WO2021059577A1 - 情報処理装置、情報処理方法、及び情報処理プログラム - Google Patents
情報処理装置、情報処理方法、及び情報処理プログラム Download PDFInfo
- Publication number
- WO2021059577A1 WO2021059577A1 PCT/JP2020/018808 JP2020018808W WO2021059577A1 WO 2021059577 A1 WO2021059577 A1 WO 2021059577A1 JP 2020018808 W JP2020018808 W JP 2020018808W WO 2021059577 A1 WO2021059577 A1 WO 2021059577A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- quality
- cells
- antibody
- input information
- cell
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B5/00—ICT specially adapted for modelling or simulations in systems biology, e.g. gene-regulatory networks, protein interaction networks or metabolic networks
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
- G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
- G16B40/20—Supervised data analysis
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12M—APPARATUS FOR ENZYMOLOGY OR MICROBIOLOGY; APPARATUS FOR CULTURING MICROORGANISMS FOR PRODUCING BIOMASS, FOR GROWING CELLS OR FOR OBTAINING FERMENTATION OR METABOLIC PRODUCTS, i.e. BIOREACTORS OR FERMENTERS
- C12M41/00—Means for regulation, monitoring, measurement or control, e.g. flow regulation
- C12M41/30—Means for regulation, monitoring, measurement or control, e.g. flow regulation of concentration
- C12M41/32—Means for regulation, monitoring, measurement or control, e.g. flow regulation of concentration of substances in solution
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12M—APPARATUS FOR ENZYMOLOGY OR MICROBIOLOGY; APPARATUS FOR CULTURING MICROORGANISMS FOR PRODUCING BIOMASS, FOR GROWING CELLS OR FOR OBTAINING FERMENTATION OR METABOLIC PRODUCTS, i.e. BIOREACTORS OR FERMENTERS
- C12M41/00—Means for regulation, monitoring, measurement or control, e.g. flow regulation
- C12M41/48—Automatic or computerized control
Definitions
- This disclosure relates to an information processing device, an information processing method, and an information processing program.
- Japanese Unexamined Patent Publication No. 2009-44974 discloses a method for constructing a prediction model for predicting cell quality.
- the cells of each sample are photographed at two or more time points with different culture times to acquire images, and each acquired image is analyzed to obtain cells.
- actual measurement data of the prediction target is prepared for each sample, the generated numerical data is used as an input value, and the prepared actual measurement data is used as a teacher value for fuzzy neural network analysis to show a combination of indicators effective for prediction.
- Perfusion culture is a method for culturing cells used in the production of biopharmacy using antibodies produced from cells.
- Perfusion culture is a culture method in which a culture medium containing cells is continuously filtered and discharged, while a fresh medium containing nutrients is continuously supplied to a culture tank.
- Perfusion culture is also referred to as continuous culture.
- parameters that affect the quality of the antibody such as process conditions in cell culture, medium components, and number of cells, can be changed even during the cell culture process. Therefore, in perfusion culture, it is preferable that the quality of the antibody after a lapse of a predetermined period from that time point can be predicted from the information on the cell culture at a certain time point. Specifically, for example, from the information at a certain time point, if the quality of the antibody after the lapse of a predetermined period is within the permissible range, the perfusion culture is continued as it is, and if it is out of the permissible range, the above parameters are set. Can be adjusted. In this case, perfusion culture can be effectively supported.
- JP-A-2009-44974 the quality of cells is predicted from images obtained by photographing cells at two different time points using a prediction model. That is, the technique described in Japanese Patent Application Laid-Open No. 2009-44974 predicts the quality of cells from the degree of change in the image of cells at two different time points. Therefore, the technique described in JP-A-2009-44974 cannot predict the quality of the antibody from the information at a certain time point, and therefore cannot effectively support the perfusion culture.
- This disclosure has been made in view of the above circumstances, and provides an information processing device, an information processing method, and an information processing program that can effectively support perfusion culture.
- the information processing apparatus of the present disclosure includes an acquisition unit that acquires input information including at least one of a process condition in cell culture, a culture medium component, and the number and diameter of cells, input information, input information, and production from cells. Based on a trained model pre-learned using the quality of the antibody and the quality of the cells, the quality of the antibody and the quality of the cells after a predetermined period of time has elapsed from the time when the acquisition unit acquired the input information. It is equipped with a prediction unit that predicts quality.
- the information processing apparatus of the present disclosure may further include an output unit that outputs a warning when the antibody quality and cell quality predicted by the prediction unit are out of the permissible range.
- the output unit may further output at least one of the information included in the input information in which the quality of the antibody and the quality of the cells are within the permissible range.
- the input of the trained model is acquired at a plurality of time points between the start of cell culture and the elapse of a predetermined period as the period for cell proliferation. It may be the input information of.
- the information processing method of the present disclosure acquires input information including at least one of a process condition in cell culture, a medium component, and the number and diameter of cells, and is produced from the input information, the input information, and the cells. Predict antibody quality and cell quality after a predetermined period of time from the acquisition of input information, based on a pre-learned model using the quality of the antibody and the quality of the cells.
- the processing is performed by the computer.
- the information processing program of the present disclosure acquires input information including at least one of a process condition in cell culture, a medium component, and the number and diameter of cells, and is produced from the input information, the input information, and the cells. Predict antibody quality and cell quality after a predetermined period of time from the acquisition of input information, based on a pre-learned model using the quality of the antibody and the quality of the cells. It is for letting the computer execute the process.
- perfusion culture can be effectively supported.
- the configuration of the cell culture apparatus 100 according to the present embodiment will be described with reference to FIG.
- the cell culture device 100 can be suitably used for cell culture for expressing an antibody in animal cells, for example.
- the cells used for expressing the antibody are not particularly limited, and examples thereof include animal cells, plant cells, eukaryotic cells such as yeast, prokaryotic cells such as Bacillus subtilis, and Escherichia coli.
- Animal cells such as CHO cells, BHK-21 cells and SP2 / 0-Ag14 cells are preferable, and CHO cells are more preferable.
- the antibody to be expressed in animal cells is not particularly limited, and for example, anti-IL-6 receptor antibody, anti-IL-6 antibody, anti-glycican-3 antibody, anti-CD3 antibody, anti-CD20 antibody, anti-GPIIb / IIIa antibody, Examples thereof include anti-TNF antibody, anti-CD25 antibody, anti-EGFR antibody, anti-Her2 / neu antibody, anti-RSV antibody, anti-CD33 antibody, anti-CD52 antibody, anti-IgE antibody, anti-CD11a antibody, anti-VEGF antibody and anti-VLA4 antibody.
- Antibodies include not only monoclonal antibodies derived from animals such as humans, mice, rats, hamsters, rabbits and monkeys, but also artificially modified antibodies such as chimeric antibodies, humanized antibodies and bispecific antibodies.
- the obtained antibody or fragment thereof can be purified uniformly.
- the separation and purification method used for ordinary polypeptides may be used.
- an antibody can be separated and purified by appropriately selecting and combining a chromatography column such as affinity chromatography, a filter, ultrafiltration, salting out, dialysis, SDS polyacrylamide gel electrophoresis and isoelectric focusing.
- a chromatography column such as affinity chromatography, a filter, ultrafiltration, salting out, dialysis, SDS polyacrylamide gel electrophoresis and isoelectric focusing.
- the concentration of the obtained antibody can be measured by measuring the absorbance or by an enzyme-linked immunosorbent assay (ELISA) or the like.
- the cell culture apparatus 100 includes a culture vessel 10 containing a cell suspension containing cells and a filter membrane for subjecting the cell suspension extracted from the culture vessel 10 to a membrane separation treatment. Includes a filter unit 20 having 24.
- the cell culture apparatus 100 further has a flow path 32 as a circulation flow path for returning the component blocked by the filter membrane 24 to the culture vessel 10, and a component that has passed through the filter membrane 24 of the cell suspension to the outside of the filter unit 20. Includes a flow path 33 for discharging into.
- the cell culture apparatus 100 includes a flow path 38 for supplying a fresh medium to the culture vessel 10, and a pump P3 provided in the middle of the flow path 38.
- the culture container 10 is a container for containing a cell suspension containing cells and a medium used for antibody expression. Inside the culture vessel 10, a stirring device 11 having a stirring blade is provided. By rotating the stirring blade of the stirring device 11, the medium contained together with the cells inside the culture vessel 10 is stirred, and the homogeneity of the medium is maintained.
- the cell culture apparatus 100 in order to prevent the concentration of cells in the culture vessel 10 from becoming excessively high, a part of the cells in the culture vessel 10 (for example, about 10%) at an appropriate timing within the culture period.
- the cell bleeding process is performed.
- the cells in the culture vessel 10 are discharged to the outside of the culture vessel 10 via the flow path 39.
- One end of the flow path 31 is connected to the bottom of the culture vessel 10, and the other end is connected to the inflow port 20a of the filter unit 20.
- a pump P1 is provided in the middle of the flow path 31 to extract the cell suspension contained in the culture vessel 10 and send it to the filter unit 20.
- the filter unit 20 is a filter membrane 24 that divides the container 21 and the space inside the container 21 into a supply side 22 and a permeation side 23, and performs a membrane separation treatment on the cell suspension extracted from the culture vessel 10. , Equipped with. Further, the filter unit 20 has an inflow port 20a into which the cell suspension flows in and an outflow port 20b from which the cell suspension flows out on the supply side 22. The cell suspension extracted from the culture container 10 passes through the filter membrane 24 while flowing into the inside of the container 21 from the inflow port 20a and flowing out from the outflow port 20b to the outside of the container 21.
- the filter unit 20 sends a permeation component to the permeation side 23 while flowing a liquid to be membrane-separated along the membrane surface of the filter membrane 24 (that is, in a direction parallel to the membrane surface) (cross).
- Membrane separation processing is performed by the flow) method.
- the tangential flow method which is a method of membrane separation treatment using the filter membrane 24, forms a flow in which the cell suspension extracted from the culture vessel 10 circulates in one direction in parallel along the membrane surface of the filter membrane 24.
- the cell suspension may be one that forms a flow in which the cell suspension reciprocates alternately in parallel along the membrane surface of the filter membrane 24.
- a KrosFlo perfusion culture flow path device (KML-100, KPS-200, KPS-600) manufactured by Spectrum Laboratories can be preferably used. Further, when forming a flow that reciprocates alternately, the ATF system manufactured by REPLIGEN can be preferably used.
- the relatively large-sized component contained in the cell suspension does not permeate through the filter membrane 24, flows out from the outlet 20b to the outside of the container 21, and is returned to the inside of the culture container 10 via the flow path 32. That is, of the cell suspension extracted from the culture vessel 10, the component blocked by the filter membrane 24 is returned to the inside of the culture vessel 10 via the flow path 32.
- the relatively small-sized component contained in the cell suspension permeates through the filter membrane 24 and is discharged to the outside of the container 21 from the discharge port 20c provided on the permeation side 23.
- a flow path 33 provided with a pump P2 is connected to the discharge port 20c of the filter unit 20, and the components discharged to the permeation side 23 are discharged from the discharge port 20c to the outside of the container 21 via the flow path 33. It is discharged.
- the filter membrane 24 is used for the purpose of separating cells and components unnecessary for cell culture.
- components unnecessary for cell culture include cell carcasses, cell crushed products, DNA, HCP, antibodies, waste products, and the like. That is, the filter membrane 24 has a separation performance suitable for blocking the permeation of cells while allowing components unnecessary for cell culture to permeate.
- the components unnecessary for cell culture discharged from the culture vessel 10 as described above are sent to the antibody purification step, which is the next step.
- the information processing device 40 includes a CPU (Central Processing Unit) 41, a memory 42, a storage unit 43, a display unit 44 such as a liquid crystal display, an input unit 45 such as a keyboard and a mouse, and an external I / F. (InterFace) 46 is included.
- the CPU 41, the memory 42, the storage unit 43, the display unit 44, the input unit 45, and the external I / F 46 are connected to the bus 47.
- the measuring unit 48 is connected to the external I / F 46.
- Examples of the information processing device 40 include a personal computer, a server computer, and the like.
- the storage unit 43 is realized by an HDD (Hard Disk Drive), an SSD (Solid State Drive), a flash memory, or the like.
- the learning program 50 and the information processing program 52 are stored in the storage unit 43 as a storage medium.
- the CPU 41 reads the learning program 50 from the storage unit 43, expands it into the memory 42, and executes the expanded learning program 50. Further, the CPU 41 reads the information processing program 52 from the storage unit 43, expands it into the memory 42, and executes the expanded information processing program 52. Further, the learning data 54 and the learned model 56 are stored in the storage unit 43.
- the measuring unit 48 includes various measuring devices for measuring the process conditions in cell culture using the cell culture device 100, the medium components, and the number and diameter of cells.
- process conditions include the rotation speed of the stirring device 11 per unit time (hereinafter referred to as “stirring rotation speed”), the aeration rate per unit volume of the medium contained in the culture vessel 10, and the culture medium 10.
- the temperature of the medium to be cultivated can be mentioned.
- the medium components include the amount of nutrient components, the amount of metabolic components, the amount of dissolved gas (for example, the amount of dissolved oxygen) and the like of the medium contained in the culture vessel 10.
- the learning data 54 is produced from the process conditions in cell culture, which are explanatory variables, the medium components, the number and diameter of cells, and the cells, which are objective variables corresponding to the explanatory variables.
- It is a data set for learning that includes a plurality of sets of antibody quality and cell quality.
- the number of cells means the sum of the number of living cells and the number of dead cells.
- antibody quality include antibody concentration, antibody agglutination amount, antibody degradation product amount, immature sugar chain amount, and the like.
- cell quality include cell viability and cell viability.
- the quality of the antibody and the quality of the cells may be one of these index values or a combination of a plurality of these index values. Further, the quality of the antibody and the quality of the cells may be evaluation values obtained by determining one or a plurality of combinations thereof in a plurality of stages (for example, four stages A to D) according to a predetermined determination standard.
- each data contained in the learning data 54 includes process conditions measured regularly (for example, once a day), medium components, and the number of cells in the past cell culture. And diameter.
- Each data contained in the learning data 54 further indicates the quality of the antibody after a predetermined period (for example, 2 days) has elapsed from the measurement time of the process conditions, the medium components, and the number and diameter of cells. And cell quality.
- the predetermined period is not limited to two days, and may be, for example, one day, three or more days, or a period other than one day (for example, 12 hours).
- the trained model 56 is a model that has been trained in advance using the training data 54.
- An example of the trained model 56 is a neural network model.
- the trained model 56 is generated by the information processing device 40 in the learning phase described later.
- the information processing apparatus 40 includes an acquisition unit 60 and a learning unit 62.
- the CPU 41 executes the learning program 50, it functions as the acquisition unit 60 and the learning unit 62.
- the acquisition unit 60 acquires the learning data 54 from the storage unit 43.
- the learning unit 62 generates the learned model 56 by training the model using the learning data 54 acquired by the acquisition unit 60 as teacher data. Then, the learning unit 62 stores the generated learned model 56 in the storage unit 43.
- the trained model 56 in which the process conditions, the medium components, the number and diameter of cells are input, and the quality of the antibody and the quality of cells are output is obtained. Generated.
- an error backpropagation method is used for learning by the learning unit 62.
- the trained model 56 may be a deep neural network model having a plurality of intermediate layers.
- step S10 of FIG. 7 the acquisition unit 60 acquires the learning data 54 from the storage unit 43.
- step S12 as described above, the learning unit 62 generates the trained model 56 by training the model using the learning data 54 acquired in step S10 as the teacher data. Then, the learning unit 62 stores the generated learned model 56 in the storage unit 43.
- step S12 ends, the learning process ends.
- the information processing apparatus 40 includes an acquisition unit 70, a prediction unit 72, a determination unit 74, a derivation unit 76, and an output unit 78.
- the CPU 41 executes the information processing program 52, it functions as an acquisition unit 70, a prediction unit 72, a determination unit 74, a derivation unit 76, and an output unit 78.
- the acquisition unit 70 includes the process conditions in cell culture using the cell culture device 100, the medium components, and the medium components measured by the measurement unit 48 at a predetermined periodic timing (for example, once a day). Obtain the number and diameter of cells.
- This process condition, the medium component, and the number and diameter of cells are examples of input information input to the trained model 56.
- the prediction unit 72 has elapsed a predetermined period (for example, 2 days) from the time when the acquisition unit 70 acquires the input information based on the input information acquired by the acquisition unit 70 and the trained model 56. Predict later antibody quality and cell quality. Specifically, the prediction unit 72 inputs the input information acquired by the acquisition unit 70 into the trained model 56. As described above, the trained model 56 inputs the process conditions, the medium components, the number and diameter of cells, and outputs the quality of the antibody and the quality of the cells after a predetermined period of time has elapsed. It is a learned model.
- a predetermined period for example, 2 days
- the output from the trained model 56 becomes a predicted value of the antibody quality and the cell quality after a predetermined period has elapsed from the time when the input information is acquired by the acquisition unit 70.
- the prediction unit 72 predicts the quality of the antibody and the quality of the cells after a predetermined period of time has elapsed from the time when the input information is acquired by the acquisition unit 70.
- the determination unit 74 determines whether the antibody quality and cell quality predicted by the prediction unit 72 are within the permissible range or out of the permissible range.
- This permissible range is experimentally determined in advance according to, for example, the cell type and the index value used as the quality of the antibody. For example, when the amount of antibody aggregates is used as the quality of the antibody, the determination unit 74 determines that the quality of the antibody is out of the permissible range when the amount of antibody aggregates is equal to or greater than the threshold value, and the amount of antibody aggregates is high. If it is below the threshold, it is determined that the quality of the antibody is within the permissible range.
- the derivation unit 76 is one of the information included in the input information in which the antibody quality and the cell quality are within the permissible range. Derivation of two pieces of information. Specifically, the derivation unit 76 derives the contribution of each explanatory variable (that is, each input) of the trained model 56 to the objective variable (that is, the output), and selects the explanatory variable having the highest contribution as an antibody. It is derived as one piece of information in which the quality of the cell and the quality of the cell are within the permissible range.
- the derivation process of the degree of contribution by the derivation unit 76 will be specifically described.
- the input layer of the trained model 56 has three nodes, node 1 to node 3, the intermediate layer has two nodes, node A and node B, and the output layer.
- the output layer Explains an example in which has one node O.
- the derivation unit 76 derives the inner product of the weights as the contribution degree for each input node.
- the derivation unit 76 derives the contribution of the node 1 according to the following equation (1).
- W 1A in the equation (1) is a weight from node 1 to node A
- WAO is a weight from node A to node O.
- W 1B in the equation (1) is a weight from node 1 to node B
- W BO is a weight from node B to node O. It is possible to derive the contribution degree for the node 2 and the node 3 in the same manner.
- Contribution of node 1 W 1A x W AO + W 1B x W BO ... (1)
- the derivation unit 76 also derives change information indicating how the derived explanatory variable having the highest contribution is changed according to the sign of the contribution.
- the explanatory variable having the highest contribution is the stirring rotation speed of the stirring device 11 included in the process conditions and the quality of the antibody is the amount of antibody aggregates.
- the derivation unit 76 derives the information indicating that the stirring rotation speed is reduced as the above change information.
- the sign of the contribution degree is a negative sign
- the derivation unit 76 derives the information indicating that the stirring rotation speed is increased as the above change information.
- the derivation unit 76 derives, for example, a plurality of explanatory variables in descending order of contribution, instead of one explanatory variable having the highest contribution, as information that the antibody quality and cell quality are within the permissible range.
- one or more explanatory variables whose contribution is equal to or greater than the threshold value may be derived.
- the output unit 78 outputs a warning to the display unit 44 when the determination unit 74 determines that the quality of the antibody and the quality of the cells are out of the permissible range. Further, in this case, the output unit 78 further outputs the information derived by the out-licensing unit 76 to the display unit 44. With these outputs, the warning screen shown in FIG. 10 is displayed on the display unit 44 as an example. As shown in FIG. 10, on the warning screen according to the present embodiment, a warning message indicating that the quality of the antibody and the quality of the cells are out of the permissible range is displayed after a lapse of a predetermined period. Further, on the warning screen according to the present embodiment, explanatory variables in which the quality of the antibody and the quality of the cells are within the permissible range, and change information indicating how to change the explanatory variables are also displayed.
- the output unit 78 When the explanatory variable derived by the derivation unit 76 is a parameter that can be controlled by the information processing device 40, the output unit 78 outputs the change information derived by the derivation unit 76 to the control target. May be changed. Specifically, for example, when the derivation unit 76 derives information indicating that the stirring rotation speed is reduced as change information, the output unit 78 outputs the change information to the motor that controls the stirring rotation speed. As a result, the stirring rotation speed is reduced by a predetermined rotation speed.
- the operation of the information processing apparatus 40 according to the present embodiment in the operation phase will be described.
- the CPU 41 executes the information processing program 52
- the quality prediction process shown in FIG. 11 is executed.
- the quality prediction process shown in FIG. 11 is executed at a regular timing such as once a day after the start of perfusion culture.
- step S20 of FIG. 11 the acquisition unit 70 acquires the input information measured by the measurement unit 48.
- step S22 as described above, the prediction unit 72 inputs the input information acquired in step S20 into the trained model 56 to obtain a predetermined period from the time when the acquisition unit 70 acquires the input information. Predict antibody quality and cell quality after lapse.
- step S24 the determination unit 74 determines whether or not the antibody quality and cell quality predicted in step S22 are out of the permissible range. If this determination is affirmative, the process proceeds to step S26.
- step S26 the derivation unit 76 derives the explanatory variable having the highest contribution as the information included in the input information in which the quality of the antibody and the quality of the cells are within the permissible range. Further, as described above, the derivation unit 76 also derives change information indicating how the derived explanatory variable having the highest contribution is changed according to the sign of the contribution.
- the trained model 56 may be retrained according to the number of batches. In this case, the weights between the nodes of the trained model 56 are also updated. Therefore, in the present embodiment, the process of step S26 is executed every time the quality prediction process is executed.
- step S28 the output unit 78 outputs the warning to the display unit 44 and outputs the information derived in step S26 to the display unit 44 as described above.
- the warning screen shown in FIG. 10 is displayed on the display unit 44 as an example.
- the user confirms the warning screen displayed on the display unit 44 by the process of step S28, and adjusts the parameters such as the process conditions and the medium components that affect the quality of the antibody and the quality of the cells.
- the process of step S28 is completed, the quality prediction process is completed. Also, when the determination in step S24 is a negative determination, the quality prediction process ends.
- the quality of the antibody and the quality of the cells are predicted after a predetermined period has elapsed from the time when the acquisition unit 70 acquired the input information. Therefore, the user can grasp the quality of the antibody and the quality of the cells after the lapse of a predetermined period, and can adjust the parameters that affect the quality of the antibody and the quality of the cells in the perfusion culture. Therefore, perfusion culture can be effectively supported.
- the optimum culture conditions are determined by a small-scale test in which perfusion culture is performed under various process conditions and culture conditions such as medium components on a relatively small scale.
- the quality of the antibody and cells is confirmed by a medium-volume trial in which perfusion culture is performed under the culture conditions determined by the small-scale test on a relatively medium scale.
- the antibody is produced by perfusion culture on a relatively large scale.
- the small amount test is performed for 30 days, for example. In order to effectively support the perfusion culture, it is preferable that the period of this small amount test can be shortened.
- the quality of the antibody improves with the passage of time.
- the quality of the antibody may be stable or decreased depending on the process conditions in the perfusion culture and the culture conditions such as the medium components (the conditions shown by the solid line in the example of FIG. 13) (FIG. In the example of 13, the conditions indicated by the one-dot chain line and the two-dot chain line) are applied. That is, the final quality of the antibody, such as after 30 days from the start of perfusion culture, varies depending on the conditions. Therefore, in the information processing apparatus 40 according to the present embodiment, the period of the small amount test is shortened by predicting the final antibody quality and the cell quality from the measurement data at a plurality of time points in the cell growth phase.
- the learning data 54 is produced from the process conditions in cell culture, which are explanatory variables, the medium components, the number and diameter of cells, and the cells, which are objective variables corresponding to the explanatory variables. It is a data set for learning that includes a plurality of sets of antibody quality and cell quality.
- a plurality of time points that is, within the cell growth period
- a predetermined period n as the period for cell growth (that is, within the cell growth period).
- the learning data 54 includes the process conditions (every day in the example of FIG. 15), the medium components, and the number and diameter of cells. Further, in the present embodiment, the process conditions, the medium components, and the number and diameter of cells, which are associated with the process conditions, the medium components, and the number and diameter of cells acquired at a plurality of time points, are finally set.
- the learning data 54 also includes the quality of the antibody and the quality of the cells after a predetermined period m has elapsed since the acquisition.
- n in FIGS. 14 and 15 is 10 and m is 20.
- n and m are not limited to this example, and for example, n is 5 and m is 9, that is, the process conditions at a plurality of time points for 5 days after the start of cell culture, the medium components, and the cells.
- the number and diameter may be associated with the quality of the antibody and the quality of the cells 14 days after the start of cell culture.
- the information processing apparatus 40 includes an acquisition unit 80 and a learning unit 82.
- the CPU 41 executes the learning program 50, it functions as the acquisition unit 80 and the learning unit 82.
- the acquisition unit 80 acquires the learning data 54 from the storage unit 43.
- the learning unit 82 generates the learned model 56 by training the model using the learning data 54 acquired by the acquisition unit 80 as teacher data. Then, the learning unit 82 stores the generated learned model 56 in the storage unit 43.
- Model 56 is generated.
- the trained model 56 may be a deep neural network model having a plurality of intermediate layers. Further, as the trained model 56, a model other than the neural network may be applied.
- step S40 the acquisition unit 80 acquires the learning data 54 from the storage unit 43.
- step S42 as described above, the learning unit 82 generates the trained model 56 by training the model using the learning data 54 acquired in step S40 as the teacher data. Then, the learning unit 82 stores the generated learned model 56 in the storage unit 43.
- step S42 ends, the learning process ends.
- the information processing apparatus 40 includes an acquisition unit 90, a prediction unit 92, and an output unit 94.
- the CPU 41 executes the information processing program 52, it functions as an acquisition unit 90, a prediction unit 92, and an output unit 94.
- the acquisition unit 90 used the cell culture device 100 measured by the measurement unit 48 at a plurality of time points from the start of cell culture to the elapse of a predetermined period n as the period for cell growth. Obtain a plurality of sets of process conditions in cell culture, medium components, and the number and diameter of cells. This process condition, the medium component, and the number and diameter of cells are examples of input information input to the trained model 56.
- the prediction unit 92 has a predetermined period from the time when the acquisition unit 90 last acquires the input information based on the plurality of sets of input information acquired by the acquisition unit 90 at a plurality of time points and the trained model 56. Predict antibody quality and cell quality after m. Specifically, the prediction unit 92 inputs a plurality of sets of input information acquired by the acquisition unit 90 into the trained model 56. As described above, the trained model 56 uses a plurality of sets of process conditions, medium components, and the number and diameter of cells as inputs, and the quality of the antibody and the quality of the cells after a predetermined period m has elapsed. It is a model learned with quality as output.
- the output from the trained model 56 becomes a predicted value of the antibody quality and the cell quality after a predetermined period m has elapsed from the time when the acquisition unit 90 last acquired the input information.
- the prediction unit 92 predicts the quality of the antibody and the quality of the cells after a predetermined period m has elapsed from the time when the acquisition unit 90 last acquired the input information.
- the output unit 94 displays the antibody quality and cell quality predicted by the prediction unit 92 by outputting them to the display unit 44. This allows the user to know the final antibody quality and cell quality at the end of the cell growth phase. Therefore, the user can perform a medium-dose trial by selecting the condition that is expected to have the highest antibody quality and cell quality among various conditions in the low-dose test at an early stage.
- the operation of the information processing apparatus 40 according to the present embodiment in the operation phase will be described.
- the CPU 41 executes the information processing program 52
- the quality prediction process shown in FIG. 20 is executed.
- the quality prediction process shown in FIG. 20 is performed at the start of perfusion culture.
- step S50 of FIG. 20 the acquisition unit 90 acquires the input information measured by the measurement unit 48.
- step S52 the acquisition unit 90 determines whether or not a predetermined period n has elapsed as a period for cell proliferation after the quality prediction process is started (that is, after the cell culture is started). If this determination is a negative determination, the process returns to step S50, and if the determination is affirmative, the process proceeds to step S54. Therefore, by repeatedly executing step S50 periodically until a predetermined period n elapses as the period for cell proliferation, the acquisition unit 90 acquires a plurality of sets of input information at a plurality of time points. To.
- step S54 the prediction unit 92 finally obtains the input information based on the plurality of sets of input information acquired in step S50 and the trained model 56 at a plurality of time points.
- the quality of the antibody and the quality of the cells are predicted after a predetermined period m has elapsed from the time of acquisition.
- step S56 the output unit 94 displays the antibody quality and cell quality predicted in step S54 by outputting to the display unit 44.
- the quality prediction process is completed.
- the information processing apparatus 40 uses the trained model according to the first embodiment in the period after the cell proliferation phase (for example, the stable phase of FIG. 13) as in the first embodiment. 56 may be used to monitor or adjust parameters for antibody quality and cell quality within acceptable limits.
- the various processors include CPUs, which are general-purpose processors that execute software (programs) and function as various processing units, as well as circuits after manufacturing FPGAs (Field Programmable Gate Arrays) and the like.
- Dedicated electricity which is a processor with a circuit configuration specially designed to execute specific processing such as programmable logic device (PLD), ASIC (Application Specific Integrated Circuit), which is a processor whose configuration can be changed. Circuits and the like are included.
- One processing unit may be composed of one of these various processors, or a combination of two or more processors of the same type or different types (for example, a combination of a plurality of FPGAs or a combination of a CPU and an FPGA). It may be composed of a combination). Further, a plurality of processing units may be configured by one processor.
- one processor is configured by a combination of one or more CPUs and software, as represented by a computer such as a client and a server.
- the processor functions as a plurality of processing units.
- SoC System On Chip
- the various processing units are configured by using one or more of the above-mentioned various processors as a hardware structure.
- an electric circuit in which circuit elements such as semiconductor elements are combined can be used.
- the mode in which the learning program 50 and the information processing program 52 are stored (installed) in the storage unit 43 in advance has been described, but the present invention is not limited to this.
- the learning program 50 and the information processing program 52 are recorded on a recording medium such as a CD-ROM (Compact Disc Read Only Memory), a DVD-ROM (Digital Versatile Disc Read Only Memory), and a USB (Universal Serial Bus) memory. May be provided at. Further, the learning program 50 and the information processing program 52 may be downloaded from an external device via a network.
- the processor With a memory built in or connected to the processor, The processor Obtain input information including at least one of process conditions in cell culture, medium components, and number and diameter of cells. Based on the input information and a learned model pre-learned using the input information, the quality of the antibody produced from the cell, and the quality of the cell, the input information is predetermined from the time when the input information is acquired. An information processing device that predicts the quality of the antibody and the quality of the cells after the lapse of a period.
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Chemical & Material Sciences (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- General Health & Medical Sciences (AREA)
- Biotechnology (AREA)
- Data Mining & Analysis (AREA)
- Organic Chemistry (AREA)
- Zoology (AREA)
- Wood Science & Technology (AREA)
- Biophysics (AREA)
- Theoretical Computer Science (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Analytical Chemistry (AREA)
- Software Systems (AREA)
- Public Health (AREA)
- Bioethics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Epidemiology (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Biochemistry (AREA)
- Sustainable Development (AREA)
- Microbiology (AREA)
- General Engineering & Computer Science (AREA)
- Genetics & Genomics (AREA)
- Molecular Biology (AREA)
- Physiology (AREA)
- Computer Hardware Design (AREA)
- Apparatus Associated With Microorganisms And Enzymes (AREA)
Abstract
Description
図1を参照して、本実施形態に係る細胞培養装置100の構成を説明する。細胞培養装置100は、例えば、動物細胞において、抗体を発現させるための細胞培養に好適に用いることができる。
ノード1の寄与度=W1A×WAO+W1B×WBO・・・(1)
開示の技術の第2実施形態を説明する。なお、本実施形態に係る細胞培養装置100の構成は、第1実施形態と同一であるため、説明を省略する。本実施形態に係る情報処理装置40のハードウェア構成は、記憶部43に記憶される学習用データ54及び学習済みモデル56以外は第1実施形態と同一であるため、本実施形態では、学習用データ54及び学習済みモデル56について説明する。
プロセッサと、
前記プロセッサに内蔵又は接続されたメモリと、を備え、
前記プロセッサは、
細胞培養におけるプロセス条件と、培地成分と、細胞の数及び径との少なくとも1つを含む入力情報を取得し、
前記入力情報と、前記入力情報、前記細胞から産生される抗体の品質及び前記細胞の品質を用いて予め学習された学習済みモデルとに基づいて、前記入力情報を取得した時点から予め定められた期間を経過した後の前記抗体の品質及び前記細胞の品質を予測する
情報処理装置。
Claims (6)
- 細胞培養におけるプロセス条件と、培地成分と、細胞の数及び径との少なくとも1つを含む入力情報を取得する取得部と、
前記入力情報と、前記入力情報、前記細胞から産生される抗体の品質及び前記細胞の品質を用いて予め学習された学習済みモデルとに基づいて、前記入力情報を前記取得部が取得した時点から予め定められた期間を経過した後の前記抗体の品質及び前記細胞の品質を予測する予測部と、
を備えた情報処理装置。 - 前記予測部により予測された前記抗体の品質及び前記細胞の品質が許容範囲外である場合、警告を出力する出力部
を更に備えた請求項1に記載の情報処理装置。 - 前記出力部は、更に、前記抗体の品質及び前記細胞の品質が前記許容範囲内となる前記入力情報に含まれる情報の少なくとも1つを出力する
請求項2に記載の情報処理装置。 - 前記学習済みモデルの入力は、細胞培養を開始してから細胞が増殖する期間として予め定められた期間が経過するまでの間の複数の時点で取得された複数の前記入力情報である
請求項1から請求項3の何れか1項に記載の情報処理装置。 - 細胞培養におけるプロセス条件と、培地成分と、細胞の数及び径との少なくとも1つを含む入力情報を取得し、
前記入力情報と、前記入力情報、前記細胞から産生される抗体の品質及び前記細胞の品質を用いて予め学習された学習済みモデルとに基づいて、前記入力情報を取得した時点から予め定められた期間を経過した後の前記抗体の品質及び前記細胞の品質を予測する
処理をコンピュータが実行する情報処理方法。 - 細胞培養におけるプロセス条件と、培地成分と、細胞の数及び径との少なくとも1つを含む入力情報を取得し、
前記入力情報と、前記入力情報、前記細胞から産生される抗体の品質及び前記細胞の品質を用いて予め学習された学習済みモデルとに基づいて、前記入力情報を取得した時点から予め定められた期間を経過した後の前記抗体の品質及び前記細胞の品質を予測する
処理をコンピュータに実行させるための情報処理プログラム。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP20867457.2A EP4015616A4 (en) | 2019-09-24 | 2020-05-11 | INFORMATION PROCESSING DEVICE, INFORMATION PROCESSING METHOD AND INFORMATION PROCESSING PROGRAM |
JP2021548321A JPWO2021059577A1 (ja) | 2019-09-24 | 2020-05-11 | |
US17/690,389 US20220199195A1 (en) | 2019-09-24 | 2022-03-09 | Information processing apparatus, information processing method, and information processing program |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2019173364 | 2019-09-24 | ||
JP2019-173364 | 2019-09-24 |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US17/690,389 Continuation US20220199195A1 (en) | 2019-09-24 | 2022-03-09 | Information processing apparatus, information processing method, and information processing program |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2021059577A1 true WO2021059577A1 (ja) | 2021-04-01 |
Family
ID=75166912
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2020/018808 WO2021059577A1 (ja) | 2019-09-24 | 2020-05-11 | 情報処理装置、情報処理方法、及び情報処理プログラム |
Country Status (4)
Country | Link |
---|---|
US (1) | US20220199195A1 (ja) |
EP (1) | EP4015616A4 (ja) |
JP (1) | JPWO2021059577A1 (ja) |
WO (1) | WO2021059577A1 (ja) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2022248935A1 (en) * | 2021-05-27 | 2022-12-01 | Lynceus Sas | Machine learning-based quality control of a culture for bioproduction |
US11567488B2 (en) | 2021-05-27 | 2023-01-31 | Lynceus, Sas | Machine learning-based quality control of a culture for bioproduction |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009044974A (ja) | 2007-08-16 | 2009-03-05 | Univ Nagoya | 細胞の品質を予測する予測モデルの構築法、予測モデルの構築用ブログラム、該プログラムを記録した記録媒体、予測モデルの構築用装置 |
JP2018117567A (ja) * | 2017-01-25 | 2018-08-02 | 株式会社日立製作所 | 細胞培養装置 |
WO2018142702A1 (ja) * | 2017-01-31 | 2018-08-09 | 株式会社ニコン | 培養支援装置、観察装置、及びプログラム |
WO2019181234A1 (ja) * | 2018-03-19 | 2019-09-26 | 富士フイルム株式会社 | 生産物の製造方法 |
JP2019173364A (ja) | 2018-03-28 | 2019-10-10 | 鹿島建設株式会社 | シールド掘進機及び巨礫の処理方法 |
WO2020021860A1 (ja) * | 2018-07-24 | 2020-01-30 | 富士フイルム株式会社 | 細胞培養支援装置の作動プログラム、細胞培養支援装置、細胞培養支援装置の作動方法 |
WO2020039683A1 (ja) * | 2018-08-22 | 2020-02-27 | 富士フイルム株式会社 | 細胞培養支援装置の作動プログラム、細胞培養支援装置、細胞培養支援装置の作動方法 |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6801000B2 (ja) * | 2016-12-01 | 2020-12-16 | 富士フイルム株式会社 | 細胞画像評価装置および細胞画像評価制御プログラム |
EP3732485A1 (de) * | 2017-12-29 | 2020-11-04 | F. Hoffmann-La Roche AG | Vorhersage des metabolischen zustands einer zellkultur |
-
2020
- 2020-05-11 JP JP2021548321A patent/JPWO2021059577A1/ja active Pending
- 2020-05-11 WO PCT/JP2020/018808 patent/WO2021059577A1/ja active Application Filing
- 2020-05-11 EP EP20867457.2A patent/EP4015616A4/en active Pending
-
2022
- 2022-03-09 US US17/690,389 patent/US20220199195A1/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009044974A (ja) | 2007-08-16 | 2009-03-05 | Univ Nagoya | 細胞の品質を予測する予測モデルの構築法、予測モデルの構築用ブログラム、該プログラムを記録した記録媒体、予測モデルの構築用装置 |
JP2018117567A (ja) * | 2017-01-25 | 2018-08-02 | 株式会社日立製作所 | 細胞培養装置 |
WO2018142702A1 (ja) * | 2017-01-31 | 2018-08-09 | 株式会社ニコン | 培養支援装置、観察装置、及びプログラム |
WO2019181234A1 (ja) * | 2018-03-19 | 2019-09-26 | 富士フイルム株式会社 | 生産物の製造方法 |
JP2019173364A (ja) | 2018-03-28 | 2019-10-10 | 鹿島建設株式会社 | シールド掘進機及び巨礫の処理方法 |
WO2020021860A1 (ja) * | 2018-07-24 | 2020-01-30 | 富士フイルム株式会社 | 細胞培養支援装置の作動プログラム、細胞培養支援装置、細胞培養支援装置の作動方法 |
WO2020039683A1 (ja) * | 2018-08-22 | 2020-02-27 | 富士フイルム株式会社 | 細胞培養支援装置の作動プログラム、細胞培養支援装置、細胞培養支援装置の作動方法 |
Non-Patent Citations (1)
Title |
---|
See also references of EP4015616A4 |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2022248935A1 (en) * | 2021-05-27 | 2022-12-01 | Lynceus Sas | Machine learning-based quality control of a culture for bioproduction |
US11567488B2 (en) | 2021-05-27 | 2023-01-31 | Lynceus, Sas | Machine learning-based quality control of a culture for bioproduction |
Also Published As
Publication number | Publication date |
---|---|
US20220199195A1 (en) | 2022-06-23 |
EP4015616A4 (en) | 2022-10-12 |
EP4015616A1 (en) | 2022-06-22 |
JPWO2021059577A1 (ja) | 2021-04-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20220199195A1 (en) | Information processing apparatus, information processing method, and information processing program | |
Khanal et al. | Developments and opportunities in continuous biopharmaceutical manufacturing | |
Chopda et al. | Recent advances in integrated process analytical techniques, modeling, and control strategies to enable continuous biomanufacturing of monoclonal antibodies | |
Yongky et al. | Process intensification in fed-batch production bioreactors using non-perfusion seed cultures | |
Bielser et al. | Semi‐continuous scale‐down models for clone and operating parameter screening in perfusion bioreactors | |
CN112662551B (zh) | 一种细胞培养控制方法以及系统 | |
WO2022042510A1 (zh) | 蛋白表达量的预测方法、装置、计算机设备和存储介质 | |
US20220251498A1 (en) | Facilities and processes to produce biotherapeutics | |
Stewart-Ornstein et al. | Model-guided optogenetic study of PKA signaling in budding yeast | |
CN109411016A (zh) | 基因变异位点检测方法、装置、设备及存储介质 | |
JP2022544928A (ja) | 細胞培養プロセスにおけるプロセス変数を測定するための方法 | |
Thakur et al. | Process analytical technology (PAT) implementation for membrane operations in continuous manufacturing of mAbs: Model-based control of single-pass tangential flow ultrafiltration | |
WO2021059578A1 (ja) | 情報処理装置、情報処理方法、及び情報処理プログラム | |
Allenby et al. | Image analyses for engineering advanced tissue biomanufacturing processes | |
Dürauer et al. | Sensors and chemometrics in downstream processing | |
Banerjee et al. | Multi-objective optimization of continuous CHO cell clarification using acoustic wave separation | |
Khuat et al. | Applications of machine learning in biopharmaceutical process development and manufacturing: Current trends, challenges, and opportunities | |
Whitford | Single‐use systems support continuous bioprocessing by perfusion culture | |
PERRINO | PH. D. THESIS IN | |
JP2023522695A (ja) | 治療用タンパク質を製造するためのクロマトグラフィーパラメータの選択 | |
CN113382793A (zh) | 实时影响顺序层析 | |
WO2023090015A1 (ja) | 情報処理装置、情報処理装置の作動方法、情報処理装置の作動プログラム、校正済み状態予測モデルの生成方法、並びに校正済み状態予測モデル | |
US20240067918A1 (en) | Apparatus for determining cell culturing condition using artificial intelligence and operation method thereof | |
Andersson et al. | Adaptable control policies for variable liquid chromatography columns using deep reinforcement learning | |
Narayanan et al. | Consistent value creation from bioprocess data with customized algorithms: Opportunities beyond multivariate analysis |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 20867457 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2021548321 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 20867457.2 Country of ref document: EP |
|
ENP | Entry into the national phase |
Ref document number: 2020867457 Country of ref document: EP Effective date: 20220318 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |