Method, device, and program for searching medium composition and estimating cell characteristics
The prediction, acquisition, estimation and evaluation process is performed by the processor, and the medium composition is optimized, which solves the problem of medium optimization of cells with unknown characteristics and achieves efficient optimization of cell culture results.
Patent Information
- Application Number
- CN202380032264.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-03-30
- Filing Date
- 2023-03-24
- Publication Date
- 2024-11-22
AI Technical Summary
During the optimization of medium for specific cells, it is difficult to effectively explore the optimal medium composition, especially when cell characteristics are unknown, and prior art is difficult to achieve optimization within a limited development period and experiment number.
A series of processes are performed by the processor, including predicting, obtaining, estimating, evaluating and exploring the determination of the medium composition, repeated until the end benchmark is met, to optimize the medium composition to improve the proliferativeness of the cells or the amount of material production. This method uses prediction and estimation of cell characteristics, adjusts the medium composition, and gradually optimizes the culture results.
The efficient optimization of culture results for cells with unknown characteristics is achieved, reducing cost and time overhead, and improving the exploration efficiency of culture medium composition.
Smart Images

Figure CN119013414A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method, device and program for exploring culture medium composition and estimating cell characteristics, and in particular to a method, device and program for exploring culture medium composition particularly suitable for a specific cell line. Background Art
[0002] In recent years, due to the popularization of gene editing technology, for example, the diversification of various cells such as CHO cells (Chinese Hamster Ovary cells) that produce antibody medicines, HEK cells (Human Embryonic Kidney cells) that produce viruses for gene therapy, and iPS cells (Induced Pluripotent Stem cells) for regenerative medicine has been promoted. In general, there are many cases of optimizing cell culture conditions in order to improve cell proliferation or the production of various products, and there are also many cases of wanting to optimize culture media for specific cells. In this case, as a general solution, the following method is used: based on some known culture medium compositions, the content of one or more compositions is changed little by little, and cells are actually cultured using a large amount of culture medium to maximize target productivity such as proliferation.
[0003] For example, Patent Document 1 describes a technique for constructing a functional environment map that indicates the intensity of activation or inhibition of basic cell functions and developing an optimized cell culture medium formulation using the functional map.
[0004] Previous technical literature
[0005] Patent Literature
[0006] Patent Document 1: Japanese Patent Application No. 2014-503220 Summary of the invention
[0007] Technical issues to be solved by the invention
[0008] When optimizing culture media for specific cells (intrinsic cells), it is believed that the following are effective: (1) understanding the characteristics of unknown cells and (2) specifying the composition of the culture media based on the characteristics. However, sometimes it is desirable to explore the best culture media for cells whose characteristics are not necessarily known. This is because understanding the characteristics of cells itself requires costs in terms of money or time, and due to industrial constraints, the analysis used to understand the characteristics of cells is sometimes difficult. Therefore, for cells with unknown characteristics, it is necessary to explore and propose the best culture media composition within the scope of limited information.
[0009] Generally speaking, in optimization problems such as (2) above, a specific optimal medium composition is gradually changed and optimized. In contrast, in (1) above, in order to clarify cell characteristics, the response to various medium compositions is generally studied. In the case of searching for the optimal medium composition for cells with unknown characteristics, there is a problem that the time and cost equivalent to the general solution described above may be required.
[0010] Thus, in the conventional technology, it is difficult to achieve both (1) and (2) within a limited development period and number of experiments.
[0011] The present invention has been made in view of such circumstances, and an object of the present invention is to provide a method, an apparatus, and a program for exploring medium composition and estimating cell characteristics, which can efficiently optimize culture results even for cells with unknown characteristics.
[0012] Means for solving technical problems
[0013] In order to achieve the above-mentioned purpose, the method involved in the first embodiment of the present invention performs an exploration of the culture medium composition for cells and an estimation of cell characteristics, wherein the processor executes the following steps: a prediction step, for the target cells, using the cell characteristics or estimated values of the target cells, predicting the culture results for the assigned culture medium composition; an acquisition step, obtaining the culture results of the target cells for a part or all of the assigned culture medium composition; an estimation step, estimating the cell characteristics of the target cells using the culture medium composition, and the obtained culture results or the predicted culture results; an evaluation step, evaluating the necessity of further estimating the cell characteristics of the target cells based on the obtained culture results; an exploration and determination step, exploring and determining the culture medium composition, i.e., the candidate culture medium composition, for the culture result to be obtained in the next acquisition step based on the necessity and the prediction step; and a control step, repeating the prediction step, the acquisition step, the estimation step, the evaluation step, and the exploration and determination step until the determined end criterion is met.
[0014] According to the first method, the cell characteristics or estimated values of the target cells are used to predict the culture results for the assigned culture medium composition, and the culture results for part or all of the target cells for the assigned culture medium composition are obtained. The cell characteristics are estimated using the culture medium composition, the obtained culture results or the predicted culture results, and then, based on the obtained culture results, the necessity of further estimating the cell characteristics is evaluated. Based on the necessity and the prediction of the culture results, the candidate culture medium composition for which the culture results should be obtained next is explored and determined, and the above steps are repeated until the determined end criterion is met. In this way, by repeating the prediction and acquisition of culture results, the estimation of cell characteristics, the evaluation of the necessity of additional estimation of cell characteristics, and the exploration and determination of candidate culture medium compositions, the culture results of cells with unknown characteristics can also be effectively optimized.
[0015] In the first embodiment and the following embodiments, the "given medium composition" refers to a "specific medium composition" or an "arbitrary medium composition", and the amount of each component is clearly defined. The "given medium composition" is not limited to a "known" medium that is generally popular or has already been produced, but may also be a "new" medium that is intended to be produced in the future, i.e., has not yet been produced. However, an "unknown" medium with an unknown amount of components is not included in the "given medium composition".
[0016] Furthermore, in the acquisition step, for example, actual culture results may be obtained in a culture evaluation system (experimental system), or culture results based on simulation or the like may be acquired without actually performing culture.
[0017] The method according to the second aspect is based on the first aspect, wherein the processor performs the evaluation in the evaluation step on the assumption that the change of the culture result relative to the change of the medium composition is continuous. In the second aspect, the evaluation is performed on the assumption that "the culture result of the cell will not change significantly relative to a small change in the composition of the medium".
[0018] The method according to the third aspect is based on the first or second aspect, wherein the processor performs the estimation in the estimation step on the assumption that the cell characteristics of the target cell are similar to the cell characteristics of one or more cells in the known cell population. The third aspect estimates the cell characteristics of the known cell population as "prior knowledge".
[0019] The method involved in the fourth aspect is based on any one of the first to third aspects, wherein the processor regards the culture result as the observation history, regards the predicted unit in the prediction process as the unknown function, regards the shape of the unknown function as the cell characteristic, and performs the prediction process, the estimation process, the acquisition process, the evaluation process, and the exploration and determination process by solving the source code problem for estimating the unknown function that best fits the observation history. The source code problem is a problem of optimizing the unknown function with a limited number of trials based on the observation history, and the fourth aspect allocates specific components to the source code problem.
[0020] The method according to the fifth aspect is based on the fourth aspect, wherein the processor expresses the predicted unit, the unknown function and / or the similar function relative to the unknown function in the prediction process through cell simulation including a metabolic pathway model.
[0021] The method involved in the sixth embodiment is based on the fifth embodiment, wherein the processor inputs the culture data of the target cell, determines the parameters of the metabolic pathway model based on the culture data, and estimates the cell characteristics based on the determined parameters. The sixth embodiment specifically stipulates the method for estimating the cell characteristics. In addition, the parameters of the metabolic pathway model constitute a part of the cell characteristics in the present invention.
[0022] The method involved in the seventh embodiment is based on the sixth embodiment, wherein the culture data includes one or more of the total number of cells of the target cells, the amount of cell secretory substances, the amount of cell-produced substances, the amount of cell metabolites, and the amount of culture medium components, and / or one or more of the total number of cells of the target cells, the amount of cell secretory substances, the amount of cell-produced substances, the amount of cell metabolites, and the amount of culture medium components, and the processor separates the parameters into input factors to the target cells and output factors from the target cells in the estimation process, and estimates the cell characteristics in a manner that the input factors and output factors at any time are established. The seventh embodiment further specifically stipulates the estimation of the cell characteristics in the sixth embodiment. In addition, in the seventh embodiment, "in a manner that the input factors and output factors are established" means that the output factors as a result of calculating the input factors based on the cell characteristics constitute the cell characteristics in a manner close to the actual output factors.
[0023] The method according to the eighth aspect is based on any one of the first to seventh aspects, wherein the processor predicts the biological behavior quantity of the target cell based on the medium composition and culture conditions given in the prediction step, changes the cell environment of the target cell based on the prediction result of the biological behavior quantity, and repeats the prediction of the biological behavior quantity and the change of the cell environment based on the changed cell environment, thereby predicting the culture result. The eighth aspect specifically defines a method for predicting the culture result.
[0024] The method according to the ninth aspect is based on the eighth aspect, wherein the processor sets the time change of the total number of cells, the time change of the substance produced by the target cells and the amount of the substance, and the time change of the substance contained in the culture environment of the target cells as the biological behavior quantity, and performs the prediction, and sets the total number of cells of the target cells and the substance produced by the target cells and the amount of the substance as the cell environment, and performs the change. The ninth aspect further specifically specifies the prediction of the eighth aspect.
[0025] The method according to the tenth aspect is any one of the first to ninth aspects, wherein the processor, in the evaluation step, evaluates the degree of deviation between the estimated value and the true value of the cell characteristic based on the predicted culture result and the obtained culture result, and in the search and determination step, searches and determines the medium composition candidate based on the degree of deviation. The tenth aspect specifically defines a method for evaluating the necessity of further characteristic estimation and searching and determining the medium composition candidate.
[0026] The method according to the eleventh aspect is based on any one of the first to ninth aspects, wherein the processor inputs a pair of a culture medium composition and a culture result in the culture medium composition in the evaluation step, evaluates the sufficiency of the search, and searches and determines a candidate culture medium composition based on the evaluation result of the sufficiency in the search and determination step. The eleventh aspect specifically defines another method for evaluating the necessity of further characteristic estimation and searching and determining a candidate culture medium composition.
[0027] The method according to the twelfth aspect is based on any one of the first to eleventh aspects, wherein the processor determines that the end criterion is satisfied and ends the repetition when the difference between the predicted culture result and the obtained culture result and / or the number of repetitions satisfies a certain condition in the control step. The twelfth aspect specifically specifies an aspect of the end criterion of the processing.
[0028] The method according to the thirteenth aspect is the method according to any one of the first to twelfth aspects, wherein the target cell is any one of a CHO cell, a HEK cell, and an iPS cell. The thirteenth aspect specifically specifies an example of the target cell.
[0029] The method according to the fourteenth embodiment is based on any one of the first to thirteenth embodiments, wherein the culture result is one or more of the antibody production amount or virus production amount of the target cell, the proliferation of the target cell, and the differentiation induction success rate of the target cell into a specific tissue. The fourteenth embodiment provides an example of the culture result. In addition, in the present invention, it is preferred to use the culture result corresponding to the type of the target cell.
[0030] In order to achieve the above-mentioned purpose, the device involved in the fifteenth mode of the present invention explores the culture medium composition for cells and estimates the cell characteristics, wherein it has a processor, which uses the cell characteristics or estimated values of the object cells to predict the culture results for the assigned culture medium composition for the object cells, obtains the culture results of the object cells for a part or all of the assigned culture medium composition, estimates the cell characteristics of the object cells using the culture medium composition and the obtained culture results or the predicted culture results, evaluates the necessity of further estimating the cell characteristics of the object cells based on the obtained culture results, explores and determines the culture medium composition for which the culture results should be obtained next, i.e., the candidate culture medium composition, based on the necessity and the prediction, and repeats the prediction, acquisition, estimation, evaluation, exploration and determination until the determined end criterion is met.
[0031] According to the fifteenth aspect, similarly to the first aspect, the culture result can be effectively optimized even for cells with unknown characteristics. In addition, the device according to the fifteenth aspect may have the same structure as the second to fourteenth aspects.
[0032] In order to achieve the above-mentioned purpose, the program involved in the sixteenth mode of the present invention enables the processor to perform an exploration of the culture medium composition of the cell and an estimation of the cell characteristics, wherein it enables the processor to execute the following steps: a prediction step, for the target cell, using the cell characteristics or estimated values of the target cell, predicting the culture result for the assigned culture medium composition; an acquisition step, obtaining the culture result of the target cell for a part or all of the assigned culture medium composition; an estimation step, estimating the cell characteristics of the target cell using the culture medium composition, and the obtained culture results or the predicted culture results; an evaluation step, evaluating the necessity of further estimating the cell characteristics of the target cell based on the obtained culture results; an exploration and determination step, based on the necessity and the prediction step, exploring and determining the culture medium composition, i.e., the candidate culture medium composition, for the culture result to be obtained in the next acquisition step; and a control step, repeating the prediction step, the acquisition step, the estimation step, the evaluation step, and the exploration and determination step until the determined end criterion is met.
[0033] According to the sixteenth embodiment, as with the first and fifteenth embodiments, the culture results can be effectively optimized for cells with unknown characteristics. In addition, the program involved in the sixteenth embodiment can also have the same structure as the second to fourteenth embodiments. In addition, a non-temporary tangible recording medium (for example, various optical and magnetic recording devices, semiconductor memories) with computer-readable codes recording the programs of these embodiments can also be cited as a method of the present invention. In addition, the above-mentioned "non-temporary tangible recording medium" does not include non-tangible recording media such as carrier signals and propagation signals themselves.
[0034] Effects of the Invention
[0035] As described above, according to the method, device, and program of the present invention for exploring the composition of a culture medium and estimating cell characteristics, the culture results can be effectively optimized even for cells whose characteristics are unknown. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 This diagram conceptually shows the workflow of the entire system.
[0037] Figure 2 It is a block diagram showing the structure of the discovery and estimation system according to the first embodiment.
[0038] Figure 3 It is a block diagram showing the structure of the processing unit.
[0039] Figure 4 This is a diagram conceptually showing the structure of the exploration estimation system and the training evaluation system.
[0040] Figure 5 This is a flowchart showing the process of searching for medium composition and estimating cell characteristics according to the first embodiment.
[0041] Figure 6 This is a conceptual diagram for explaining the necessity evaluation of additional estimation of culture medium composition.
[0042] Figure 7 It is a diagram showing the genealogical tree of CHO cells.
[0043] Figure 8 This is a diagram showing how prior knowledge is used in exploration.
[0044] Fig. 9 This is a diagram showing the appearance of exploration based on history.
[0045] Fig.10 This is a diagram showing an example of the search results related to the DG44 strain.
[0046] Fig.11 This is a diagram showing an example of the search results related to the K1 strain.
[0047] Fig.12 It is a diagram showing the structure of a culture result prediction unit.
[0048] Fig.13 : is a flowchart showing the process of predicting the culture results.
[0049] Fig.14 This is a diagram schematically showing a simulator for a biological behavior quantity prediction process.
[0050] Fig.15 It is a schematic diagram showing the metabolic pathway of a cell.
[0051] Fig.16 This is a diagram for explaining the outline of the Michaelis-Menten equation.
[0052] Fig.17 This is a diagram schematically showing a conventional cell metabolism model.
[0053] Fig.18 This is a schematic diagram of creating a cell death model.
[0054] Fig.19 It is a diagram showing the structure of a cell property estimation unit.
[0055] Fig. 20 is a flowchart showing a method for creating a mathematical model of a cell.
[0056] Fig.21 This is a diagram for explaining the feature extraction process.
[0057] Fig. 22 This is a flowchart showing the process of creating a mathematical model.
[0058] Fig.23 This figure explains the process of creating a mathematical model. DETAILED DESCRIPTION
[0059] Embodiments of the method, device, and program for searching for culture medium composition and estimating cell characteristics of the present invention are described in detail. In the description, the drawings are referred to as needed.
[0060] [Overview of the process]
[0061] Below, mainly to the case of exploring the composition of the culture medium when the antibody-producing CHO cell is set as the object and the productivity of the optimization target is set as its antibody production amount, the example is described. However, the object cell of the present invention can also be any one of CHO cells, HEK cells for producing viruses for gene therapy, and iPS cells for regenerative medicine. In addition, the productivity as the target can also be one or more of the proliferation of the object cell, the antibody production amount or the virus production amount, and the differentiation induction success rate of the object cell to a specific tissue. In the case of selecting these object cells or productivity, it is also possible to constitute an embodiment or embodiment of the same purpose as the following content. In addition, the culture method can also appropriately select a well plate (microtiter plate, microplate), a flask, batch, fed culture, perfusion culture, etc.
[0062] Here, the culture medium refers to a mixture of nutrients used to culture cells, generally including amino acids, vitamins, metals, etc., but the components are not limited thereto, and therefore, it can be expressed as a multidimensional vector (or, each dimension is a point in a multidimensional space corresponding to each nutrient (hereinafter referred to as "culture medium composition space")). In the present invention, the amount or content ratio of part or all of the composition (nutrients) of such a culture medium can be optimized. In addition, the culture medium composition space is generally multidimensional (the number of dimensions varies depending on the number of nutrients), but in the following description, it is sometimes expressed in one dimension for simplicity (described later). Figure 6 , Figure 8 horizontal axis, etc.).
[0063] [Workflow of the present invention]
[0064] exist Figure 1 The workflow (overview of the processing) assumed in the present invention is shown in . The culture medium prompting side (for example, the exploration estimation system 10 described later) prompts a plurality of culture medium candidates (step S1), and the evaluation side (for example, the user of the culture evaluation system 600 described later) uses the culture medium candidate to actually culture a certain cell (target cell), obtain the culture result, and evaluate the result (step S2). The evaluation side (for example, the culture evaluation system 600 itself) can also obtain and evaluate the culture result through simulation, etc. without relying on actual culture (step S2). The evaluation result is notified to the prompting side (step S3), and the cycle is repeated.
[0065] In this case, the purpose of the system as a whole is to obtain the best possible culture results while suppressing the number of groups or repetitions of the culture medium presented, or under its constraints. The number of groups or repetitions of the culture medium presented is, for example, one cycle of steps S1 to S3, in which five culture media are presented and evaluated, and the cycle is set to about 3 to 5, but it can also be a different condition. In addition, when repeating the cycle, the culture medium contained in the presented group can also be changed according to the evaluation result.
[0066] In addition, Figure 1 In the workflow, for example, the culture medium is a CHO cell culture medium (CHO: Chinese Hamster Ovary), the cells are certain CHO cells (not only original CHO cells, but also CHO cells of various systems produced by mutation or change, such as CHO-DG44, CHO-K1, CHO-S, etc.), and the culture result is the amount of antibody produced after culturing the cells for a specified period of time. However, as described above, the culture medium, cells, and culture results of the present invention are not limited to these examples.
[0067] [First embodiment]
[0068] Figure 2 1 is a block diagram showing the structure of the exploration estimation system 10 and the system associated therewith according to the first embodiment of the present invention. The exploration estimation system 10 is a system (exploration estimation system) for exploring the composition of the culture medium for cells and estimating the characteristics of cells, and can be realized using a computer. Figure 2 As shown, the exploration estimation system 10 includes a processing unit 100 (processor), a storage unit 200, a display unit 300, and an operation unit 400, which are connected to each other to send and receive required information. For these components, various settings can be adopted, and each component can be set in one part (inside a frame, in a room, etc.), or can be set in separate places and connected via a network. In addition, the exploration estimation system 10 can be connected to an external server 500 and an external database 510 via a network NW such as the Internet to obtain information such as cell characteristics, culture medium composition, and cell strains as needed.
[0069] The culture evaluation system 600 is a system for acquiring, evaluating, and managing cell culture data, and is connected to the exploration estimation system 10 via the network NW. As described later, the exploration estimation system 10 presents a plurality of culture medium candidates, and the user of the culture evaluation system 600 actually cultures cells (target cells) for the culture medium candidates to obtain cell culture results. The culture evaluation system 600 can also obtain cell culture results by simulations that do not rely on actual culture. The exploration estimation system 10 obtains the culture results of cells for the given culture medium composition from the culture evaluation system 600.
[0070] [Structure of the processing unit]
[0071] Figure 3 1 is a block diagram showing the structure of the processing unit 100 (processor). The processing unit 100 includes a culture result prediction unit 102, a cell characteristic estimation unit 104, an exploration estimation unit 110, an input / output control unit 120, a communication control unit 122, a ROM 130 (ROM: Read Only Memory), and a RAM 140 (RAM: Random Access Memory). The exploration estimation unit 110 includes a culture result acquisition unit 112, an additional estimation evaluation unit 114, a culture medium composition prompt unit 116, and a repeat control unit 118. In addition, the input / output control unit 120 and the communication control unit 122 control input / output and communication with an external server 500, an external database 510, a culture evaluation system 600, etc. The details of the exploration and estimation processing of the processing unit 100 using these units will be described later.
[0072] The functions of each part of the processing unit 100 described above can be implemented using various processors. Among the various processors, for example, a general-purpose processor, i.e., a CPU (Central Processing Unit), which executes software (programs) to implement various functions is included. In addition, among the various processors described above, processors, i.e., programmable logic devices (PLDs), such as FPGA (Field Programmable Gate Array), which can change the circuit structure after manufacturing, are also included. In addition, processors, i.e., dedicated circuits, such as ASIC (Application Specific Integrated Circuit), which have a circuit structure specially designed to perform specific processing, are also included in the various processors described above.
[0073] The functions of each part can be implemented by one processor or by combining multiple processors. In addition, multiple functions can also be implemented by one processor. As an example of multiple functions being composed of one processor, there is first a method in which a processor is composed of a combination of one or more CPUs and software, such as represented by computers such as clients and servers, and the processor is implemented as multiple functions. Second, there is a method in which a processor is used to implement the functions of the entire system, such as represented by a system on chip (System On Chip: SoC). In this way, various functions are composed of one or more of the above-mentioned various processors as a hardware structure. In addition, more specifically, the hardware structure of these various processors is a circuit (circuitry) composed of a combination of circuit elements such as semiconductor elements.
[0074] When the above-mentioned processor or circuit executes software (program), the processor (computer) readable code of the executed software is pre-stored in a non-temporary tangible recording medium such as ROM130, and the processor refers to the software. The software pre-stored in the non-temporary tangible recording medium includes a program (a program for exploring the composition of the culture medium for cells and estimating the cell characteristics; hereinafter referred to as "exploration estimation method") for executing the method of the present invention (a method for exploring the composition of the culture medium for cells and estimating the cell characteristics; hereinafter referred to as "exploration estimation program"). The code can also be recorded in various non-temporary tangible recording media such as optical magnetic recording devices and semiconductor memories instead of ROM130. When performing processing using software, for example, RAM140 is used as a temporary storage area. In addition, for example, data stored in an EEPROM (Electronically Erasable and Programmable Read Only Memory) not shown in the figure can also be referred to.
[0075] The above-mentioned “non-transitory tangible recording medium” does not include intangible recording media such as carrier signals and propagation signals themselves.
[0076] [Structure of storage unit]
[0077] The storage unit 200 is composed of various non-transitory tangible recording media such as optical and magnetic recording media and semiconductor memories, and its input and output control unit, and stores information such as culture medium composition, cell characteristics, and culture results.
[0078] [Structure of display unit and operation unit]
[0079] The display unit 300 includes a monitor 310 (display device) and can display input information, information stored in the storage unit 200, and results of processing performed by the processing unit 100. The operation unit 400 includes a keyboard 410 and a mouse 420 as input devices and / or pointing devices, and the user can perform operations required for executing the exploration estimation method and the exploration estimation program of the present invention via these devices and the screen of the monitor 310.
[0080] [Exploring the treatment of estimation methods]
[0081] Figure 4 is a diagram conceptually showing the configuration of the discovery estimation system 10 and the training evaluation system 600. Figure 5 1 and 2 are flowcharts showing the processing of the exploratory estimation method of the first embodiment. The exploratory estimation method and the processing of the exploratory estimation program of the first embodiment will be described below with reference to these figures. In order to simplify the description, some elements of the exploratory estimation system 10 are omitted from illustration as needed.
[0082] In the first embodiment, an example of searching for the composition of a culture medium is described with antibody-producing CHO cells as the target and the productivity to be optimized being the amount of antibody produced.
[0083] Here, the culture medium composition is set to X i , set the cell characteristics to C and the antibody production to Y i (The subscript i indicates that there can be multiple cells, and the cell type is assumed to be one in a series of workflows.) i Can be freely constructed, C unknown, Y i The true value of Y is the value that can only be obtained through experiments. i X i To solve this problem, a system combining the following elements was constructed.
[0084] In addition, in the first embodiment, Y as the optimization target i is the amount of antibody produced, but in Y i The experiment data may be appropriately accompanied by other experimental data (for example, the time series of the number of cells, the amount of antibody produced, the change of the culture medium composition, and other accompanying data). Here, "experimental data" is, for example, the time series of the number of cells, the amount of antibody produced, the change of the culture medium composition, etc. The cell characteristic estimation unit 104 (processing unit 100; processor) can use these data to estimate the cell characteristics. That is, the cell characteristic estimation unit 104 can estimate the cell characteristics not only by using the culture results but also by using the experimental data in the middle (refer to the "Specific structure and processing example of the cell characteristic estimation unit" described later).
[0085] [Prediction of culture results]
[0086] The culture result prediction unit 102 (processor) predicts the culture result of the target cells in response to the composition of the medium given (step S100: prediction step). The prediction of the culture result can be expressed as Y i ^ =F(X i , C), that is, the culture result prediction unit 102 inputs the culture medium composition X i and cell characteristics C, predicting the amount of antibody produced Y i ^ Here, the superscript symbol "^ (hat-shaped)" indicates a predicted value (the same applies below). However, in practice, since the true cell characteristic C is unknown, the estimated value C^ of the cell characteristic is input instead.
[0087] The above-mentioned F can be a function, or a predictor obtained through machine learning, or a simulator of a model library, etc. When a simulator is used, a simple simulation that can be quickly calculated can also be performed. The details of the specific structure and processing example of this culture result prediction unit 102 will be described later (refer to "Specific structure and processing example of the culture result prediction unit"). In addition, the above-mentioned true value "Y i "The experimental system is also regarded as a kind of (including true C)F, called F0.
[0088] [Tips on culture medium composition]
[0089] The culture medium composition presenting unit 116 (processor) presents the culture medium composition X to the culture evaluation system 600. i The user of the culture evaluation system 600 uses the culture medium composition X i The target cells are actually cultured to obtain the culture results. Alternatively, the culture evaluation system 600 may also obtain the culture results through simulation (eg, high-precision simulation). The culture evaluation system 600 provides the exploration estimation system 10 with the culture results obtained through actual culture or simulation.
[0090] [Obtaining culture results]
[0091] The culture result acquisition unit 112 (processor) acquires the culture result of the target cells for a part or all of the medium composition given (step S110: acquisition step). The acquired culture result is the medium composition X i and antibody production Y i Pairing ( Figure 4 {X i ,Y i}; also referred to below as “historical R”).
[0092] [Estimation of cell characteristics]
[0093] The cell characteristic estimation unit 104 (processor) estimates the cell characteristics of the target cells using the culture medium composition and the culture results predicted in step S100 (step S110: estimation process). Here, "cell characteristics" can be defined in various ways biologically or cytologically, but in the present invention, "cell characteristics" are defined as "a mechanism for determining the productivity of cells (the proliferation of the above-mentioned cells or the amount of antibody produced, etc.) when the culture medium composition is given." Specifically, the "cell characteristics" of this embodiment are mechanisms for determining the amount of antibody produced Y for the culture medium composition X, which can be represented by, for example, a function, a machine learning model, or a cell mathematical model. Therefore, for example, the coefficients of a function, the parameters of a machine learning model, or the parameters of a metabolic pathway model in a cell mathematical model can be considered to be part of the cell characteristics of the present invention.
[0094] The estimation of the cell characteristics by the cell characteristics estimation unit 104 in step S110 can be expressed as C^=G({X i ,Y i That is, the cell characteristic estimation unit 104 receives one or more "medium composition X" i and its antibody production Y i When the pairing of Y is performed (the above-mentioned “history R”), the cell characteristics are estimated. Here, as mentioned above, Y i It may also be accompanied by experimental data. Here, G may also be a function that returns one of the functions F, such as F^=F_C^(X) according to the above-mentioned cell characteristics (broadly regarded as returning C^). In addition, the cell characteristic estimation unit 104 may also be a parameter determiner for machine learning, or a model estimator for a model library, etc. The details of the specific structure and processing examples of the cell characteristic estimation unit 104 will be described later (refer to "Specific structure and processing examples of the cell characteristic estimation unit").
[0095] Here, finally, when some history R is obtained, first, the history R is input to G (cell characteristic estimation unit 104) to estimate the cell characteristic C^, and then a plurality of culture medium composition candidates {X^} are input to F (culture result prediction unit 102). i}, the medium composition X that produces the maximum antibody yield Y is returned as the optimal medium. However, assuming that the culture is repeated, how to construct the medium composition that should obtain the culture result, that is, the medium composition candidate {X i Therefore, in the first embodiment, the next element is further introduced into the discovery estimation system 10 .
[0096] [Necessity evaluation of additional estimates]
[0097] The additional estimation evaluation unit 114 (processor) evaluates the necessity of further estimating the cell characteristic (step S130: evaluation step). The additional estimation evaluation unit 114 can evaluate the necessity of the additional characteristic by, for example, the following method.
[0098] (Evaluation method 1)
[0099] Evaluation method 1 can be expressed as L = V1 ({Y i ^,Y i That is, the additional estimation evaluation unit 114 inputs the predicted value Y of the antibody production amount based on the predictor (culture result prediction unit 102). i ^(predicted culture results) and the true value Y of the antibody production based on the culture i The pairing of (representing the difference between the predicted culture result and the obtained culture result; hereinafter also referred to as the "actual difference") is used to evaluate the deviation between the estimated value and the true value of the cell characteristic. i ^Close to Y i , it is appropriate to judge that C is also close to C. For example, the deviation between the function F^ that takes C^ instead of C itself and the true unknown function F0 can also be evaluated.
[0100] The additional estimation evaluation unit 114 can evaluate that "the necessity of adding characteristics is high" when the evaluated deviation is large (exceeding the determined deviation), and can evaluate that "the necessity of adding characteristics is low" when the evaluated deviation is small (below the determined deviation). When the evaluation is "the necessity of adding characteristics is low", the repetition control unit 118 can determine that "the end criterion is satisfied" in step S140 and end the repetition.
[0101] (Evaluation method 2)
[0102] Evaluation method 2 can be expressed as L = V2 ({X i ,Y i That is, the additional estimation evaluation unit 114 receives the history R (culture medium composition X i and medium composition X i The culture results in Y i (e.g., antibody production) pairing), the adequacy of the exploration can be evaluated. i If the distribution of the candidates for the possible culture medium composition X in the above-mentioned culture medium composition space has been sufficiently covered, the additional estimation evaluation unit 114 may also evaluate that it is difficult to make the estimated value of the cell characteristic closer to the true value.
[0103] In the evaluation method 2, the additional estimation evaluation unit 114 can make an evaluation by assuming that "the culture result has continuity with respect to the change of the medium composition". This is because the cell culture result can be assumed in most cases not to change significantly with respect to a small change in the medium composition. Therefore, the additional estimation evaluation unit 114 can also make an evaluation as "in the above-mentioned medium composition space, with respect to X i Changes in Y i The need for additional estimation in the area with small changes in X is low, whereas the evaluation is "relative to X i Changes in Y i For example, in the area where the change is large, additional estimation is necessary. Figure 6 In the conceptual diagram for explaining the necessity of additional estimation, the horizontal axis represents the culture medium composition space, and the vertical axis represents the culture result, the additional estimation evaluation unit 114 can evaluate the area near the area indicated as "high" as "because the i Changes in Y i The area near "Low" is evaluated as "Since the change in Yi is small relative to the change in Xi, the need for additional estimation is low."
[0104] [Exploration and determination of the next culture medium composition candidates]
[0105] The repeat control unit 118 (processor) causes the culture medium composition prompt unit 116 (processor) to explore and determine the culture medium composition for which the culture result should be obtained next, i.e., the culture medium composition candidate (steps S140, S150: control process, exploration and determination process), based on the necessity of additional estimation evaluated in step S130 and the prediction result in the prediction process.
[0106] Specifically, when there is no need for additional estimation (YES in step S140: control process), the repetition control unit 118 (processor) ends the processing because the end criterion is satisfied. When there is a need for additional estimation, the repetition control unit 118 causes the culture medium composition presentation unit 116 (processor) to search and determine the culture medium composition for which the culture result should be obtained next, i.e., the culture medium composition candidate, because the end criterion is not satisfied (if NO in step S140, the process proceeds to step S150: control process, search and determination process).
[0107] Furthermore, in step S140 , the repetition control unit 118 may determine that “the termination criterion is satisfied” and terminate the repetition when the number of repetitions exceeds a predetermined number of times.
[0108] The culture medium composition prompting unit 116 can search and determine the culture medium composition candidate as follows according to the evaluation method used by the additional estimation evaluation unit 114. Specifically, when the additional estimation evaluation unit 114 uses the evaluation method 1, the culture medium composition prompting unit 116 preferably sets the culture medium composition candidate X in the range (region) where the exploration is not sufficient in the culture medium composition space when the deviation is large, and when the deviation is small, it preferentially sets the culture medium composition candidate X where the culture result Y is expected to increase based on the current estimation result. In addition, when the additional estimation evaluation unit 114 uses the evaluation method 2, the culture medium composition prompting unit 116 sets the culture medium composition candidate X in the region (relative to X) when the need for additional estimation is high. i Changes in Y i On the contrary, when the need for additional estimation is low, priority is given to setting X where Y is expected to increase based on the current estimation result.
[0109] [Effects of the First Embodiment]
[0110] According to the exploration estimation system 10 (exploration estimation method, exploration estimation system, exploration estimation program) of the first embodiment, by repeating the prediction and acquisition of culture results, the estimation of cell characteristics, the evaluation of the necessity of additional estimation of cell characteristics, and the exploration and determination of candidate culture medium compositions, the culture results can be effectively optimized even for cells with unknown characteristics.
[0111] [Additional Embodiments of the Method for Estimating Cell Characteristics]
[0112] The following describes an additional implementation method for estimating cell characteristics. For example, if limited to CHO cells, it is known that various CHO cell lines are not "equally spaced" from each other, but can be modeled as a pedigree tree through derivative relationships, etc. Specifically, Figure 7 As shown, among CHO cells, there are multiple cell lines having different characteristics from the original CHO cells (CHO-ori), such as "CHO-K1", "CHO-S", and "CHO-DG44".
[0113] also, Figure 7 The pedigree tree is recorded in the following non-patent document 1.
[0114] [Non-patent document 1] "Wurm, FM; Wurm, MJ Cloning of CHO Cells, Productivity and Genetic Stability A Discussion. Processes 2017, 5, 20." [Retrieved on March 1, 2022], Internet (https: / / www.researchgate.net / publication / 316314849_Cloning_of_CHO_Cells_Productivity_and_Genetic_Stabili ty-A_Discussion / link / 59410f0045851566e1beae32 / download)
[0115] In addition, please cite the above paper according to http: / / creativecommons.org / licenses / by / 4.0 / .
[0116] The relationship of the pedigree tree as described above is not limited to CHO cells, and there is also a pedigree tree in HEK cells, and iPS cells may also produce such a relationship through long-term culture. The "cells" mentioned in the present invention generally include, for example, host cells with a certain gene established by culture or change, and antibody-producing cells (sometimes referred to as "cell strains") in which antibody-producing genes are inserted, but such systems or cell strains can be defined by genes, or can be defined with reference to conventionally known systems (in the case of CHO cells, "CHO-K1", "CHO-S", "CHO-DG44", etc.). In addition, there are various definitions in "cell strains", for example, it can be defined as "cells separated from an organism, or cells that have undergone certain processing on genes, etc., can be in a state where they can be stably proliferated and cultured for a long time while maintaining certain properties."
[0117] Based on the definition of this cell system, it is possible to grasp the cell characteristics of various CHO cell lines that can be obtained through some methods, and to form a set Cs of cell characteristics such as {C1, C2, C3, ...} in advance. Therefore, even if the cell characteristic C is unknown, the cell belongs to any cell system, and therefore, it can be expected that the characteristics are close to the representative cells of any cell system. That is, in the case of known cell characteristics of CHO cell lines, it can be assumed that "the true cell characteristic C of the unknown cell is close to any one of the cell characteristics contained in the set Cs" (such an assumption is called "prior knowledge"). In addition, especially in the case where it is known that the unknown cell is a derivative cell of a specific cell system, it can also be assumed that C is close to C_k (the known information that C belongs to a specific cell system is called "prior knowledge").
[0118] Here, the underlined notation "C_k" means that the cell characteristics are set for a certain information using the prior knowledge "k". In the above example, there is the prior knowledge that "a certain unknown cell is close to a specific cell characteristic Ck among the cell characteristics included in the set Cs".
[0119] Then, for example, the initial value of C can be set to C = C_k. Alternatively, in the system that takes in C, the initial value can also be set to F = F_k. Figure 7 For example, even if it is unknown whether a certain cell is CHO-Sc, if it is known that it is a CHO-S0 derived cell line, "C=C_S0" or "F=F_S0" is set as the initial setting, thereby expecting more efficient exploration than when "C=C_ori" is set as the initial setting. The subsequent processing can apply the same method as the first embodiment described above.
[0120] In this way, the cell property estimation unit 104 (processor) can perform exploration or estimation by assuming that the cell property of the target cell is similar to the cell property of one or more cells in the known cell population (i.e., using prior knowledge). Figure 8 This exploration using prior knowledge is represented in Figure 8 The unknown function is estimated in . The gray-shaded area (the width of the vertical axis) represents the estimated width of the cell characteristics for a certain medium composition (a point on the horizontal axis). In addition, the medium composition actually corresponds to a point in the medium composition space which is a multidimensional space, but in Figure 8 For convenience, we use one-dimensional space ( Figure 8 When estimating cell characteristics, knowledge (prior knowledge) such as known functions approaching unknown functions is used.
[0121] [How to apply the solution to the source code problem]
[0122] Furthermore, the specific implementation method is explained by applying the source code problem. The source code problem refers to the problem of optimizing an unknown function based on the observation history with a limited number of trials. In the present invention, for such a generalized problem, in addition to allocating specific components such as cells, cell characteristics, and culture medium composition, domain-specific elements are also introduced (the necessity evaluation of additional estimates assuming the continuity of the results, the use of prior knowledge related to the similarity of characteristics, etc.). Specifically, in this method, the culture results are regarded as the observation history, the predicted units in the prediction process are regarded as the unknown function, and the shape of the unknown function is regarded as the cell characteristic, and the unknown function that best fits the observation history is estimated.
[0123] Specifically, for example, using GP-UCB (Gaussian Process Upper Confidence Bound), we construct Fgp_t(X, C^) = {μ_t(X) + σ_t(X), μ_t(X) - σ_t(X)} (see Fig. 9 ). Here, Fgp refers to the function of the width (interval) of F, μ represents the response prediction value of F (i.e., F^), σ represents the unreliability of the prediction, and F^ is updated based on the history, so it is different at the time t of each trial. In addition, t(X), σ_t(X), and t(X, C^) are the unreliability of the culture medium composition X and the cell characteristic C in the t-th trial, and the estimated value C^ of the cell characteristic for the culture medium composition X.
[0124] Among the multiple culture medium compositions X, the culture medium composition X that maximizes Fupper_t(X_C)=μ_t(X)+σ_t(X) (i.e., the upper limit of the interval) is one of the culture medium composition candidates that the culture medium composition prompting unit 116 should prompt. The culture medium composition prompting unit 116 can, for example, set the culture medium composition X that maximizes each Fupper in a plurality of intervals set randomly or appropriately as a prompt set (prompted culture medium composition candidates). In this way, the culture medium composition prompting unit 116 can be configured. In GP-UCB, based on a certain history {X i ,Y i}, fixed F(X i ,C)=μ_t(X i )=Y i , further reduce X i The surrounding σ_t. Thus, the cell characteristic estimation unit 104 and the additional estimation evaluation unit 114 can be configured. That is, Fgp is updated, and the candidates of the range to be searched are also updated.
[0125] Furthermore, for example, if the function for a known cell is set to F(X, C_k)=μ_k(X), it can be initially set to F_0(X, C)=μ_k(X), so the assumption of the similarity of the above-mentioned cell characteristics can also be taken into account.
[0126] Fig. 9 This is a situation of exploration based on GP-UCB (a diagram showing the predicted state after a certain number of explorations). The horizontal axis of the diagram corresponds to the culture medium composition, and the vertical axis corresponds to the estimated value of the cell characteristics. In addition, in the diagram, the solid line represents the predicted value of the cell characteristics relative to a certain culture medium composition, and the gray area represents the unreliability relative to the predicted value. The cross symbol represents the observation result so far, and the circle represents the next exploration estimate candidate. Fig. 9Part (a) of FIG. 1 shows a state where the culture medium composition corresponding to the circle is determined as a candidate for exploration estimation. The culture medium composition presentation unit 116 selects a point where the cell characteristic becomes maximum or maximum as a candidate for exploration estimation. Fig. 9 Part (b) of FIG. 1 shows a state where the observed value corresponding to the circle in part (a) is obtained and the unreliability of the part (the part of the circle with a dotted line) converges. In this state, the culture medium composition prompting unit 116 can further determine the culture medium composition corresponding to the circle (the point where the cell characteristic becomes maximum or maximum) as the next exploration estimation candidate.
[0127] A cell simulator or a similar function thereof may also be applied to the above-mentioned culture result prediction unit 102. When a cell simulator is applied, the predicted unit, unknown function, and / or a similar function to the unknown function in the prediction process can be expressed by cell simulation including a metabolic pathway model (refer to the "Specific structure and processing example of the culture result prediction unit" described later).
[0128] [Example]
[0129] The exploration effect of the present invention based on simulation is shown. Here, a cell simulator (culture result prediction unit 102) is constructed by the method described in the "Specific structure and processing example of the culture result prediction unit" below. However, in the following examples, DG44 is the DG44 original cell strain of CHO cells, X is the cell strain derived from DG44, and K is the K original cell strain (for the relationship between these cell strains of CHO cells, refer to Figure 7 system diagram).
[0130] In this embodiment, for
[0131] (1) When the unknown cell is set to DG44 (the same system as X)
[0132] (2) When the unknown cell is set to K,
[0133] Furthermore, for
[0134] (a) Using prior knowledge X
[0135] (b) No prior knowledge of X is used
[0136] In the case of a simulator, the effect of the exploration of the fourth method of the present invention (the method using the solution to the source code problem) was studied. In addition, under the condition that the number of culture media prompted at one time was set to five, it was studied whether the culture medium composition exceeding the reference culture medium could be found in a short five-trial run. Fig.10 , 11 The results are shown in FIG. Fig.10 This is the result when the unknown cell is set to DG44. Fig.11 is the result when the unknown cell is set to K. In addition, Fig.10 , 11 In the figure, the horizontal axis represents the number of exploration trials, and the vertical axis represents the amount of antibody production (an example of cell characteristics).
[0137] As a result, for either the case of "using prior knowledge X" or the case of "not using prior knowledge X", as the number of trials increased, a culture medium composition that produced antibodies in an amount exceeding that of the baseline culture medium was successfully discovered through limited exploration, and in particular, for cell lines of the same system, it was confirmed that the use of prior knowledge was effective.
[0138] [Specific structure and processing example of the culture result prediction unit]
[0139] The specific configuration and processing example of the above-mentioned culture result prediction unit 102 (processor) will be described. Fig.12 1 is a diagram showing a configuration example of the culture result prediction unit 102. In the example of the figure, the culture result prediction unit 102 includes a culture environment input unit 102A, a biological behavior quantity prediction unit 102B, a cell environment change unit 102C, a repeat unit 102D, and an output unit 102E. These units can be configured using various processors and memories (ROM, RAM, etc.) in the same manner as described above with respect to the processing unit 100.
[0140] 《Methods for predicting cell culture outcomes》
[0141] Fig.13 This is a flowchart showing a method for predicting the culture results of target cells by the culture result prediction unit 102 based on the above-mentioned structure. The method comprises: a culture environment input step for inputting a culture environment consisting of a culture medium composition and culture conditions for culturing cells; a biological behavior quantity prediction step for predicting the biological behavior quantity based on the culture environment input in the culture environment input step; and a cell environment change step for changing the cell environment based on the biological behavior quantity predicted by the biological behavior quantity prediction step. In addition, there is a repetition step for repeating the biological behavior quantity prediction step and the cell environment change step for predicting the biological behavior quantity based on the cell environment changed by the cell environment change step until a prescribed time as the culture time has passed. There is an output step for outputting the cell environment changed by the cell environment change step after a prescribed time has passed.
[0142] Next, each process is described.
[0143] <<Cultivation environment input process (step S12)>>
[0144] The culture environment input unit 102A performs a culture environment input process (step S12). The culture environment input process is a process for inputting a culture environment consisting of a culture medium composition and culture conditions for culturing cells. The culture environment is input by the user and processed by the culture environment input unit 102A.
[0145] The composition of the culture medium includes the culture medium components and the amounts of the culture medium components of the culture medium for culturing cells. The culture conditions are set conditions for optimizing the culture process of cells, and the culture conditions include the culture method, the size and type of the culture container, oxygenation, supplementation of the culture medium and nutrient sources, addition of pH adjusters and carbon dioxide, removal of the culture medium containing growth-inhibiting byproducts, and harvesting of the target product.
[0146] Examples of the culture method include the presence or absence of sterilization treatment, whether the culture medium is liquid or solid, the culture temperature, etc. In addition, the culture conditions are not only the conditions at the start of the culture, but also the presence or absence of oxygenation and the amount of addition, the presence or absence of supplementation of the culture medium and nutrient sources and the amount of supplementation, the addition of pH adjusters and carbon dioxide, the presence or absence of removal of the culture medium containing growth-inhibiting byproducts and the amount of supplementation, and the presence or absence of harvesting of the target product and the amount of harvest, etc., during the culture period.
[0147] <<Biological behavior quantity prediction process (step S14)>>
[0148] The biological behavior amount prediction unit 102B performs a biological behavior amount prediction process (step S14). The biological behavior amount prediction process is a process for predicting the biological behavior amount based on the culture environment input in the culture environment input process. The biological behavior amount refers to the time change amount of the total number of cells, the time change amount of the substances and amounts produced by the cells, and the time change amount of the substances and amounts contained in the culture environment. The substances produced by the cells refer to antibodies produced by the cells, etc. The substances contained in the culture environment refer to culture medium components and byproducts, etc.
[0149] Fig.14 : is a diagram schematically showing a simulator for predicting biological behavior quantities. Fig.14 As shown, the biological behavior prediction process is performed using the action mechanism model (biological action mechanism model) and the machine learning model. The action mechanism model uses various mathematical models inside and outside the cell. The cell metabolism model is used for the inside of the cell, and the cell metabolism model uses the absorption rate F of each component of the culture medium absorbed by the cell. A , calculate the cell proliferation rate F g and the rate of antibody production F p For the outside of the cells, a culture medium model that calculates the concentration changes of each component of the culture medium was used. In addition, a model obtained by machine learning was used for cell death.
[0150] First, the cell metabolism model performed inside the cell is described. The cell metabolism model can obtain the proliferation rate of the cell and the production rate of the antibody by the mechanism of the cultured cell and the cell culture simulation method that reproduces the bioreaction technology. As a cell culture simulation method, it can be carried out by a method including a modeling method of metabolic flux analysis (MFA) or flux balance analysis (FBA) including a genome-scale metabolic model.
[0151] "Flux" or "metabolic flux" refers to the speed at which molecules pass through a target pathway or reaction. Among the factors that control flux are the speed of the catalyst of the enzyme in the pathway, the availability (durability) of the substrate, the concentration of the enzyme in the cell, and the proximity of the enzyme in the pathway. "Metabolic flux analysis" is a method for obtaining the amount of molecules that move from these factors. "Flux balance analysis" is an analytical method that focuses on stoichiometry and metabolic flux. In the case where constants related to metabolism cannot be fully determined, the behavior range and characteristics of the metabolic circuit that becomes the target are analyzed based on the structure of the metabolic reaction and basic constraints such as the law of conservation of matter.
[0152] Next, use Fig.15 Describe the metabolism within cells. Fig.15 This is a schematic diagram showing an example of a cell's metabolic pathway. For a cell metabolism model, first, the substrate used for metabolism in the culture medium and the upper and lower limits of the nutrients taken in, i.e., the intake constraints, are obtained. Fig.15 In the equation, A, B, C, ... are used to represent substances produced by metabolism. F1, F2, F3, ... are functions representing the time variation of the concentration of each substance. For example, F1 is the flow rate of substance A, and F2 is the flow rate of substance B produced by metabolism.
[0153] exist Fig.15 In the example shown, the conversion from substance A to substance B is performed in equal amounts by the law of conservation of matter. Similarly, substance B is equal to the total amount of conversion to substance C and substance E. In this way, the amount of substance conversion is determined by the amount of matrix and nutrients initially taken in by the cell, etc. In addition, as a parameter that limits the conversion of substances, there is a reaction rate. For example, as described above, the conversion from substance A to substance B is performed in equal amounts, but within a certain unit time, due to the limitation of the reaction rate, there is an upper limit in the conversion to substance B, and sometimes not all substance A is converted to substance B. As an intake constraint, a parameter that limits the conversion of the substance is obtained, and the intake constraint is obtained by a formula that mathematically models the mechanism of cell intake of culture medium components, and is applied when calculating the culture results.
[0154] As a mathematical model for calculating the reaction rate, for example, the Michaelis-Menten equation can be used. Fig.16The outline of the Michaelis equation is shown in FIG. The Michaelis equation is a formula related to the reaction speed V of the enzyme. When the substrate concentration S is low, the reaction speed V is proportional to the substrate concentration S. When the substrate concentration S is high, it converges to the maximum speed Vmax independently of the substrate concentration S. In addition, as other mathematical models, Fick's diffusion law can be used. Fick's diffusion law is a formula for obtaining the amount of flow through a unit area per unit time, which is proportional to the substrate concentration gradient on both sides of the diffusion coefficient D and the membrane. In the present embodiment, as a constraint condition for taking in the culture medium, the Michaelis equation or Fick's diffusion law is used, thus, it is a reaction speed faster than the reaction speed and flow obtained by the mathematical model, but the conditions within the range of the conversion amount based on the law of conservation of matter can be simulated. Therefore, more accurate cultivation prediction can be performed.
[0155] Simulation-based calculations can also use multiple mathematical models that mimic the state of cells in parallel. For example, it is possible to use a mathematical model that allows cells to proliferate and a mathematical model that allows cells to produce biological products without proliferating, and change the ratio between the multiple models according to the culture state such as the concentration of the culture medium.
[0156] Next, the culture medium model performed outside the cell is described. The culture medium model is a model for obtaining the concentration change of the culture medium around the cell when metabolism occurs under the conditions of the above-mentioned cell metabolism model, and a cell signaling model expressed by a differential equation can be used. As for the concentration change, the concentration change of each component can be obtained by solving the ordinary differential equation for time t using the Runge-Kutta method.
[0157] Here, conventional cell metabolism models including a cell death model are described. Fig.17 This is a diagram showing the outline of the previous cell metabolism model. The main culture activities of cells during the culture period are divided into cell proliferation, cell antibody production, and cell death (activity cessation). Fig.17 As shown, these activities include a proliferation phase in which cells proliferate at the beginning of culture, and an antibody production phase in which the cell proliferation rate decreases and the antibody production rate increases in the middle of culture. Then, in the late stage of culture, there is a cell death phase in which the activity of producing cells stops and the cells die. In previous simulations of culture prediction, these three culture activities were performed using a cell metabolism model, with a proliferation model used for cell proliferation, a production model used for antibody production, and a cell death model used for cell death. However, in the cell death model, a simple model that is constantized was implemented, and sufficient accuracy was not obtained. Therefore, in this embodiment, as shown in FIG. Fig.14 As shown, by simulating the cell death model using the learned model obtained by machine learning, the predicted results of the culture are calculated with high accuracy.
[0158] Fig.18This is a schematic diagram of creating a cell death model through machine learning. As a method for creating a learned model based on machine learning, the random forest method can be used. In the case of constructing a cell death model through machine learning, the number of living cells, the concentrations of twenty kinds of amino acids, and the concentration of ammonia are input as data to the input side. In addition, the survival rate of cells is input as data to the output side. As for the data used, it is actually cultivated, for example, 80% of the measured data is used for learning, and 20% is used for testing. In this way, a cell death model can be created.
[0159] In addition, the creation of the learning completion model based on machine learning is not limited to the cell death model. For data items such as cell death that are difficult to model based on a mechanism, a learning completion model is created by machine learning, thereby, the predicted result of cultivation is calculated with higher accuracy. As such data items, it is possible to enumerate the production inhibition of proliferation inhibition and antibodies, etc. In the case of creating a proliferation inhibition model by machine learning, as the data used on the output side, the proliferation amount (proliferation ratio) of cells can be used. In addition, in the case of creating an antibody production inhibition model, as the data used on the output side, the production amount (production ratio) of antibodies can be used.
[0160] In the biological behavior amount prediction step of the present embodiment, by using a learned model created by machine learning for cell death, the biological behavior amount can be measured with higher accuracy. Therefore, the prediction result of the culture can be calculated with high accuracy.
[0161] The biological behavior quantity prediction process (step S14) can be performed based on the culture conditions (medium component concentrations) acquired in the culture condition input process (step S12) and the input constraints, by a CHO cell simulator including these models (a device or system that simulates the biological behavior of CHO cells using a mechanism of action model and a learned model of cell death created through machine learning).
[0162] <Cell environment changing step (step S16)>
[0163] The cell environment changing unit 102C performs a cell environment changing step (step S16). The cell environment changing step changes the cell environment based on the biological behavior amount predicted by the biological behavior amount prediction step. The cell environment refers to the total number of cells and the substances and amounts produced by the cells.
[0164] The biological behavior prediction step calculates the time variation of the number of cells after the culture time t, the time variation of the substances and amounts produced by the cells, and the time variation of the substances and amounts contained in the culture environment. The cell environment modification step reflects the changes in the number of cells and the components in the culture medium after the culture time t in the cell environment.
[0165] Repeating Process
[0166] The repeating unit 102D controls the execution of the repeating step. The repeating step is a step of repeating the biological behavior amount prediction step (step S14) and the cell environment change step (step S16).
[0167] After the cell environment change process (step S16) is completed, the repetitive part 102D determines whether the specified culture time has passed (step S18). In the case of the specified culture time (for example, 14 days) for calculation (in the case of YSE), the calculation is terminated and the output process is turned to (step S24). In the case of not passing the specified culture time (in the case of NO), it is determined whether to supplement the culture medium and nutrients (step S20). In the case of not supplementing the culture medium and nutrients (in the case of NO), the biological behavior quantity prediction process (step S14) is returned. In the case of supplementing the culture medium and nutrients (in the case of YES), the culture medium concentration change process (step S22) is performed.
[0168] In the culture environment input process (step S12), when the culture condition is set to supplement the culture medium and nutrients after a predetermined time, the culture medium concentration change process (step S22) is performed. The supplement of the culture medium and nutrients includes the supplement of pH regulator and carbon dioxide in addition to the culture medium and nutrients. The culture medium concentration change process changes the culture medium concentration by adding the culture medium and nutrients to the cell environment at the time when the cell environment change process is completed.
[0169] In addition, the culture medium concentration changing step not only supplements the culture medium and nutrients, but also removes the culture medium containing proliferation-inhibiting byproducts and changes the culture medium concentration caused by harvesting the target product. In the culture environment input step (step S12), when the culture medium containing proliferation-inhibiting byproducts and the harvesting of the target product are set as culture conditions, the culture medium concentration is changed by removing the proliferation-inhibiting byproducts and the harvested target product from the cell environment at the time when the cell environment changing step ends.
[0170] When the process is repeated, based on the cell environment changed in the cell environment change process (step S16) or the culture medium concentration changed in the culture medium concentration change process (step S22), the input constraint conditions are obtained, and the biological behavior amount prediction process (step S14) and the cell environment change process (step S16) are performed. Afterwards, the biological behavior amount prediction process (step S14) and the cell environment change process (step S16) are repeated until the prescribed culture time (step S18) has passed.
[0171] After the prescribed culture period has passed (if YES in step S18), the calculation is terminated. As a result, the number of cells during the culture period and the amount of biological products or ammonia generated can be obtained. In addition, by plotting the biological behavior at each time of the repeated process, the cell proliferation curve and the transition of antibody production can be obtained.
[0172] <<Output process (step S22)>>
[0173] The output unit 102E performs an output step (step S22). The output step is a step of outputting the cell environment changed in the cell environment changing step (step S16).
[0174] As the result outputted by the output step, the proliferation number of cells during the culture period and the amount of biological products generated, such as antibodies or ammonia, etc. can be outputted. In addition, the proliferation curve of cells and the transition of antibody production can be outputted. The outputted result can output the above-mentioned result after a predetermined time, or the above-mentioned result after an arbitrary period of time during the culture can be outputted.
[0175] [Specific structure of the cell property estimation unit and example of processing]
[0176] A specific configuration and processing example of the above-mentioned cell characteristic estimation unit 104 (processor) will be described. Fig.19 1 is a diagram showing a configuration example of the cell characteristic estimation unit 104. In the example of the figure, the cell characteristic estimation unit 104 includes a culture data input unit 104A, a feature extraction unit 104B, a mathematical model creation unit 104C, and an output unit 104D. These units can be configured using various processors and memories (ROM, RAM, etc.) in the same manner as described above with respect to the processing unit 100.
[0177] As described below, the cell property estimation unit 104 can create a mathematical model of the cell, and estimate the cell property using the created mathematical model of the cell.
[0178] 《Methods for creating mathematical models of cells》
[0179] Fig. 20: is a flowchart showing the method for creating a mathematical model of a cell according to the present embodiment. The method according to the present embodiment comprises: a culture data input step (step S13) for inputting culture data of a cell; a feature extraction step (step S15) for extracting feature values of cell activity from the culture data input in the culture data input step; a mathematical model creation step (step S17) for creating a mathematical model of a cell according to the feature values of cell activity extracted in the feature extraction step; and an output step (step S19) for outputting the mathematical model created in the mathematical model creation step.
[0180] Hereinafter, each step will be described.
[0181] <<Cultivation data input process (step S13)>>
[0182] The culture data input unit 104A performs a culture data input step (step S13). The culture data input step is a step of inputting the culture data of cells. The culture data can be input according to the input instruction operation of the user.
[0183] The culture data may also include one or more of the total number of cells, the amount of cell secretory substances, the amount of cell-produced substances, the amount of cell metabolites, and the amount of culture medium components, and / or one or more of the temporal changes in the total number of cells, the temporal changes in the amount of cell secretory substances, the temporal changes in the amount of cell-produced substances, the temporal changes in the amount of cell metabolites, and the temporal changes in the amount of culture medium components. Here, cell secretory substances refer to substances that are not target substances among substances produced by cells and output outside the cells, and as an example, ammonia, by-products, etc. can be cited. Cell production substances refer to target substances among substances produced by cells and output outside the cells, and as an example, antibodies, etc. can be cited. Cell metabolites refer to substances that exist inside the cells among substances produced by cells. These culture data can use culture data that have actually been cultured.
[0184] <<Feature value extraction process (step S15)>>
[0185] The feature extraction unit 104B performs a feature extraction step (step S15). The feature extraction step extracts the feature of cell activity from the culture data input in the culture data input step.
[0186] The characteristic amount of cell activity refers to the concentration of a medium component, the consumption amount or consumption rate of a medium component, the production amount or production rate of a cell secretory substance, the production amount or production rate of a cell production substance, and the production amount or production rate of a cell metabolite. In the characteristic amount extraction step, the characteristic amount of cell activity is extracted from the culture data input in the culture data input step.
[0187] Fig.21This is a diagram for explaining the feature extraction process. Fig.21 , the time series culture data of the concentration of the medium component (exemplified as component X1) and the time series culture data of the antibody production amount (Y) are shown. In the feature extraction step, for example, from Fig.21 The concentration vectors of the medium components X1 to Xn at the time of culture time t are extracted as shown by arrow A from the culture data of the time series of the concentration of the medium components shown. In addition, as shown by arrow B, the consumption rates ΔX1, t to ΔXn, t of the medium components X1 to Xn are extracted. As for the consumption rate ΔX1, t of the medium component X1, the consumption rate ΔX1, t at time t is obtained by dividing the consumption amount ΔX1 of the medium component X1 from time t to time (t+1) by the culture time. The consumption rates ΔX2, t to ΔXn, t of other medium components can also be obtained in the same way. Similarly, the production rate ΔYt of antibody Y is extracted from the culture data of the time series of antibody production as shown by arrow C. As for the production rate ΔYt of antibody Y, the production rate ΔYt at time t is obtained by dividing the increase ΔY of antibody Y from time t to time (t+1) by the culture time.
[0188] In addition, regarding the concentration of the culture medium components, it is preferred to calculate the theoretical upper limit of the culture medium components that can be absorbed by the cells, or the consumption rate of the culture medium components by the cells and extract the concentration vector of the culture medium components. Fig.15 An example of a cell's metabolic pathway is described.
[0189] In the above Fig.15 In the example of the metabolic pathway of the described cell, the conversion from substance A to substance B is carried out in equal amounts by the law of conservation of matter. Similarly, substance B is equal to the total amount of the conversion to substance C and substance E. In this way, the amount of substance conversion is determined according to the amount of matrix and nutrients initially taken in by the cell. In addition, as a parameter that limits the conversion of substances, there is a reaction rate. For example, as described above, the conversion from substance A to substance B is carried out in equal amounts, but within a certain unit time, because there is a limitation on the reaction rate, there is an upper limit in the conversion to substance B, and sometimes not all substances A are converted to substance B. Therefore, in the case of a high concentration of a culture medium component, sometimes a difference is generated between the actual concentration of the culture medium component and the concentration of the culture medium component that can be consumed by culture. Therefore, in the case of creating a cell mathematical model with the actual concentration of the culture medium component, it is worried that the created cell mathematical model is not created based on cell characteristics. Therefore, the theoretical upper limit of the culture medium that the cell can absorb or the consumption rate of the culture medium is calculated and used as a characteristic amount of cell activity, thereby, a cell mathematical model based more on cell characteristics can be created.
[0190] The theoretical upper limit of the amount that the cells can absorb or the rate of consumption of the culture medium can be calculated using, for example, the Michaelis-Menten equation. Fig.16 The outline of the Michaelis equation is described. The Michaelis equation is a formula related to the reaction rate V of the enzyme. When the substrate concentration S is low, the reaction rate V is proportional to the substrate concentration S. When the substrate concentration S is high, it converges to the maximum speed Vmax regardless of the substrate concentration S. In addition, as another mathematical model, Fick's diffusion law can be used. Fick's diffusion law is a formula for obtaining the amount of flow per unit area per unit time, which is proportional to the diffusion coefficient D and the substrate concentration gradient on both sides of the membrane. By calculating the culture medium concentration using the Michaelis equation or Fick's diffusion law, a cell mathematical model based more on cell characteristics can be created.
[0191] 《Mathematical model creation process (step S17)》
[0192] The mathematical model creation unit 104C performs a mathematical model creation step (step S17). The mathematical model creation step is a step of creating a cellular mathematical model based on the feature value of the cell activity extracted in the feature value extraction step.
[0193] Fig. 22 This is a flowchart showing the process of creating a mathematical model. Fig.23 This figure explains the process of creating a mathematical model.
[0194] The cell mathematical model creation step separates the characteristic values of the cell activity extracted in the characteristic value extraction step into input factors to the target cell and output factors from the target cell (separation step: step S32). In the separation step, the concentration of the culture medium component is selected as the input factor. In addition, as the output factor, at least one of the consumption amount or consumption rate of the culture medium component, the production amount or production rate of the cell secretion substance, the production amount or production rate of the cell production substance, and the production amount or production rate of the cell metabolite is selected.
[0195] Next, if Fig.23 As shown, by exploring F(C) that satisfies the input factors and output factors at any time, a cell mathematical model is created (explored) (creation process: step S34). Here, the concentrations of the culture medium components X1 to Xn at any time t are used as input factors, and the consumption rates ΔX1, t to ΔXn, t of the culture medium components X1 to Xn at time t and the production rate ΔYt of the antibody Y at time t are used as output factors to create a model. The creation of a cell mathematical model is performed by creating a cell mathematical model that reproduces the characteristic quantities of cell activity in which the input factors and output factors are established at any time t1, t2, t3, t4, etc.
[0196] In addition, the cell mathematical model can be based on the cell mathematical model that reproduces the characteristic amount of cell activity, and the mathematical model creation unit 115 can change the cell mathematical model that becomes the basis to generate multiple cell mathematical models (generation process: step S36). Fig. 22 As shown in FIG. 1 , the change of the cell mathematical model can be performed by randomly changing the cell mathematical model that serves as the reference. Fig.23 As shown, the numerical values of 2 rows and 4 columns of the central mathematical model of the cell have been changed relative to the benchmark mathematical model of the cell (left side). In addition, the numerical values of 3 rows and 2 columns of the mathematical model of the cell on the right have been changed relative to the benchmark mathematical model of the cell. In addition, when the changes are random, a genetic algorithm can be used to create it. A genetic algorithm is a method of creating while repeating operations such as mutation or crossover. Mutation in a genetic algorithm refers to the operation of remodeling a part of a mathematical model of the cell. Crossover refers to a method of selecting an intersection point of two mathematical models of the cell, replacing it from that position backward, and creating another mathematical model of the cell.
[0197] A cell mathematical model that reproduces the characteristic amount of cell activity can be created by the mathematical model creation step (step S17). In addition, by creating a cell mathematical model that changes a part of the cell mathematical model based on the mathematical model, even if the culture data estimated by the cell mathematical model that reproduces the characteristic amount of cell activity cannot reproduce the actual culture data, it is easy to explore other cell mathematical models that can reproduce the culture data.
[0198] <<Output process (step S19)>>
[0199] The output unit 104D performs an output step (step S19). The output step is a step of outputting the mathematical model of the cell created in the mathematical model creation step (step S17).
[0200] In the output step, the plurality of cell mathematical models created in the mathematical model creation step can be output.
[0201] According to the cell characteristic estimation unit 104 (a method for creating a cell mathematical model, a cell mathematical model creation program, and a cell mathematical model creation device) described above, a cell mathematical model can be created using culture data of actual culture. Therefore, information at the genetic level is not required, and a cell mathematical model can be created based on culture data.
[0202] Although the embodiments and other examples of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications are possible.
[0203] Explanation of symbols
[0204] 10Explore the Estimation System
[0205] 100 Processing Department
[0206] 102 Cultivation Result Prediction Department
[0207] 102A Cultivation Environment Input Department
[0208] 102B Biological Behavior Quantity Prediction Department
[0209] 102C Cell Environment Change Department
[0210] 102D Repeating Section
[0211] 102E output unit
[0212] 104 Cell property estimation unit
[0213] 104A Training Data Input Department
[0214] 104B Feature extraction unit
[0215] 104C Mathematical Model Creation Department
[0216] 104D output unit
[0217] 110 Exploration Estimation Department
[0218] 112 Training Result Acquisition Department
[0219] 114 Additional Estimation and Evaluation Department
[0220] 115 Mathematical Model Creation Department
[0221] 116 Culture medium composition tips
[0222] 118 Control Department
[0223] 120 Input and output control unit
[0224] 122 Communication Control Unit
[0225] 130ROM
[0226] 140RAM
[0227] 200 Storage Department
[0228] 300 Display Unit
[0229] 310 Monitor
[0230] 400 Operation Department
[0231] 410 Keyboard
[0232] 420 Mouse
[0233] 500 External Server
[0234] 510 External Database
[0235] 600 training evaluation system
[0236] Steps of the S1 to S3 workflow
[0237] S12, S14, S16, S18, S20, S22 Steps of the culture result prediction method
[0238] S13, S15, S17, S19 Steps in the method of creating a mathematical model of cells
[0239] S32, S34, S36 Steps in the mathematical model creation process
[0240] S100~S150 explore the steps of the estimation method
Claims
1. A method for exploring the composition of a culture medium for cells and estimating cell characteristics, in, The processor performs the following steps: a prediction step of predicting, for the target cells, a culture result for the given medium composition using the cell characteristics or estimated values of the target cells; an obtaining step of obtaining a culture result of the target cell with respect to a part or all of the medium composition given; an estimating step of estimating cell characteristics of the target cell using the medium composition and the obtained culture result or the predicted culture result; an evaluation step of evaluating the necessity of further estimating the cell characteristics of the target cells based on the obtained culture results; A searching and determining step of searching and determining, based on the necessity and the predicting step, a medium composition, i.e., a candidate medium composition, for obtaining the culture result in the obtaining step; as well as The control process repeats the prediction process, the acquisition process, the estimation process, the evaluation process, and the search and determination process until the determined end criterion is satisfied.
2. The method according to claim 1, in, In the evaluation step, the processor performs the evaluation assuming that changes in the culture result with respect to changes in the medium composition are continuous.
3. The method according to claim 1 or 2, in, In the estimating step, the processor performs the estimation assuming that the cell property of the target cell is similar to the cell property of one or more cells in the known cell population.
4. The method according to any one of claims 1 to 3, in, the processor, The culture result is regarded as an observation history, the predicted unit in the prediction step is regarded as an unknown function, and the shape of the unknown function is regarded as the cell characteristic. The prediction step, the estimation step, the acquisition step, the evaluation step, and the search and determination step are performed by solving a source code problem for estimating an unknown function that best fits the observation history.
5. The method according to claim 4, in, The processor expresses the predicted unit, the unknown function and / or a similar function relative to the unknown function in the prediction process through a cell simulation including a metabolic pathway model.
6. The method according to claim 5, in, the processor, inputting the culture data of the target cells, determining the parameters of the metabolic pathway model according to the culture data, An estimation of the cell property is performed based on the determined parameters.
7. The method according to claim 6, in, The culture data includes one or more of the total number of cells of the target cells, the amount of a cell secretory substance, the amount of a cell produced substance, the amount of a cell metabolite, and the amount of a culture medium component, and / or one or more of the total number of cells of the target cells, the amount of a cell secretory substance, the amount of a cell produced substance, the amount of a cell metabolite, and the amount of a culture medium component, In the estimating step, the processor separating the parameter into an input factor to the subject cell and an output factor from the subject cell, The cell properties are estimated in such a way that the input factors and output factors hold true at any time.
8. The method according to any one of claims 1 to 7, in, The processor performs the following processing in the prediction step, thereby predicting the culture result: predicting the biological behavior of the target cells based on the medium composition and culture conditions given, Based on the predicted result of the biological behavior quantity, the cell environment of the target cell is changed; Based on the changed cell environment, the prediction of the biological behavior amount and the change of the cell environment are repeated.
9. The method according to claim 8, in, the processor, The prediction is performed by using the temporal change of the total number of cells, the temporal change of the substance produced by the target cells and the temporal change of the substance and the temporal change of the substance contained in the culture environment of the target cells and the temporal change of the amount as the biological behavior quantity, The change is performed using the total number of cells of the target cells, the substance produced by the target cells, and the amount of the substance as the cell environment.
10. The method according to any one of claims 1 to 9, in, the processor, In the evaluation step, the degree of deviation between the estimated value and the true value of the cell characteristic is evaluated based on the predicted culture result and the obtained culture result, In the searching and determining step, the medium composition candidate is searched and determined based on the deviation.
11. The method according to any one of claims 1 to 9, in, the processor, In the evaluation step, the pair of the culture medium composition and the culture result in the culture medium composition is input to evaluate the adequacy of the exploration. In the searching and determining step, the medium composition candidates are searched and determined based on the evaluation result of the adequacy.
12. The method according to any one of claims 1 to 11, in, In the control step, when the difference between the predicted culture result and the obtained culture result and / or the number of repetitions satisfies a predetermined condition, the processor determines that the end criterion is satisfied and ends the repetition.
13. The method according to any one of claims 1 to 12, in, The target cell is any one of a CHO cell, a HEK cell, and an iPS cell.
14. The method according to any one of claims 1 to 13, in, The culture result is based on one or more of the amount of antibodies or viruses produced by the subject cells, the proliferation of the subject cells, and the success rate of inducing differentiation of the subject cells into a specific tissue.
15. A device for exploring the composition of a culture medium for cells and estimating cell characteristics, in, It has a processor, The processor performs the following processing: For a target cell, predicting a culture result for a given medium composition using a cell property or an estimated value of the target cell, obtaining the culture results of the target cells for a part or all of the medium composition given, estimating the cell characteristics of the target cell using the medium composition and the obtained culture result or the predicted culture result, Based on the culture results obtained, evaluating the necessity of further evaluating the cell characteristics of the target cells, Based on the necessity and the prediction, searching and determining a medium composition that should obtain the culture result next, i.e., a medium composition candidate, The predicting, the obtaining, the estimating, the evaluating, and the exploring and determining are repeated until a determined end criterion is met.
16. A program for causing a processor to perform exploration of culture medium composition for cells and estimation of cell characteristics, in, It causes the processor to perform the following steps: a prediction step of predicting, for the target cells, a culture result for the given medium composition using the cell characteristics or estimated values of the target cells; an obtaining step of obtaining the culture result of the target cells for a part or all of the medium composition given; an estimating step of estimating cell characteristics of the target cell using the medium composition and the obtained culture result or the predicted culture result; an evaluation step of evaluating the necessity of further estimating the cell characteristics of the target cells based on the obtained culture results; A searching and determining step of searching and determining, based on the necessity and the predicting step, a medium composition, i.e., a candidate medium composition, for obtaining the culture result in the obtaining step; as well as The control process repeats the prediction process, the acquisition process, the estimation process, the evaluation process, and the search and determination process until the determined end criterion is satisfied.
17. A recording medium which is non-transitory and computer-readable, in, The program according to claim 16 is recorded therein.
