cell culture process;

[0066] Predicted values ​​of unmeasurable parameters in the model; and

[0067] Receive one or more new measurable parameter values.

[0068] In a system according to the above aspects, the one or more unmeasurable parameters may include the concentration of lysed cells during cell culture and / or the concentration of biological materials that have a toxic effect on living cells.

[0069] In a system based on the above aspects, the concentration of lysed cells can be tracked by modeling the breakdown of dead cells.

[0070] In systems based on the above aspects, the generation of toxic biomaterials can be modeled as a function of living cell density.

[0071] In systems based on the above aspects, modeling of dead cell decomposition can involve adjusting cell death rate by the concentration of lysed cells and / or the concentration of toxic biological materials.

[0072] In some implementations and instances, the coupled ordinary differential equations may include the following equations (4) and (5) related to modeling the decomposition of dead cells:

[0073] and

[0074]

[0075] Where x l It is the concentration of lysed cells, kJ. l It is the rate of dead cells that transform into lysed cells, x d It is the concentration of dead cells, F h It is the harvest flow rate during cell culture, F. b V is the cell effluent flow rate during cell culture, V is the volume of the bulk solution during cell culture, and μ d It is the cell death rate.

[0076] Cell death rate μ d It can be defined by the following equation (9):

[0077]

[0078] Where k d The mortality rate is in the absence of cell lysis. It is the value of the toxin, which can include x. l The concentration m of metabolites, or the value of biological materials. And k t It is a constant representing the increase in mortality from toxins.

[0079] In a system according to the above aspects, the processor can be further configured as follows:

[0080] Receive information indicating the desired cell growth during cell culture; and

[0081] Based on the results of the simulated cell culture process, the optimal operating conditions for achieving the desired cell growth are determined.

[0082] In a system according to the above aspects, the processor may be further configured as follows:

[0083] Generate one or more control signals for controlling one or more devices to perform a cell culture process under determined optimal operating conditions; and

[0084] Output one or more control signals generated.

[0085] The subject matter described in this application can be implemented as a method or system, possibly in the form of one or more computer program products. The subject matter described in this application can be implemented in a data signal or a machine-readable medium, wherein the medium is contained in one or more information carriers, such as a CD-ROM, DVD-ROM, semiconductor memory, or hard disk. Such a computer program product can cause a data processing apparatus to perform one or more of the operations described in this application.

[0086] Furthermore, the subject matter described in this application can also be implemented as a system including a processor and memory coupled to the processor. The memory can encode one or more programs to cause the processor to execute one or more methods described in this application. In some instances, the system can be a general-purpose computer system. In other instances, the system can be a special-purpose computer system including embedded systems.

[0087] In some cases, any of the above aspects, as well as any of the various implementations and examples described herein, can provide one or more of the following advantages:

[0088] Provides a single upstream biological process model that describes cell growth curves across multiple scenarios, including:

[0089] - The exponential growth, stagnation, and death phases of batch feeding operations, and

[0090] - Media exchange operations such as infusion and enhancement processes;

[0091] Implement the Bayesian method to identify parameters, that is:

[0092] - A powerful global optimization method for identifying highly correlated dynamic coefficients, and

[0093] - Statistical methods for model validation, estimation of uncertainties in dynamic coefficients, and uncertainty in predictive power.

[0094] The application areas of the various aspects, implementation schemes, and examples described herein may include, but are not limited to, process development, process optimization, process simulation (e.g., digital twins), process monitoring, and advanced control of upstream biological processes. At a higher level, in some cases, applications based on any of the aspects, implementation schemes, and examples described herein can be developed to provide products that automatically identify system-level biological process growth models. Additionally, when system coefficients have a physical interpretation, process understanding can be generated to improve the understanding of growth dynamics.

[0095] Brief description of the attached figures

[0096] Details of one or more embodiments are set forth in the exemplary drawings and description below. Other features will be apparent from the description, drawings, and claims. However, it should be understood that even though embodiments are described separately, individual features of different embodiments may be combined into further embodiments.

[0097] Figure 1 A schematic diagram of an exemplary cell culture process system is shown.

[0098] Figure 2 An exemplary flowchart for simulating the cell culture process is shown.

[0099] Figure 3 An exemplary flowchart is shown for Bayesian model parameters and predictions of unmeasurable states.

[0100] Figure 4 An exemplary hardware configuration of a computer that can be used to implement at least a portion of the system described herein is shown. Invention Details

[0102] The examples will be described in detail below with reference to the accompanying drawings. It should be understood that various modifications can be made to the examples. In particular, one or more elements of an example can be combined and used in other examples to form new examples.

[0103] System Configuration

[0104] Figure 1 A schematic diagram of an exemplary cell culture process system is shown. Figure 1 The exemplary system shown includes an upstream process system 1, a computing device 20, and a data storage device 30.

[0105] The upstream processing system 1 may include a bioreactor 10 and a culture medium exchange assembly 12. Cell culture processes can be carried out in the bioreactor 10. The culture medium exchange assembly 12 can separate the bulk fluid from the cells in the cell-free harvest stream and the cell effluent stream. The culture medium exchange assembly 12 can be used to collect desired biological components from the bioreactor 10 as harvest material using at least one separation method.

[0106] The computing device 20 can be connected to the upstream process system 1 via (a) a wired and / or wireless communication network. The computing device 20 can obtain data regarding the operation of the upstream process system 1. For example, the computing device 20 can receive one or more values ​​measured by one or more sensors disposed in the upstream process system 1. Furthermore, the computing device 20 can provide one or more control signals to control the upstream process system 1. The computing device 20 can be configured to perform methods according to various embodiments and examples described herein. For example, the computing device 20 can be configured to identify a model of a cell culture process performed by the upstream process system 1 and simulate the cell culture process, as described below. Furthermore, for example, the computing device 20 can be configured to control the upstream process system 1 based on the simulation results of the cell culture process. The storage device 30 can store information used by the computing device 20 and / or information generated by the computing device 20.

[0107] Cell culture process model

[0108] The growth kinetics of cell culture processes can be described using Mono growth kinetics, such as those described in equations (1) to (12).

[0109] x t =x v +x d +x l (1)

[0110]

[0111]

[0112]

[0113]

[0114]

[0115]

[0116] μ eff =μ max θ sub θ quad θ inh (8)

[0117]

[0118]

[0119]

[0120]

[0121]

[0122] In the above equations (1)-(13),

[0123] ·x t It is the total cell concentration.

[0124] ·x v It is the concentration of live cells.

[0125] ·x d It is the concentration of dead cells.

[0126] ·x l It is the concentration of lysed cells.

[0127] ·x viab It is vitality (e.g., the percentage of living cells).

[0128] ·μ eff It is the effective cell growth rate.

[0129] ·μ d It is the cell death rate.

[0130] ·F b It is the cell outflow rate (e.g., containing substances identical to those in the container, cells, and bulk fluid).

[0131] •V is the volume of the bulk solution.

[0132] ·k l It is the cell lysis rate (e.g., the ratio of dead cells to lysed cells).

[0133] ·F h It is the harvest flow rate (e.g., the bulk fluid separating cells).

[0134] ·μ max It is the maximum cell growth rate.

[0135] · It is the value of biological materials.

[0136] ·k b It is the generation rate of biological materials.

[0137] · It is the value of the toxin, which can include x. l m or

[0138] ·k t It is a constant representing the increase in mortality from toxins.

[0139] ·s i It is the value of the i-th substance, which can be a state (x, m) or an independent variable (u), where x can be x t x v x d x l and / or x viab m can be the concentration of the metabolite (e.g., m as described below). i and / or m f，i Furthermore, u can involve one or more operational conditions.

[0140] ·θ s，i It is the coefficient of the i-th substance whose growth is inhibited.

[0141] ·q i It is the i-th quadratic parameter, which can be either the state (x, m) or the independent variable (u).

[0142] ·θ i，opt It is q i The value, in which growth was not suppressed,

[0143] ·θ q，i It is described as q i Growth inhibition amount and θ i，opt The deviation coefficient,

[0144] ·I i It is the value of the i-th inhibition parameter, which can be a state (x, m) or an independent variable (u).

[0145] ·k i It is the coefficient that exceeds the subsequent growth inhibition.

[0146] ·m i It is the concentration of the i-th metabolite.

[0147] ·δ i (t) is the specific consumption / secretion rate of the i-th metabolite at the current time.

[0148] ·F f It is the flow rate of the fresh culture medium, and,

[0149] ·m f，i It is F f The concentration of the i-th metabolite.

[0150] Among the parameters used in equations (1)-(13), the live cell concentration x v Dead cell concentration x d Vitality x viab Cell outflow velocity F b Volume of bulk solution V, harvest flow rate F h The value of biological materials The value of the i-th substance s i The i-th quadratic parameter q i The value of the i-th suppression parameter I i The concentration m of the i-th metabolite i The specific consumption / secretion rate δ of the i-th metabolite at the current time. i (t), Fresh culture medium flow rate F f The concentration m of the i-th metabolite in the fresh culture medium stream f，i This can be considered an example of the "measurable parameter" of this disclosure.

[0151] In addition, the concentration of lysed cells x l Effective cell growth rate μ eff Cell death rate μ d Cell lysis rate k l Maximum cell growth rate μ max The generation rate k of biological materials b , toxin value The constant k represents the increase in mortality from toxins. t The coefficient θ of the i-th substance whose growth is inhibited. s，i Among them, the growth of q is uninhibited i The value θ i，opt Describe when q i Deviation from θ i，opt The coefficient θ of the inhibition amount of time growth q，i The coefficient k that inhibits growth after exceeding the limit i This can be considered an example of the “unmeasurable parameter” of this disclosure.

[0152] In some cases, the values ​​of the biological materials listed above are used as one of the exemplary "measurable parameters". It can be treated as one of the "non-measurable parameters". Values ​​of biological materials. Its effects can be observed through its use as a toxin.

[0153] As a description of the effective growth rate μ eff The alternative to equation (8) is that the effective growth rate can be described by data-driven methods, such as partial least squares (PLS) regression models or machine learning (ML) models of the following forms.

[0154]

[0155] In equation (14), It can be a data-driven regression function, such as PLS or ML, and δ(t) is the specific consumption / production rate of the selected metabolite at the current time.

[0156] Typically, product concentration or other quality-type indicators may be included in the kinetic model. These types of measures can be modeled using equation (15).

[0157]

[0158]

[0159] In equation (15), γ i Let be the i-th type of biological material, which can be a target protein or a quality metric function, and the function... It can be PLS, latent structural orthogonal projection (OPLS), ML, or other suitable regression algorithms.

[0160] The above equations can be considered as systematic equations for cell culture processes involving batch feeding or culture medium exchange. For example, they can be derived through... Figure 1 The system shown is used for cell culture.

[0161] Part of the system equations unique to this disclosure may be the modeling of lysed cells in equations (4) and (5), and the treatment of bulk fluid toxicity in equation (9). Lysed cells are dead cells that have decomposed and become part of the bulk fluid. The presence of lysed cells is known, but they are not tracked due to a lack of measurements quantifying their concentration. Including lysed cells as a key driver in the system may be a robust global optimization approach, which can be provided by a Bayesian identification method, where there may be sufficient information to obtain information about the material balance between live cells, dead cells, and lysed cells from typical process measurements, including live cell density (VCD) and cell viability.

[0162] A second unique aspect of this disclosure could be toxicity modeling. A challenge in current state-of-the-art modeling of batch or fed-batch processes during their operation may be the alteration of cell manipulation as the batch progresses. In this disclosure, lysed cells, metabolites, and biomaterials (x) are included. l m and The accumulation of various biological materials, including [specific examples of biological materials], can be modeled and treated as toxins to influence cell death and optionally growth rate. This could allow for the identification of individual models describing behavior throughout an entire batch (e.g., a fed-batch cycle).

[0163] Modern biomanufacturing processes often involve media exchange (e.g., intensified and perfusion treatments). Media exchange may involve replacing a portion of the used bulk medium with fresh medium (e.g., containing toxins as described above). Currently, modeling media exchange processes may require identifying separate model coefficients for each type of operation. Including the cumulative toxic effects of biological material could allow models calibrated on batch / feed batch operations to be transferred to media exchange, and ultimately allow for modeling multiple process modalities using a single model. This is advantageous for developing operational strategies during process development and for evaluating various cell lines based solely on data collected from batch / feed batch operations.

[0164] A challenge in modeling cell culture systems can be generating a single system-level model that relates to multiple modalities. For example, as mentioned above, it is often possible to identify a separate model (or at least model coefficients) for exponential growth relative to quiescent phase, and the transferability of the model between batch feeding and media exchange is often not considered. Modeling of unmeasured lysed cell concentration states (from x in equation (4)) l It may be possible to identify a single model for exponential growth, fixed to media exchange in fed-batch operations, thereby enabling the identification of a single model that is transferable between metabolic phases and fed-batch operations relative to media exchange operations.

[0165] Finding the optimal set of coefficients can be a challenging global optimization problem. Some of the many challenges include:

[0166] High correlation between coefficients;

[0167] Lack of model validation standards; and

[0168] There is a lack of methods to determine the significance of coefficients. For linear orthogonal systems, classical statistical methods for determining the significance of parameters can be applied. On the other hand, there is no equivalent direct method for nonlinear systems such as the Monod model mentioned above.

[0169] Model parameter identification and simulation

[0170] Figure 2 An exemplary flowchart shows the simulation process trajectory for a new set of operating conditions (u). Specifically, Figure 2 An exemplary use of the model is shown, where model parameters (e.g., coefficients) are identified from a set of data and the process trajectory is simulated for a new set of operating conditions (u). The model can be identified from batch / feed batch operations and simulated for a set of media exchange operating conditions.

[0171] It can be done Figure 1 The computing device 20 shown performs Figure 2 The exemplary flowchart shown is shown below.

[0172] refer to Figure 2 Data can be collected regarding at least one operation of the cell culture process (step S10). The collected data may include values ​​of measurable parameters encompassing one or more operating conditions. One or more operating conditions may be represented by one or more independent variables (u) and may include, for example, one or more of the following: fresh culture medium feed flow rate, fresh culture medium composition, cell effluent flow rate, harvest flow rate, temperature, pH, dissolved oxygen, agitation rate, viable cell target (e.g., for perfusion), head gas (e.g., O2 and / or CO2), operating volume, etc. Furthermore, in some instances, the collected data may include measurable state values, such as viable cell concentration values ​​x. v Dead cell concentration x d Vitality x viab The concentration m of the i-th metabolite i And / or the concentration m of the i-th metabolite in the fresh feed stream f,i (i = 1, 2, ..., N, where N may be the number of metabolites). The collected data may also include values ​​for biological materials. Biological materials can be used as toxins.

[0173] Therefore, in step S10, the computing device 20 can obtain information for the purpose of... Figure 1 The measurable parameter value measured for at least one operation of the cell culture process performed at the upstream processing system 1 shown.

[0174] The collected data can be used for system identification to find model parameters (e.g., coefficients) using Bayesian methods (step S20). In other words, the computing device 20 can use Bayesian inference to estimate the values ​​of unmeasurable parameters in the cell culture process model using the measurable parameter values ​​obtained in step S10. Unmeasurable parameters can be considered as coefficients of the model to be estimated. The model can describe the cell culture process using coupled ordinary differential equations including, for example, equations (3)-(5).

[0175] The estimation of the coefficients can be achieved within a Bayesian framework. From the probabilistic model of the system described in this paper, the posterior distributions of these parameters can be computed. Therefore, as a result, values ​​corresponding to the maximum values ​​in the posterior distributions of these coefficients, and confidence regions where the coefficient values ​​fall within a specific probability range, can be obtained.

[0176] Treating X(t) as a vector of values ​​describing the Mono growth kinetics model at time t, the coupled ordinary differential equations (ODEs) can be written as equation (17).

[0177]

[0178] The system may consist of K coupled ODEs (e.g., state vectors), and β may be the parameter vector to be inferred. f may be one or more functions describing the dynamics of the system (e.g., equations (3)-(5)). The initial conditions x0 may be known or unknown, and therefore may need to be estimated.

[0179] Consider a set of noisy batch experiments that produce a set of observations Y at different time points in K states. The probability p(Y|X) can be obtained, where X is a shortcut to X(β,x0). The exact definition of probability may depend on how the noise generated by the experiments is modeled. If the noise is considered to be normally distributed around the most probable value of X, then the standard deviation of the normal distribution around the most probable value may need to be considered. The value of the standard deviation can also be time-dependent. For example, the value of the standard deviation may increase over time because the further the ODE is from the initial conditions, the more likely experimental error is to occur.

[0180] Therefore, by combining the likelihood with the prior distribution p(β) on the parameters, we can obtain the posterior distribution as shown in the following equation (18).

[0181]

[0182] Prior distributions describe known information about the parameters to be estimated. For example, if the parameters are known to be within a certain range, a uniform distribution between the range boundaries can be used as the prior distribution. Depending on the prior knowledge available about the parameters of the dynamic model, other prior distributions can be applied. Z may be a marginal possibility that is often difficult to handle. For this reason, numerical approximations are often used to generate inferences. These approximations may require the use of Monte Carlo Markov chains (MCMCs) to simulate large samples. Advances in sampling methods and continuous improvements in available computational power have made it possible to perform such computations. Parameter estimation can be achieved using Monte Carlo methods such as nested sampling and non-reverse samplers (NUTS), but is not limited to these.

[0183] Figure 3 An exemplary workflow for estimating Bayesian model parameters and unmeasurable states is shown. Figure 3 The exemplary workflow shown can be Figure 2 An example of the detailed process of step S20. Therefore, the computing device 20 can execute... Figure 2 When step S20 is shown, it is executed. Figure 3 The example workflow shown is as follows.

[0184] refer to Figure 3This can provide a mechanical model for Bayesian inference, such as coupled ODEs as defined in equation (17) above (step S200). For the parameters to be estimated (in other words, unknown model parameters, such as unmeasurable parameters in this disclosure), a prior distribution can be determined based on available knowledge about the parameters to be estimated (step S202). The prior distribution can be used for the Monte Carlo sampling described above (step S204) and the Design of Experiments (DoE) (step S206).

[0185] Further reference Figure 3 In step S206, taking DoE into account, data about the measurable state can be collected (step S208). The measurable state can be understood as the measurable parameter in this disclosure. For example, in... Figure 2 The data collected in step S10 can correspond to the data collected in step S10. Figure 3 The data collected in step S208. Using the results of Monte Carlo sampling and the data collected in step S208, the posterior distribution of the parameter to be predicted can be obtained (step S210). From the mechanical model and the posterior distribution of the unknown parameters, the values ​​of the unknown parameters, such as the values ​​of unmeasurable parameters, can be predicted (step S212).

[0186] An example of an unknown parameter representing an unknown state of a system is lysed cell density, or in other words, the concentration of lysed cells. Lysed cells are cells that have died and broken down into the bulk fluid. They are typically invisible and unmeasurable. The concentration of lysed cells x l It can be an indicator of toxicity, growth inhibition, and increased mortality. In fed-batch operations, these lysed cells may accumulate, leading to a shift in metabolic activity. During culture medium exchange, fresh medium (F...) f ) may be introduced into the system, and some cells may be expelled (F b Furthermore, a portion of the used culture medium (e.g., in bioreactors containing metabolic byproducts and bulk fluid containing lysed cells) may be separated from the cells for harvesting (F h Replacing used culture medium with fresh medium can reduce the accumulation of lysed cells and related biological products, thereby restoring metabolic activity to a state similar to that in fresh medium.

[0187] Given the available prior knowledge about the parameters and the available data, the Bayesian parameter estimation described above can provide the posterior distribution of the parameters. Therefore, the contribution of the data to making the posterior distribution more informative may indeed depend on the new information brought by the data. Thus, appropriate experimental designs may be needed to create system variations that minimize the correlation between parameter estimates and thus maximize the information. Thorough studies of simulations of dynamic systems and possible experimental settings may be necessary to create a set of experiments that maximize the information provided by the data.

[0188] Refer again Figure 2 After estimating the values ​​of the unmeasurable parameters of the model in step S20, new operating conditions can be provided (step S30), and at least one process trajectory for simulating the state of the cell culture process and the biological materials can be generated (step S40). In other words, after the estimation in step S20, the computing device 20 can receive one or more new measurable parameter values ​​related to the operating conditions of the cell culture process in step S30. Furthermore, the computing device 20 can use the following in step S40: the model of the cell culture process; the estimated values ​​of the unmeasurable parameters in the model; and the received one or more new measurable parameter values ​​to simulate the cell culture process.

[0189] Examples of new operating conditions provided at step S30 may include, but are not limited to, fresh culture medium feed flow rate, fresh culture medium composition, cell effluent flow rate, harvest flow rate, temperature, pH, dissolved oxygen, stirring rate, live cell target (e.g., for perfusion), top gas (e.g., O2 and / or CO2), operating volume, etc.

[0190] In step S40, one or more simulated process trajectories can be generated for the state and biological materials, which may include, but are not limited to, the total cell concentration x. t , live cell concentration x v Dead cell concentration x d Concentration of lysed cells x l Vitality x viab The concentration m of the i-th metabolite i (i = 1, 2, ..., N, where N can be the number of metabolites) and / or the value of biological material.

[0191] Once one or more simulated process trajectories of the state and biological material are generated at step S40, these trajectories may be provided along with independent variables representing new operating conditions (step S50). For example, computing device 20 may provide a display showing at least one image in which one or more simulated process trajectories are plotted or drawn, along with information indicating the values ​​of the new operating conditions received at step S30. In some instances, the plotted trajectories and the information indicating the values ​​of the new operating conditions may be displayed on a display device (not shown) of computing device 20. Optionally or additionally, the plotted trajectories and the information indicating the values ​​of the new operating conditions may be transmitted from computing device 20 to another device besides computing device 20 and displayed on a display of that device.

[0192] In some instances, as referenced above... Figure 2 and 3Following the described simulated cell culture process, the simulation results can be used to determine the optimal operating conditions for achieving the desired cell growth during the cell culture process. For example, computing device 20 can receive information indicating the desired cell growth during the cell culture process and determine the optimal operating conditions for achieving the desired cell growth based on the results of the simulated cell culture process as described above. The information indicating the desired cell growth can be provided by a user using computing device 20 or an input device connected to computing device 20. The determination of the optimal operating conditions for achieving the desired cell growth can, for example, be performed by identifying the simulated process trajectory that best approximates the desired cell growth from the simulation results of the cell culture process and determining the operating conditions that generated one or more process trajectories in the simulation.

[0193] In some cases, the determined optimal operating conditions can be displayed to the user on the display of the computing device 20 and / or another device connected to the computing device 20. The user can then operate the upstream process system 1 to perform the cell culture process under the determined operating conditions. In this way, displaying the determined optimal operating conditions can provide improved and continuous human-computer interaction when operating the upstream process system 1.

[0194] Furthermore, in some instances, the determined optimal operating conditions can be used to control the upstream process system 1. In these instances, the computing device can generate one or more control signals to control the upstream process system 1 to operate under the determined optimal operating conditions, and output the generated one or more control signals to the upstream processing system 1. The upstream processing system 1 can then operate under the determined optimal operating conditions to achieve the desired cell growth during cell culture. Therefore, in these instances, the computing device 20 can implement methods for controlling the cell culture process.

[0195] In some cases, as referenced above Figure 2 and 3 The described model parameter identification and simulation can be enabled:

[0196] • Generate a single system-level model describing growth behavior.

[0197] - In multiple stages of batch replenishment operations, and

[0198] - Between batch feeding and media exchange operations; and

[0199] • Bayesian methods for parameter identification are provided.

[0200] - Powerful global optimization solution

[0201] - Uncertainty in parameter prediction makes it possible to statistically prove the accuracy of parameter prediction, and

[0202] - The uncertainty of the predicted model.

[0203] Furthermore, in some instances, as referenced above... Figure 2 and 3 The described model parameter identification and simulation can be integrated into simulation applications. Simulation applications can be used for one or more of the following:

[0204] Process simulation

[0205] - It can simulate growth curves for various feed or process variables (e.g., pH, temperature, dissolved oxygen).

[0206] - Enables a simulated growth curve for media exchange operations on models calibrated on streaming batch processing, which can be understood as similar to a scaling tool.

[0207] Process optimization

[0208] - By using models in optimization routines, optimal feed and process trajectories can be determined to maximize growth or follow a desired growth trajectory.

[0209] Process control

[0210] - Growth models can be used as part of a model predictive control strategy to adjust feed and process conditions to track the desired growth curve.

[0211] Hardware configuration

[0212] Figure 4 Exemplary hardware configurations of computers that can be used to implement at least a portion of the systems described above are shown. For example, Figure 1 The computing device 20 shown can be used Figure 4 The computer shown in Figure 7 is used to implement this. Figure 4The computer 7 shown includes a central processing unit (CPU) 70, a graphics processing unit (GPU) 88, system memory 72, a network interface 74, a hard disk drive (HDD) interface 76, an external disk drive interface 78, and an input / output (I / O) interface 80. These components of the computer are coupled to each other via a system bus 82. The CPU 70 can perform arithmetic, logical, and / or control operations by accessing the system memory 72. The system memory 72 can store information and / or instructions for use in conjunction with the CPU 70. The system memory 72 may include volatile and non-volatile memory, such as random access memory (RAM) 720 and read-only memory (ROM) 722. A basic input / output system (BIOS) containing basic routines may be stored in the ROM 722, which facilitates the transfer of information between components within the computer 7, such as during startup. The system bus 82 can be any of several types of bus architectures, including a memory bus or memory controller, a peripheral bus, and a local bus using any of a variety of bus architectures. The CPU 70 can also be connected to one or more sensors (not shown) via one or more corresponding interfaces (not shown) and bus 82. The sensors can measure physical conditions or states, including but not limited to: temperature, pH, pressure, etc. Additionally, the sensors can include other types of measurement or detection devices, including but not limited to imaging devices, microphones, spectral sensors, etc. The controller can control physical conditions or states, including but not limited to: temperature, flux, agitation, etc.

[0213] The computer may include a network interface 74 for communicating with other computers and / or devices via a network.

[0214] In addition, the computer may include a hard disk drive (HDD) 84 for reading and writing to a hard disk (not shown), and an external disk drive 86 for reading or writing to a removable disk (not shown). The removable disk may be a disk for the disk drive or an optical disk such as a CD-ROM for an optical disk drive. The HDD 84 and the external disk drive 86 are connected to the system bus 82 via an HDD interface 76 and an external disk drive interface 78, respectively. The drives and their associated computer-readable media provide a non-volatile storage of computer-readable instructions, data structures, program modules, and other data for a general-purpose computer. Data structures may include relevant data for implementing the exemplary methods and variations thereof as described herein. The relevant data may be organized in a database, such as a relational database or an object database.

[0215] Although the exemplary environment described herein employs hard disks (not shown) and external disks (not shown), those skilled in the art will understand that other types of computer-readable media, such as magnetic tape cassettes, flash memory cards, digital video disks, random access memory, read-only memory, etc., which can store computer-accessible data, may also be used in the exemplary operating environment.

[0216] Multiple program modules may be stored on a hard disk, external disk, ROM 722, or RAM 720, including an operating system (not shown), one or more application programs 7202, other program modules (not shown), and program data 7204. The application program may include at least a portion of the functions described above.

[0217] Computer 7 can be connected to an input device 92, such as a mouse and / or keyboard, and a display device 94, such as a liquid crystal display, via corresponding I / O interfaces 80a and 80b and a system bus 82. If computer 7 is implemented as a tablet computer, for example, a touch panel for displaying information and receiving input can be connected to computer 7 via corresponding I / O interfaces and a system bus 82. Furthermore, in some instances, although... Figure 4 As not shown, the computer 7 can also be connected to a printer and / or an imaging device such as a camera via a corresponding I / O interface and system bus 82.

[0218] As used as Figure 4 As a supplement or alternative to the implementation of computer 7 shown herein, some or all of the functions of the exemplary embodiments described herein can be implemented as one or more hardware circuits. Examples of such hardware circuits may include, but are not limited to, large-scale integration (LSI), reduced instruction set computing (RISC), application-specific integrated circuits (ASIC), and field-programmable gate arrays (FPGA).

Claims

1. A computer-executed method for simulating a cell culture process, comprising: S30: Receive one or more measurable parameter values ​​related to one or more operating conditions of the cell culture process, the one or more operating conditions including one or more of the following: fresh culture medium feed flow rate, fresh culture medium composition, cell effluent flow rate, harvest flow rate, temperature, pH, dissolved oxygen, stirring rate, live cell target, top gas, and operating volume. S40: Use the following to simulate the cell culture process: The model of the cell culture process; The estimated values ​​of the unmeasurable parameters in the model; and The received values ​​of one or more measurable parameters; Receive information indicating the desired cell growth during cell culture; as well as Based on the results of the simulated cell culture process, the optimal operating conditions for achieving the desired cell growth were determined. The cell culture process model described therein utilizes a coupled ordinary differential equation comprising measurable parameters and non-measurable parameters to describe the cell culture process, wherein one or more of the measurable parameters relate to the one or more operating conditions of the cell culture process. One or more of the aforementioned unmeasurable parameters are related to cell lysis during the cell culture process; One or more of the unmeasurable parameters mentioned above include one or more of the following: The concentration of lysed cells during cell culture. The concentration of biological materials that have toxic effects on living cells. This involves tracking the concentration of lysed cells by modeling the breakdown of dead cells. The generation of toxic biomaterials was modeled as a function of living cell density. The aforementioned modeling of the decomposition of dead cells involves adjusting the cell death rate by the concentration of lysed cells or the concentration of toxic biomaterials, and The estimated value of the unmeasurable parameter is estimated using Bayesian inference, based on measurable parameter values ​​measured for at least one operation of the cell culture process.

2. The computer-executed method as described in claim 1, further comprising: S10: Obtain the value of a measurable parameter for at least one operation of the cell culture process; S20: Using the obtained measurable parameter values, Bayesian inference is used to estimate the values ​​of the unmeasurable parameters in the model, and the values ​​of the unmeasurable parameters are used as the estimated values ​​of the unmeasurable parameters during the simulated cell culture process.

3. The method of claim 1 or 2, further comprising: Generate one or more control signals for controlling one or more devices to perform the cell culture process to operate under determined optimal operating conditions; as well as Output one or more control signals generated.

4. A computer program product comprising computer-readable instructions, which, when loaded and run on a computer, cause the computer to perform the method of any one of claims 1-3.

5. A system for simulating cell culture processes, comprising: Storage medium (30) storing a model of the cell culture process, the model describing the cell culture process using coupled ordinary differential equations including measurable and non-measurable parameters, wherein one or more of the measurable parameters relate to one or more operating conditions of the cell culture process, and one or more of the non-measurable parameters relate to cell lysis during the cell culture process, the one or more operating conditions including one or more of the following: fresh culture medium feed flow rate, fresh culture medium composition, cell effluent flow rate, harvest flow rate, temperature, pH, dissolved oxygen, agitation rate, live cell target, top gas, operating volume; and The processor is configured to perform the following operations: S30: Receive one or more measurable parameter values ​​related to one or more operating conditions of the cell culture process; S40: Use the following to simulate the cell culture process: The model of the cell culture process; The estimated value of the unmeasurable parameter in the model is estimated using Bayesian inference, which utilizes the values ​​of measurable parameters measured for at least one operation of the cell culture process. and One or more new measurable parameter values ​​received; Receive information indicating the desired cell growth during cell culture; as well as Based on the results of the simulated cell culture process, the optimal operating conditions for achieving the desired cell growth were determined. One or more of the unmeasurable parameters mentioned above include one or more of the following: The concentration of lysed cells during cell culture. The concentration of biological materials that have toxic effects on living cells. This involves tracking the concentration of lysed cells by modeling the breakdown of dead cells. The generation of toxic biomaterials was modeled as a function of living cell density. The aforementioned modeling of the breakdown of dead cells involves adjusting the cell death rate by the concentration of lysed cells or the concentration of toxic biomaterials.

6. The system of claim 5, wherein the processor is further configured to: S10: Obtain the value of a measurable parameter for at least one operation of the cell culture process; S20: Using the obtained measurable parameter values, Bayesian inference is used to estimate the values ​​of the unmeasurable parameters in the model, and the values ​​of the unmeasurable parameters are used as the estimated values ​​of the unmeasurable parameters during the simulated cell culture process.

7. The system of claim 5 or 6, wherein the processor is further configured to: Generate one or more control signals for controlling one or more devices to perform the cell culture process under determined optimal operating conditions; and Output one or more control signals generated.

Images