Cell culture process search method, cell culture process search program, and cell culture process search device

Abstract

To provide a cell culture process search method by which optimum culture conditions can be searched, a cell culture process search program, and a cell culture process search device.SOLUTION: The cell culture process search method includes a process condition generation step of generating a plurality of process conditions, a culture result prediction step of acquiring a culture prediction result of cells for each of the generated process conditions, the step including a simulation method including a metabolic flux analysis or a flux balance analysis using a mechanism of cells or a genome-scale metabolic model, and a step of finding an optimal process condition from the culture prediction result, and an optimized process condition acquisition step of selecting an optimal medium composition and culture condition on the basis of a proliferation amount of cells or a bioproduction generation amount of an antibody, and in the optimized process condition acquisition step, the generated process condition and the culture prediction result are input, and an optimal process is calculated using a regression model which is a learned model by machine learning.SELECTED DRAWING: Figure 7

Description

[Technical Field]

[0001] The present invention relates to a cell culture process searching method, a cell culture process searching program, and a cell culture process searching device, and in particular to a cell culture process searching method, a cell culture process searching program, and a cell culture process searching device that search for an optimal cell culture process for bioproduction, which produces chemical products using the functions of cells, microorganisms, or enzymes derived from animals or plants. [Background technology]

[0002] When examining culture conditions (culture medium, culture method, period, etc.) for bioproduction, experiments are conducted by selecting dozens of conditions from a vast number of culture condition combinations, and these experiments are repeated multiple times. A known method for selecting dozens of conditions is to use experimental design. However, with experimental design, only a very limited number of condition combinations are explored from the vast number of combinations available, and repeated experiments require significant costs (time, labor, materials, etc.).

[0003] Therefore, efficient searches for optimal conditions are being conducted using computers. In recent years, research into the analysis of metabolic pathways has progressed with the aim of controlling metabolism in fermentation and other processes. In order to search for optimal culture conditions using a computer, it is necessary to measure the structure of the metabolic pathway as well as constants related to each reaction. However, it is difficult to measure constants related to intracellular reactions, except for some model organisms. For example, Patent Document 1 listed below is known as a method for examining culture conditions for bioproduction from calculable metabolism using partially known constants.

[0004] Meanwhile, methods have been developed to understand biological behaviors such as metabolic reaction volume and growth from the structure of metabolic reaction circuits alone, without using information on reaction constants. Metabolic flux balance analysis (FBA) uses only the structure of metabolic reactions, even when metabolic constants cannot be fully measured, to analyze the behavior range and characteristics of a target metabolic circuit based on basic constraints such as the law of conservation of mass. For example, Non-Patent Document 1 below describes a method in which metabolic reactions are first described as a series of linear equations, a vector space of solutions to this system of equations is defined, this vector space is then converted into a biochemically meaningful basis, and finally, a metabolic state that maximizes a given objective function using linear programming is identified.

[0005] Furthermore, Patent Document 2 is known as a method for determining optimal metabolic circuit information by associating FBA with various environmental factors that affect metabolism, culture time, etc. Patent Documents 3 and 4, for example, are known as methods for using FBA to predict culture results taking time fluctuations into account and to obtain information for optimal cell modification. Patent Document 5 is known as a method for indirectly inferring cellular metabolic information from changes in medium composition during culture, calculating necessary components, and optimizing the medium. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2007-47994 [Patent Document 2] Japanese Patent Application Laid-Open No. 2003-180400 [Patent Document 3] US Patent Application Publication No. 2012 / 0191434 [Patent Document 4] US Patent Application Publication No. 2012 / 0185226 [Patent Document 5] Special Publication No. 2014-503220 [Non-patent literature]

[0007] [Non-Patent Document 1] Shilling, C. and Palsson, B., Proc. Nat. Acad. Sci., 95, 4193-4198, 1988 Summary of the Invention [Problem to be solved by the invention]

[0008] However, the method described in Patent Document 1 is limited to microorganisms or some biological functions that are relatively small in scale. The method described in Non-Patent Document 1 does not take into account any culture information such as culture time and culture composition. The methods described in Patent Documents 2 to 4 are not methods for predicting culture results that take time fluctuations into account, or for determining optimal culture conditions and medium compositions. The method described in Patent Document 5 is not a method for searching for optimal culture conditions and medium compositions without conducting experiments.

[0009] Thus, in the past computer-assisted searches for optimal cell culture processes, no search method has been carried out that takes into consideration the medium composition, culture method, culture period, etc.

[0010] The present invention has been made in consideration of the above circumstances, and provides a cell culture process searching method, a cell culture process searching program, and a cell culture process searching device that can search for optimal culture conditions for cell culture and bioproduction by overlooking the culture process without repeating experiments. [Means for solving the problem]

[0011] In order to achieve the object of the present invention, the cell culture process exploration method of the present invention includes a process condition generation step of generating a plurality of process conditions for culturing cells, a culture result prediction step of obtaining predicted cell culture results for each of the plurality of process conditions generated in the process condition generation step, and an optimized process condition acquisition step of finding optimal process conditions from the predicted culture results obtained in the culture result prediction step.

[0012] In one aspect of the present invention, the process conditions include at least one of a plurality of medium compositions and a plurality of culture conditions.

[0013] In one aspect of the present invention, the culture conditions are preferably set conditions including at least one of the culture method, the size and type of culture vessel, oxygen addition, supplementation of medium and nutrients, removal of medium containing growth inhibitory by-products, and harvesting of the target product.

[0014] In one aspect of the present invention, the culture conditions preferably include those designed and operated for scale-up from a small-scale process to a large-scale process.

[0015] In one aspect of the present invention, it is preferable that the plurality of process conditions generated in the process condition generating step are acquired as a matrix.

[0016] In one aspect of the present invention, the process condition generating step preferably includes selecting some of the process condition items and determining numerical values ​​for the selected items.

[0017] In one aspect of the present invention, the process condition generation step is preferably carried out by a method including at least one of determining the numerical values ​​for the process condition items based on arbitrary numerical values, determining them by generating random numbers within a specified range, determining them based on numerical values ​​obtained through experiments, and determining them based on an expansion medium mixing strategy.

[0018] In one aspect of the present invention, the determined range is preferably set using an equation that is a mathematical model of the mechanism by which an organism takes up a medium component.

[0019] In one embodiment of the present invention, the mathematically modeled equation is preferably the Michaelis-Menten equation or Fick's law.

[0020] In one aspect of the present invention, it is preferable that the method of generating random numbers determines the values ​​for the process condition items using values ​​generated by continuous uniform random numbers, continuous Gaussian random numbers, discrete random numbers, or binomial random numbers.

[0021] In one aspect of the present invention, the culture result prediction step preferably includes a cell culture simulation method that reproduces the mechanism of the cells to be cultured and the bioprocess.

[0022] In one aspect of the present invention, the cell culture simulation method preferably includes a modeling approach that includes metabolic flux analysis or flux balance analysis using a genome-scale metabolic model.

[0023] In one aspect of the present invention, the optimized process condition acquisition step preferably includes an input step of inputting the process conditions generated in the process condition generation step and the culture prediction result obtained in the culture result prediction step, a creation step of creating a trained model by machine learning using the process conditions input in the input step and the culture prediction result as learning data, and a calculation step of calculating the optimal process by solving an inverse problem using the trained model.

[0024] In one aspect of the present invention, the culture conditions are preferably set conditions including at least one of the culture method, the size and type of culture vessel, oxygen addition, supplementation of medium and nutrients, removal of medium containing growth inhibitory by-products, and harvesting of the target product.

[0025] In one aspect of the present invention, the culture conditions preferably include those designed and operated for scale-up from a small-scale process to a large-scale process.

[0026] In one aspect of the present invention, the cell is preferably a eukaryotic cell or a prokaryotic cell.

[0027] In one aspect of the present invention, the eukaryotic cells are preferably cell lines derived from animals, plants or insects, or primary cultures, or fungi.

[0028] In one aspect of the present invention, the prokaryotic cells are preferably bacteria, including Escherichia coli, Bacillus subtilis, cyanobacteria, or actinomycetes, and archaea, including methanogens, extreme halophiles, or hyperthermophiles.

[0029] In order to achieve the object of the present invention, a cell culture process exploration program according to the present invention causes a computer to execute the above-described cell culture process exploration method.

[0030] In order to achieve the object of the present invention, the cell culture process exploration device of the present invention includes a process condition generation unit that generates a plurality of process conditions for culturing cells, a culture result prediction unit that acquires a predicted cell culture result for each of the plurality of process conditions generated by the process condition generation unit, and an optimized process condition acquisition unit that finds optimal process conditions from the predicted culture result acquired by the culture result prediction unit.

[0031] In order to achieve the object of the present invention, the trained model of the present invention is a trained model used to calculate optimal process conditions, and is a trained model obtained by machine learning based on a plurality of process conditions and culture prediction results obtained from the plurality of process conditions. [Effects of the Invention]

[0032] According to the present invention, the optimal medium composition or culture conditions can be obtained without repeating experiments, which enables efficient exploration of culture processes. [Brief explanation of the drawings]

[0033] [Figure 1] FIG. 1 is a block diagram showing the configuration of a cell culture process exploration device. [Figure 2] FIG. 2 is a block diagram showing the configuration of a processing unit. [Figure 3] 1 is a flow chart illustrating a method for exploring a cell culture process. [Figure 4] FIG. 10 is a diagram illustrating a graph for calculating a numerical value using random numbers. [Figure 5]10 is a flowchart showing a culture result prediction process. [Figure 6] FIG. 1 is a schematic diagram showing cellular metabolic pathways. [Figure 7] 1 is a calculation flow of a simulation. [Figure 8] FIG. 1 is a diagram illustrating an outline of the Michaelis-Menten equation. [Figure 9] FIG. 10 is a diagram showing the results shown by the culture result prediction step. DETAILED DESCRIPTION OF THE INVENTION

[0034] Hereinafter, a cell culture process searching method, a cell culture process searching program, and a cell culture process searching apparatus according to the present invention will be described with reference to the accompanying drawings. First, a cell culture process searching apparatus for executing the cell culture process searching method of this embodiment will be described.

[0035] <Cell culture process exploration device> FIG. 1 is a block diagram showing the configuration of a cell culture process searching apparatus (hereinafter simply referred to as the “searching apparatus”) 10. The searching apparatus 10 is an apparatus that searches for optimal process conditions from among multiple process conditions and can be implemented using a computer. As shown in FIG. 1, the searching apparatus 10 includes a processing unit 100, a memory unit 200, a display unit 300, and an operation unit 400, which are interconnected to transmit and receive necessary information. These components can be installed in various configurations, such as in a single location (e.g., in a single cabinet or room) or in separate locations and connected via a network. The searching apparatus 10 can also connect to an external server 500 and an external database 510 via a network NW such as the Internet to acquire information such as culture medium composition and culture medium conditions, or conditions used in simulations, mathematical models, etc., as needed.

[0036] <Configuration of the processing unit> 2 is a diagram showing the configuration of the processing unit 100. The processing unit 100 includes a process condition generating unit 105, a culture result predicting unit 110, an optimized process condition acquiring unit 115, an output unit 120, a display control unit 125, a CPU 130 (Central Processing Unit), a ROM 135 (Read Only Memory), and a RAM 140 (Random Access Memory).

[0037] The process condition generating unit 105 generates multiple process conditions for culturing cells. The process conditions include at least one of multiple medium components and multiple culture conditions. The generation of multiple medium components involves determining the medium components and the amounts of the medium components (composition ratios of the medium components) of the medium for culturing cells, thereby generating multiple medium compositions. The determination of multiple culture conditions involves determining the culture conditions for culturing cells and numerical values ​​for each culture condition, thereby determining multiple culture conditions. The culture result predicting unit 110 calculates predicted cell culture results by simulation for the process conditions generated by the process condition generating unit 105 (the generated medium compositions and the determined culture conditions). The optimized process condition acquiring unit 115 finds optimal process conditions (medium compositions and culture conditions) from the culture prediction results obtained by the culture result predicting unit 110.

[0038] The output unit 120 outputs the process conditions generated by the process condition generation unit 105. The output unit 120 also outputs the cell culture prediction results obtained by the culture result prediction unit 110. The output unit 120 also outputs the optimal process conditions acquired by the optimized process condition acquisition unit 115. The optimized process condition acquisition unit 115 can also rearrange the culture prediction results obtained by the culture result prediction unit 110 from the best results, and the output unit 120 can output the culture prediction results in the rearranged order. The display control unit 125 controls the display of the acquired information and processing results on the monitor 310. A cell culture process search method using these functions of the processing unit 100 will be described in detail below. Note that the processing by these functions is performed under the control of the CPU 130.

[0039] The functions of each unit of the processing unit 100 described above can be realized using various processors. The various processors include, for example, a CPU, which is a general-purpose processor that executes software (programs) to realize various functions. The various processors described above also include a programmable logic device (PLD), such as an FPGA (Field Programmable Gate Array), whose circuit configuration can be changed after manufacturing. Furthermore, the various processors described above also include dedicated electrical circuits, such as an ASIC (Application Specific Integrated Circuit), which is a processor having a circuit configuration designed specifically to execute specific processing.

[0040] The functions of each section may be realized by a single processor or by a combination of multiple processors. Furthermore, multiple functions may be realized by a single processor. Examples of multiple functions configured by a single processor include, first, a configuration in which one processor is configured by a combination of one or more CPUs and software, as typified by computers such as client and server computers, and this processor realizes multiple functions. Second, a configuration in which a processor is used to realize the functions of the entire system on a single IC (Integrated Circuit) chip, as typified by a System on Chip (SoC). In this way, various functions are configured as a hardware structure using one or more of the various processors described above. Furthermore, the hardware structure of these various processors is, more specifically, an electric circuit formed by combining circuit elements such as semiconductor elements.

[0041] When the above-mentioned processor or electric circuit executes software (program), the processor (computer) readable code of the software to be executed is stored in a non-transitory recording medium such as ROM 135 (see FIG. 2), and the processor refers to the software. The software stored in the non-transitory recording medium includes a program for executing the cell culture process exploration method according to the present invention. The code may be recorded in a non-transitory recording medium such as various types of magneto-optical recording devices or semiconductor memories, instead of ROM 135. When processing using the software, for example, RAM 140 is used as a temporary storage area, and for example, an EEPROM (Electronically Erasable and Programmable Read Only Memory), not shown, is used. It is also possible to refer to data stored in

[0042] <Storage unit configuration> The memory unit 200 is composed of a non-transitory recording medium such as a DVD (Digital Versatile Disk), a hard disk, or various semiconductor memories, and its control unit, and stores the components constituting the culture medium composition and the items for determining the culture conditions, such as the culture method, the size and type of culture vessel, enzyme addition, medium and nutrient replenishment, removal of medium containing growth inhibitory by-products, and harvesting of the target product. It also stores the predicted cell culture results calculated by the culture result prediction unit 110. It also stores the optimal process conditions determined by the optimized process condition acquisition unit 115.

[0043] <Display and operation unit configuration> The display unit 300 includes a monitor 310 (display device) and can display input information, information stored in the memory unit 200, and processing results 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 to execute the cell culture process exploration method according to this embodiment via these devices and the screen of the monitor 310. Operations that the user can perform include inputting the desired cell growth amount and the production amount of bioproduction such as antibodies generated by cell growth, inputting medium components and their amounts, culture conditions, and numerical values ​​of the culture conditions when the user determines process conditions in the process condition generation unit 105, inputting the result output method in the optimized process condition acquisition unit 115, such as the number of candidate culture media to be output, specifying a mathematical model for determining numerical input constraints in the culture result prediction unit 110, and specifying the simulation method to be used.

[0044] <Processing in cell culture process exploration device> In the cell culture process searching apparatus 10 described above, a search for a cell culture process can be performed in response to a user's instruction via the operation unit 400.

[0045] <Cell culture process exploration method> 3 is a flowchart showing a cell culture process searching method of the present invention. The cell culture process searching method of the present invention includes a process condition generating step of generating a plurality of process conditions for culturing cells, a culture result predicting step of acquiring predicted cell culture results for each of the plurality of process conditions generated in the process condition generating step, and an optimized process condition acquiring step of finding optimal process conditions from the predicted culture results acquired in the culture result predicting step. The process conditions include at least one of a plurality of medium components and a plurality of culture conditions.

[0046] Each step will be described below.

[0047] <Process Condition Generation Step (Step S12)> The process condition generating unit 105 of the searching device 10 performs a process condition generating step (step S12). The process condition generating step is a step of generating a plurality of process conditions. The process conditions include at least one of a plurality of medium components and a plurality of culture conditions. The plurality of medium compositions are generated by determining the medium components of the medium in which the cells are cultured and the amounts of the medium components. The plurality of culture conditions are determined by determining the culture conditions in which the cells are cultured and numerical values ​​for each culture condition.

[0048] Regarding the formulation of the medium composition, cell culture media are composed of numerous components (over 50 components in the case of animal cell media). The formulation of the medium composition determines the components of the medium and the amount of each component, i.e., the composition ratio of the medium components.

[0049] The amounts (compositional ratios) of medium components used to generate a medium composition are preferably determined by any one of the following methods: determining the amounts based on arbitrary values, determining the amounts by generating random numbers within a specified range, determining the amounts based on experimentally determined values, and determining the amounts based on an expanded medium mixing strategy. The method of determining the amounts based on an expanded medium mixing strategy can be, for example, the method described in Martin Jordan et al., Cytotechnology, 65, 31-40, 2013. The medium composition may be generated by having a user select some medium components and having a computer select the other medium components. Furthermore, the amounts (compositional ratios) of the medium components selected by the user can also be determined. The amounts (compositional ratios) of the medium components selected by the user can be constant or variable. Alternatively, the selection of medium components and the determination of the amounts (compositional ratios) of the medium components can all be performed by a computer.

[0050] As a method for determining the numerical values ​​of the medium composition by random number generation, it is possible to use continuous uniform random numbers in which values ​​are selected with the same probability within a predetermined range, as shown by graph GA in Figure 4, or continuous normal random numbers in which values ​​are selected based on a normal distribution within a predetermined range, as shown by graph GB. It is also possible to determine the numerical values ​​using values ​​generated by discrete random numbers in which values ​​selected with a predetermined interval within a predetermined range are selected with the same probability within the predetermined range, as shown by graph GC, or binomial random numbers in which values ​​selected with a predetermined interval within a predetermined range are selected based on a normal distribution, as shown by graph GD.

[0051] Culture conditions are set conditions for optimizing the cell culture process, and include the culture method, the size and type of culture vessel, oxygen addition, supplementation of medium and nutrients, addition of pH adjusters and carbon dioxide, removal of medium containing growth-inhibitory by-products, and harvesting of the target product. As with medium composition, the culture conditions can be determined by the user, with the computer determining other culture conditions. Alternatively, the computer can determine all of the culture conditions and their numerical values. Numerical values ​​for culture conditions refer to numerically set culture conditions, such as the amount of medium, the size of the culture vessel, the amount of additives such as medium and nutrients, the rate of additive addition, and the culture temperature.

[0052] The culture method may be sterilized or not, whether the culture medium is liquid or solid, the culture temperature, etc. The size of the culture vessel may be set randomly or can be determined appropriately by the user. By changing the size of the culture vessel, culture conditions can be determined from small-scale processes to large-scale processes, and the culture results can be predicted in the subsequent culture result prediction step. This allows the design and operation of scale-up to be optimized by simulation, making it possible to accommodate cultures on a small to large scale.

[0053] Furthermore, the culture process can be set not only with the conditions at the start of culture, but also with other culture conditions, such as whether or not oxygen is added during the culture period and the amount of oxygen added, whether or not medium and nutrient sources are replenished and the amount of replenishment, addition of a pH adjuster and carbon dioxide, whether or not medium containing growth-inhibiting by-products is removed and the amount removed, and whether or not the target product is harvested and the amount harvested. These amounts added or removed and the times of addition or removal can be set as numerical values ​​of the culture conditions.

[0054] When random numbers are used to randomly determine the values ​​for each medium component and culture condition, the range of selected random numbers is preferably set using a mathematical model of the mechanism by which an organism takes up medium components. The Michaelis-Menten equation or Fick's law can be used as the mathematical model. The Michaelis-Menten equation describes the relationship between the substrate concentration and the reaction rate of an enzymatic reaction. Performing the process condition generation step within the Michaelis-Menten equation allows us to narrow down the range appropriate for the desired medium from the beginning, thereby reducing the number of medium compositions and culture conditions required in the subsequent culture result prediction step. Fick's law describes the rate of passive diffusion of drug molecules through the cell membrane directly without a carrier such as an enzyme.

[0055] The plurality of medium compositions generated in the process condition generation step and the plurality of culture conditions determined can be acquired as a medium composition matrix and a culture condition matrix, respectively. By acquiring the medium compositions and culture conditions as matrices, the culture prediction results obtained in the next culture result prediction step can be reflected in the matrices, making it possible to confirm the trends of the culture prediction results for the medium compositions and culture conditions.

[0056] <Culture result prediction step (step S14)> The culture result prediction unit 110 of the searching device 10 performs a culture result prediction step (step S14). The culture result prediction step is a step of acquiring predicted cell culture results for the plurality of process conditions generated in the process condition generation step. The predicted culture results include predictions of the number of cultured cells, as well as predictions of the amount of bioproduction such as antibodies produced by culturing the cells and the amount of by-products such as ammonia.

[0057] The culture outcome prediction step can obtain the predicted culture outcome by a cell culture simulation method that reproduces the mechanism of the cells to be cultured and the bioprocess. The cell culture simulation method can be performed by a method including a modeling approach including metabolic flux analysis (MFA) using a genome-scale metabolic model or flux balance analysis (FBA).

[0058] "Flux" or "metabolic flux" refers to the rate at which molecules pass through a target pathway or reaction. Factors that control flux include the catalytic rate 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 determining the amount of molecules moving based on these factors. "Flux balance analysis" is an analytical method that focuses on stoichiometry and metabolic flow. Even when metabolic constants cannot be fully measured, it analyzes the behavior range and characteristics of the target metabolic circuit based on the structure of metabolic reactions and basic constraints such as the law of conservation of mass.

[0059] Fig. 5 is a flowchart of the culture result prediction process. Fig. 6 is a schematic diagram showing the metabolic pathway of cells. Fig. 7 is a calculation flow of the simulation. The culture result prediction process includes a process condition acquisition process (step S32), an import constraint condition acquisition process (step S34), an optimization calculation process (step S36), and a concentration change calculation process (step S40).

[0060] <Process condition acquisition step (step S32)> The process condition acquisition step is a step of acquiring the process conditions generated in the process condition generation step (step S12).

[0061] <Import constraint condition acquisition process (step S34)> The uptake constraint condition acquisition process is a process for acquiring uptake constraint conditions, which are the upper and lower limits for the uptake of substrates, nutrients, etc. used in metabolism. In Figure 6, substances produced by metabolism are indicated by A, B, C, etc. Furthermore, F1, F2, F3, etc. are functions that indicate the change in concentration of each substance over time; for example, F1 is the flux that takes in substance A, and F2 is the flux that becomes substance B through metabolism.

[0062] In the example shown in Figure 6, substance A is converted to substance B in equal amounts due to the law of conservation of mass. Similarly, the amount of substance B is equal to the sum of the amounts converted to substances C and E. Thus, the amount of substance converted is determined by the amount of substrate and nutrients initially taken up by the cells. Reaction rate is also a parameter that limits substance conversion. For example, as mentioned above, substance A is converted to substance B in equal amounts, but due to reaction rate limitations, there is an upper limit to the conversion to substance B within a certain unit time, and not all of substance A may be converted to substance B. In the uptake constraint acquisition process, these constraints are obtained using a mathematical model of the mechanism by which organisms take up medium components and are applied when calculating the culture results. The Michaelis-Menten equation, for example, can be used as a mathematical model to calculate this reaction rate. A schematic diagram of the Michaelis-Menten equation is shown in Figure 8. The Michaelis-Menten equation relates to the reaction rate V of an 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, the reaction rate V converges to a maximum rate Vmax regardless of the substrate concentration S. Fick's law can also be used. Fick's law is an equation for determining the flux, which is the amount of fluid passing through a unit area per unit time. The flux is proportional to the diffusion coefficient D and the substrate concentration gradient on both sides of the membrane. In this embodiment, by using the Michaelis-Menten equation or Fick's law as a constraint for medium uptake, simulations can be performed without calculating conditions for a reaction rate that is faster than the reaction rate and flux determined by a mathematical model but is within the range of the conversion amount determined by the law of conservation of mass. This allows for more accurate culture predictions.

[0063] <Optimization calculation process (step S36)> In the optimization calculation step, metabolic flux (consumption rate) is calculated using a mathematical model (metabolic circuit model) related to cellular metabolism based on the medium composition (medium component concentration) and uptake constraint conditions acquired in the process condition acquisition step (step S32) and the uptake constraint condition acquisition step (step S34). The metabolic flux calculation can be performed, for example, by linear programming (LP) as shown in FIG. 6. In FIG. 6, the objective function F 11Find the condition under which F is maximized. 11 is the rate at which the target product is produced. If the desired simulation result is cell proliferation, it is the rate at which the cells grow, and if the desired bioproduction amount is the amount of antibody or other bioproducts produced, it is the rate at which by-products are produced. Within the constraints, the objective function F 11 Each function F1~F 11 , d[A] / dt=-F1, d[B] / dt=-F2, d[I] / dt=-F 11 By calculating , the metabolic flux of each reaction function at a certain time can be determined. This is the optimal metabolic flux (optimization condition) at a certain time.

[0064] In the optimization calculation step, multiple mathematical models that mimic the state of cells may be used in parallel. For example, a mathematical model of cell proliferation and a mathematical model of cell non-proliferation and only bioproduction may be used in parallel, and the ratio between the multiple models may be changed depending on the culture state, such as the medium concentration.

[0065] After the optimization calculation step (step S36) is completed, if a predetermined incubation time (for example, 14 days) for the calculation has elapsed (YES), the calculation is terminated. If the predetermined incubation time has not elapsed (NO), the next concentration change calculation step is performed (step S38).

[0066] <Concentration change calculation process (step S40)> The concentration change calculation step is a step for calculating the concentration change in the medium surrounding the cells when metabolism is performed under the optimized conditions obtained in the optimization calculation step (step S36). The concentration change is calculated by solving ordinary differential equations using the Runge-Kutta method.

[0067] After the concentration change calculation step determines the change in concentration around the cell after the elapse of the infinitesimal time t, the process returns to the uptake constraint condition acquisition step (step S34), acquires the uptake constraint condition for the concentration around the cell after the elapse of the infinitesimal time t, and then performs the optimization calculation step (step S36). Thereafter, the concentration change calculation step (step S40), the uptake constraint condition acquisition step (step S34), and the optimization calculation step (step S36) are repeated until the predetermined culture time has elapsed.

[0068] When it is determined that the culture conditions include supplementing the medium, nutrients, pH adjuster, and carbon dioxide (hereinafter also referred to as "supplementing the medium, nutrients, etc.") after a predetermined time has elapsed (in Figure 7, supplementation occurs at the start of t=2), the medium composition is calculated by adding the supplemented medium components to the medium concentration around the cells at the end of the calculation for t=1.

[0069] Furthermore, in addition to supplementing the medium and nutrients, etc., conditions may be added such that the medium containing growth-inhibiting by-products is removed and the target product is harvested during the process. In this case, too, the calculation can be performed by reflecting the changes in the medium composition around the cells in the concentration change calculation step (step S40).

[0070] After a predetermined culture period has elapsed (step S38), the calculation is terminated. As a result, at least one of the proliferation number of cells during the culture period and the bioproduction of antibodies and the like produced or the amount of by-products of antibodies and the like, such as ammonia, produced is obtained.

[0071] <Optimization process condition acquisition step (step S16)> The optimized process condition acquisition unit 115 of the searching device 10 performs an optimized process condition acquisition step (step S16). The optimized process condition acquisition step is a step of finding optimal process conditions from the culture prediction result obtained in the culture result prediction step (step S14).

[0072] The optimal process conditions can be found by selecting a predetermined number in descending order of cell proliferation amount, or a predetermined number in descending order of bioproduction amount such as antibody. The culture result prediction step produces results such as those shown in Figure 9. Of the process conditions generated in the process condition generation step (step S12), culture prediction results are obtained for each of multiple medium compositions and multiple culture conditions. By sorting these culture prediction results in descending order of the number of cultured cells or the amount of bioproduction such as antibody, the optimal medium composition and culture conditions can be selected.

[0073] Furthermore, the optimized process condition acquisition step can find optimal process conditions using machine learning (deep learning). Optimal process conditions include either an optimal medium composition or optimal culture conditions. When machine learning is performed, the multiple process conditions generated in the process condition generation step and the culture prediction results obtained in the culture result prediction step are input to the input side (input step). Next, the process conditions input in the input step and the culture prediction results are used as learning data to create a regression model (trained model) using machine learning (creation step). As a result of learning, the created regression model is used to solve an inverse problem and calculate optimal process conditions (calculation step). When solving the inverse problem, a medium composition or culture conditions that will result in a desired cell growth amount or a desired antibody production amount is calculated. As a result, when the culture prediction results obtained in the culture result prediction step are sorted in order of cell growth amount or antibody production amount, the optimal medium composition or culture conditions can be obtained using not only the top results but also the results that do not rank high, thereby enabling effective use of the calculated culture prediction results.

[0074] Furthermore, in this embodiment, the culture results are predicted using a metabolic circuit model, and information on the metabolic circuit flux and time fluctuations can be obtained during this process. Because this information corresponds to the mechanism of action of cell culture, it is possible to predict the culture results, including the mechanism. Therefore, it is possible to obtain trends in which process conditions should be selected to obtain good results, and this can be used as a reference when determining process conditions.

[0075] <Cells used in the simulation> The cells used in the cell culture process exploration method, cell culture process exploration program, and cell culture process exploration device of this embodiment are not particularly limited, and both eukaryotic cells and prokaryotic cells can be used. Examples of eukaryotic cells include cell lines derived from animals, plants, or insects, primary cultures, and fungi. Examples of prokaryotic cells include bacteria, including Escherichia coli, Bacillus subtilis, cyanobacteria, and actinomycetes, and archaea, including methanogens, extreme halophiles, and hyperthermophiles.

[0076] According to the cell culture process exploration method, cell culture process exploration program, and cell culture process exploration device of the present invention, an optimal medium composition or culture conditions can be obtained by simulation, so that the medium composition or culture conditions can be efficiently explored without repeating experiments. [Example]

[0077] The present invention will be described in more detail below with reference to examples of the present invention.

[0078] Simulations were performed to determine the optimal medium composition for culturing the CHO cell line K1. The metabolic pathway information and model for the CHO cell line K1 were obtained from BiGG Models (http: / / bigg.ucsd.edu / ), using the FBA model for the K1 line (iCHOv1_K1_final.xml). Since glucose and 20 amino acids were used as the medium composition, the culture data for glucose and 20 amino acids for the CHO cell line K1 was obtained from Appl Microbiol Biotechnol (2015) 99:4645-4657.

[0079] Next, this data was used to set the glucose and amino acid uptake parameters in the model. The Michaelis-Menten equation was used as the model for setting the parameters. Next, using these parameter values, the upper and lower limits of the medium concentration were set for each of the 20 amino acids, and 10,000 compositions (candidate medium) were generated using continuous uniform random numbers. For these candidate mediums, a culture simulation incorporating the FBA model was used to predict the results for each composition, yielding 10,000 results. The top 10 of these were selected as high-production medium candidates. Meanwhile, deep learning was performed on the 10,000 results, using the medium compositions of glucose and the 20 amino acids as input conditions. The resulting regression model was used to calculate 10 pairs of new medium compositions.

[0080] Combined with the previously obtained high-production medium candidates, we were able to obtain 20 high-production medium candidates. [Explanation of symbols]

[0081] 10. Cultivation process exploration device 100 Processing section 105 Process condition generation unit 110 Culture result prediction section 115 Optimization process condition acquisition unit 120 Output section 125 Display control unit 200 Storage section 300 Display 310 Monitor 400 Control unit 410 keyboard 420 Mouse 500 External Server 510 External Database

Claims

1. a process condition generating step of generating a plurality of process conditions for culturing cells; a culture result prediction step of acquiring a predicted result of cell culture for each of the plurality of process conditions generated in the process condition generation step, the culture result prediction step including a cell culture simulation method including a modeling approach including a mechanism of the cell to be cultured and metabolic flux analysis using a genome-scale metabolic model or flux balance analysis; an optimized process condition acquisition step of finding optimal process conditions from the culture prediction results obtained in the culture result prediction step, in which a predetermined number of cells is selected in descending order of cell proliferation amount or a predetermined number of cells is selected in descending order of antibody bioproduction amount, thereby selecting optimal medium composition and culture conditions; and The optimization process condition acquisition step includes: an input step of inputting the process conditions generated in the process condition generating step and the culture prediction result obtained in the culture result predicting step; a creation step of creating a trained model by machine learning using the process conditions input in the input step and the culture prediction result as training data; A calculation step of calculating an optimal process using the trained model, The cell culture process exploration method, wherein the trained model is a regression model.

2. The method for exploring a cell culture process according to claim 1 , wherein the process conditions include at least one of a plurality of medium compositions and a plurality of culture conditions.

3. 3. The cell culture process exploration method according to claim 2, wherein the culture conditions are set conditions including at least one of the culture method, the size and type of culture vessel, oxygen addition, supplementation of medium and nutrients, removal of medium containing growth inhibitory by-products, and harvesting of the target product.

4. The cell culture process exploration method according to claim 3 , wherein the culture conditions include design and operation for scaling up from a small-scale process to a large-scale process.

5. The cell culture process exploration method according to claim 1 , wherein the plurality of process conditions generated in the process condition generating step are acquired as a matrix.

6. The cell culture process exploration method according to claim 1 , wherein the process condition generating step selects some of the process condition items and determines numerical values ​​for the selected items.

7. 7. The cell culture process exploration method according to claim 1, wherein the process condition generation step is carried out by at least one of a method of determining the numerical values ​​for the process condition items based on arbitrary numerical values, a method of determining by generating random numbers within a specified range, a method of determining by numerical values ​​obtained by experiment, and a method of determining based on an expansion medium mixing strategy.

8. The cell culture process exploration method according to claim 7 , wherein the determined range is set using an equation that mathematically models the mechanism by which an organism takes up a medium component.

9. The cell culture process exploration method according to claim 8, wherein the mathematically modeled equation is the Michaelis-Menten equation or Fick's law.

10. 10. The cell culture process exploration method according to claim 7, wherein the method using random numbers determines the values ​​for the process condition items using values ​​generated by continuous uniform random numbers, continuous Gaussian random numbers, discrete random numbers, or binomial random numbers.

11. The cell culture process exploration method according to any one of claims 1 to 10, wherein the cells are eukaryotic cells or prokaryotic cells.

12. The method for exploring a cell culture process according to claim 11, wherein the eukaryotic cells are cell lines derived from animals, plants or insects, or primary cultures, or fungi.

13. The cell culture process exploration method according to claim 11, wherein the prokaryotic cells are bacteria including Escherichia coli, Bacillus subtilis, cyanobacteria, or actinomycetes, and archaea including methanogens, extreme halophiles, or hyperthermophiles.

14. A cell culture process searching program that causes a computer to execute the cell culture process searching method according to any one of claims 1 to 13.

15. A non-transitory computer-readable recording medium having the cell culture process exploration program according to claim 14 recorded thereon.

16. a process condition generating unit that generates a plurality of process conditions for culturing cells; a culture result prediction unit that acquires a predicted result of cell culture for each of the plurality of process conditions generated by the process condition generation unit, the culture result prediction unit executing a cell culture simulation method including a modeling approach including a metabolic flux analysis using a genome-scale metabolic model or a flux balance analysis based on the mechanism of the cell to be cultured; an optimized process condition acquisition unit that finds optimal process conditions from the culture prediction result obtained by the culture result prediction unit, and selects optimal medium composition and culture conditions by selecting a predetermined number in descending order of cell proliferation amount or a predetermined number in descending order of antibody bioproduction amount; The optimization process condition acquisition unit The process conditions generated by the process condition generating unit and the culture prediction result obtained by the culture result predicting unit are input, The input process conditions and the culture prediction results are used as learning data to create a trained model by machine learning; An optimal process is calculated using the trained model, The cell culture process exploration device, wherein the trained model is a regression model.

Images