A method, device, equipment and medium for optimizing culture medium mixing ratio

Abstract

The present invention provides a method, device, equipment and medium for optimizing the mixing ratio of culture medium, the method comprising: taking the first input variable corresponding to each culture medium mother liquid as input data, taking the culture effect of the culture medium corresponding to each culture medium mother liquid as output data, training a preset machine learning model to obtain a trained machine learning model; optimizing the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother liquid; and obtaining a target mixing ratio corresponding to the optimal culture effect of each culture medium mother liquid according to the second input variable corresponding to each culture medium mother liquid. By adopting the embodiment of the present invention, one or more optimal mixing ratios can be selected by performing data dimensionality reduction processing on the data input to the model and developing the mixing ratio based on machine learning technology, thereby further improving the accuracy of the mixing test.

Description

Technical Field

[0001] The present invention relates to the field of biotechnology, and in particular to a method, device, equipment and medium for optimizing the mixing ratio of a culture medium. Background Art

[0002] Culture medium refers to a nutrient matrix composed of different nutrients that provide growth and reproduction for microorganisms, plants or animals. It generally contains several major types of substances such as carbohydrates, nitrogen-containing substances, inorganic salts, vitamins and water. Culture medium is not only the basic substance for providing cell nutrition and promoting cell proliferation, but also the living environment for cell growth and reproduction.

[0003] However, the inventors of the present invention have discovered in their research on the prior art that, in the process of culture medium mixing and ratio optimization, the current traditional technology can only select a small amount of culture medium and analyze it according to a specific ratio. How to select mixed culture medium in large-scale culture and how to set the ratio of each mixed culture medium are the difficulties in optimizing the traditional culture medium formula. Summary of the invention

[0004] The present invention provides a method, device, equipment and medium for optimizing the mixing ratio of a culture medium, which can provide one or more optimal mixing ratios of the culture medium, thereby further improving the accuracy of the mixing test.

[0005] To achieve the above objectives, in a first aspect, an embodiment of the present invention provides a method for optimizing a culture medium mixing ratio, comprising the following steps:

[0006] Taking the first input variable corresponding to each culture medium mother solution as input data and the culture effect of the culture medium corresponding to each culture medium mother solution as output data, the preset machine learning model is trained to obtain a trained machine learning model; wherein the first input variable corresponding to each culture medium mother solution is obtained by converting according to the initial mixing ratio of each culture medium mother solution;

[0007] Optimizing the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each of the culture medium mother solutions;

[0008] According to the second input variable corresponding to each of the culture medium mother solutions, a target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions is obtained.

[0009] As an optional embodiment, the initial mixing ratio of each culture medium mother solution is converted into the first input variable corresponding to each culture medium mother solution by the following formula:

[0010]

[0011] Among them, x i is the first input variable corresponding to each of the culture medium mother solutions, i is the corresponding culture medium, the value range is 2~, n is the number of culture medium mother solutions, p i is the mixing ratio of the culture medium, and ε is an infinitesimal value greater than 0.

[0012] As an optional embodiment, the culture effect of the culture medium includes: cell viability, cell density, protein expression, sugar spectrum, and charge heterogeneity.

[0013] As one optional embodiment, the preset machine learning model includes: support vector machine regression model, K nearest neighbor model, XGBoost, ridge regression, LightGBM, random forest, GBDT, or deep learning model;

[0014] The deep learning model includes: a fully connected neural network, a convolutional neural network, a recurrent neural network or an attention model.

[0015] As one optional embodiment, the method of optimizing the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother solution includes:

[0016] Based on a preset global optimization algorithm or a preset heuristic algorithm, a regression prediction of the culture effect is performed in the search space to obtain a second input variable corresponding to the optimal culture effect of each of the culture medium mother solutions.

[0017] As one optional embodiment, the global optimization algorithm includes: genetic algorithm, greedy algorithm, annealing algorithm, ant colony algorithm, particle swarm algorithm, artificial bee colony algorithm, artificial fish swarm algorithm, shuffle frog leaping algorithm, fireworks algorithm, bacterial foraging optimization algorithm, firefly algorithm;

[0018] The heuristic algorithms include: Newton's method, quasi-Newton's method, conjugate gradient method, and gradient descent method.

[0019] As one optional embodiment, according to the second input variable corresponding to each of the culture medium mother solutions, the calculation formula for obtaining the target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions is:

[0020]

[0021] in, In order to optimize the culture effect of each of the culture medium mother solutions, the target mixing ratio is: is the second input variable corresponding to each of the culture medium mother solutions, n is the number of culture medium mother solutions, and ε is an infinitesimal value greater than 0.

[0022] An embodiment of the present invention provides a medium mixing ratio optimization device, comprising:

[0023] A machine learning model training module is used to train a preset machine learning model using the first input variable corresponding to each culture medium mother solution as input data and the culture effect of the culture medium corresponding to each culture medium mother solution as output data to obtain a trained machine learning model; wherein the first input variable corresponding to each culture medium mother solution is obtained by converting the initial mixing ratio of each culture medium mother solution;

[0024] A second input variable acquisition module is used to optimize the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother solution;

[0025] The target mixing ratio acquisition module is used to obtain the target mixing ratio corresponding to the optimal culture effect of each culture medium mother solution according to the second input variable corresponding to each culture medium mother solution.

[0026] An embodiment of the present invention provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the method for optimizing the culture medium mixing ratio described in the above embodiment is implemented.

[0027] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored computer program. When the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the vehicle positioning method described in the above embodiment.

[0028] Compared with the prior art, the embodiment of the present invention provides a method, device, equipment and medium for optimizing the culture medium mixing ratio. It can greatly shorten the formula development time for different cell lines by performing data dimensionality reduction processing on the data input into the model and developing the mixing ratio based on machine learning technology, select the mixed culture medium and provide one or more optimal mixing ratios for each cell line, thereby further improving the accuracy of the mixing experiment, reducing the learning cost of talents, and avoiding ineffective waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic flow chart of a method for optimizing a culture medium mixing ratio provided by an embodiment of the present invention;

[0030] Figure 2 It is a structural schematic diagram of a medium mixing ratio optimization device provided by an embodiment of the present invention;

[0031] Figure 3 It is a structural diagram of a terminal device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] In the first aspect, the present invention provides a method for optimizing the mixing ratio of a culture medium. Figure 1 , is a schematic flow chart of a method for optimizing a culture medium mixing ratio provided in an embodiment of the present invention, the method comprising steps S11 to S13:

[0034] S11. Take the first input variable corresponding to each culture medium mother liquor as input data, and take the culture effect of the culture medium corresponding to each of the culture medium mother liquors as output data, and train the preset machine learning model to obtain a trained machine learning model; wherein, the first input variable corresponding to each of the culture medium mother liquors is obtained by converting the initial mixing ratio of each of the culture medium mother liquors.

[0035] Specifically, the maximum addition values ​​of the mixing ratios of multiple mother liquid culture media are all set to 100%, the total value of the ratios is 100%, and the culture medium formulas of mixed mother liquids at different ratios are designed. After the above design is completed, the cells are cultured separately for 7 days, and samples are taken and counted on the third, fifth and seventh days, and the culture effect of the culture medium is tested. Obtain the changes in the cell growth process and the culture effect on the culture medium under different mixing ratios. The mixing ratio of the mother liquid culture medium, as well as relevant information such as the target product output and quality are recorded in the culture database for use by the algorithm model. The number of mother liquid culture media in the database is more than 2, and the total number of sample mixing formula databases is more than 10.

[0036] It should be noted that after the corresponding first input variable is obtained by converting the initial mixing ratio of each culture medium mother liquor, it is also necessary to perform data preprocessing on the first input variable, and input the converted and preprocessed first input variable as the input data of the model; in addition, after obtaining the culture effect of the corresponding culture medium under different mixing ratios, it is necessary to perform data cleaning, and use the culture effect of the culture medium after data cleaning as the output data of the model; wherein, the preprocessing method includes but is not limited to: standardization, zero mean centering, min-max normalization, etc.; data cleaning mainly adopts the following methods: missing value processing, outlier processing, data set balancing, data noise processing, data deduplication, and data format normalization.

[0037] S12. Optimize the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother solution.

[0038] It can be understood that the optimal culture effect of the culture medium mother solution means that the yield and quality of the target product of the culture medium mother solution are optimal.

[0039] S13. According to the second input variable corresponding to each of the culture medium mother solutions, a target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions is obtained.

[0040] It is understandable that the traditional method of developing culture medium mixing formula is based on two or several classic culture media and the culture medium mixing method is used. The culture medium mixing method is a relatively fast culture medium formula development method. It sets a reasonable ratio and mixes two or more culture media. According to the changes in cell growth and the culture effect of the culture medium, the culture medium mixing ratio is optimized to obtain the optimal culture medium.

[0041] In the existing technology, the existing traditional technology requires multiple rounds of experiments, each experiment cannot include all proportions, it takes a long time, and requires a lot of professional theoretical knowledge such as basic chemistry, biochemistry and molecular biology, cell biology, etc., and the formula obtained may not be optimal. Furthermore, the current traditional technology can only select specific proportions in some culture media for analysis, and in the process of ratio optimization, the model complexity will increase as the number of mixed culture media increases.

[0042] Specifically, the purpose of mixed culture media is to optimize the best ratio, but traditional methods are more about trying to get an excellent result or estimating that a certain culture medium is relatively high quality. Current traditional technologies can only process a small amount of experimental data and perform limited data analysis. They cannot mine potential information and perform quantitative analysis on large amounts of data. Culture medium mixing is a relatively quick way to obtain high-quality formulas, but how to choose mixed culture media and how to set the ratio of each mixed culture medium are the difficulties in optimizing traditional culture medium formulas. For example, for commercial culture media, the types and contents of ingredients in its formula are usually not public information. Therefore, especially when optimizing the ratio of mixed commercial culture media, it is impossible to use more professional theoretical knowledge such as basic chemistry, biochemistry and molecular biology, and cell biology to optimize the optimal ratio. With artificial intelligence technology, there is no need to master relevant theoretical knowledge to perform modeling optimization.

[0043] Compared with the prior art, the method for optimizing the mixing ratio of a culture medium provided in the embodiment of the present invention can greatly shorten the formulation development time of different cell lines by performing data dimension reduction processing on the data input into the model and developing the mixing ratio based on machine learning technology, select a mixed culture medium and provide one or more optimal mixing ratios for each cell line, thereby further improving the accuracy of the mixing test, reducing the learning cost of talents, and avoiding ineffective waste. Furthermore, AI technology can summarize and learn the culture effects of different mixed culture media from experimental data, without the need to master more professional theoretical knowledge such as basic chemistry, biochemistry and molecular biology, cell biology, etc., and reduce subjective interference caused by human emotional state, and quantitatively optimize the mixing ratio of the culture medium with objective data to achieve rapid and high optimization of the culture medium.

[0044] As an optional embodiment, the initial mixing ratio of each culture medium mother solution is converted into the first input variable corresponding to each culture medium mother solution by the following formula:

[0045]

[0046] Among them, x i is the first input variable corresponding to each of the culture medium mother solutions, i is the corresponding culture medium, the value range is 2~, n is the number of culture medium mother solutions, p i is the mixing ratio of the culture medium, and ε is an infinitesimal value greater than 0.

[0047] For example, in order to improve the accuracy of the model and to facilitate the recommendation of a mixing ratio with a total ratio of 100%, a data dimension reduction processing algorithm was invented for the culture medium mixing ratio data. The core algorithm is to convert the n-dimensional mother liquid culture medium mixing ratio into a multi-dimensional model input feature with a dimension minus one. Except for the mixing ratio of the first mother liquid, the mixing ratios of the remaining mother liquids correspond to one input feature.

[0048] It is worth noting that by using the above formula (1) to perform data dimensionality reduction processing, the prediction accuracy of the model can be improved, and the R square is higher. In addition, the value range space of the mixing ratio of the culture medium is converted from 0 to 1 to negative infinity to positive infinity, and while reducing the dimension, the limitation that the sum of the mixing ratios is 1 is eliminated.

[0049] As an optional embodiment, the culture effect of the culture medium includes but is not limited to: cell viability, cell density, protein expression, sugar spectrum, and charge heterogeneity.

[0050] As one optional embodiment, the preset machine learning model includes but is not limited to: support vector machine regression model, K nearest neighbor model, XGBoost, ridge regression, LightGBM, random forest, GBDT, or deep learning model;

[0051] The deep learning model includes but is not limited to: a fully connected neural network, a convolutional neural network, a recurrent neural network or an attention model.

[0052] Exemplarily, the input variables that have been reduced in dimension and preprocessed are used as model input values, and the changes in the cell growth process and the culture effect of the culture medium are used as output values ​​to construct a variety of machine learning algorithm models, including but not limited to: support vector machine regression model, K nearest neighbor model, XGBoost, ridge regression, LightGBM, random forest, GBDT, or deep learning model; the deep learning model includes but is not limited to: fully connected neural network, convolutional neural network, recurrent neural network or attention model; preferably a support vector machine regression model. When the amount of data is sufficient, the model can accurately predict the culture effect of the culture medium under different input variables. If the model accuracy is not enough and the data is insufficient, continue to design a mixed culture medium to obtain a trained machine learning model.

[0053] It is worth noting that the regression model established through computer AI technology can greatly shorten the development time of culture media for different cell lines, and can develop culture media for multiple different cell lines at the same time, providing one or more optimal mixing ratios for each cell line, reducing the learning cost of talents, improving the accuracy of experiments, and avoiding invalid waste. In addition, the use of machine learning models for parallel analysis can reduce repetitive experiments, effectively improve analysis efficiency, thereby speeding up research and development and saving time.

[0054] As one optional embodiment, the method of optimizing the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother solution includes:

[0055] Based on a preset global optimization algorithm or a preset heuristic algorithm, a regression prediction of the culture effect is performed in the search space to obtain a second input variable corresponding to the optimal culture effect of each of the culture medium mother solutions.

[0056] As one optional embodiment, the global optimization algorithm includes but is not limited to: genetic algorithm, greedy algorithm, annealing algorithm, ant colony algorithm, particle swarm algorithm, artificial bee colony algorithm, artificial fish swarm algorithm, shuffle frog leaping algorithm, fireworks algorithm, bacterial foraging optimization algorithm, firefly algorithm;

[0057] The heuristic algorithm includes but is not limited to: Newton's method, quasi-Newton's method, conjugate gradient method, and gradient descent method.

[0058] Exemplarily, based on the trained machine learning model, a deep mining numerical simulation is performed, and a global optimization algorithm or a heuristic algorithm is used to search for input variables in the search space for regression prediction of the culture effect; the heuristic algorithm includes but is not limited to: genetic algorithm, greedy algorithm, annealing algorithm, ant colony algorithm, particle swarm algorithm, artificial bee colony algorithm, artificial fish swarm algorithm, shuffle frog leaping algorithm, fireworks algorithm, bacterial foraging optimization algorithm, and firefly algorithm; the global optimization algorithm includes but is not limited to: Newton method, quasi-Newton method, conjugate gradient method, and gradient descent method commonly used in deep learning; preferably gradient descent method, the gradient descent method is preferably SGD, Momentum, Adagrad, RMSprop, Adam, Nadam. Using the above optimization algorithm, the optimal input variable for the culture medium culture effect predicted by model simulation is obtained.

[0059] As one of the optional embodiments, illustratively, when there are a large number of n mother liquid culture media, in order to reduce the number of mother liquids used, the use of m of them can be limited, and the remaining mother liquid culture media can be used at a ratio of 0%. The methods of limiting the use of mother liquids include but are not limited to: random limitation, specified limitation, and full combination exhaustive limitation. Random limitation means randomly selecting and limiting the use of m of the n mother liquid culture media; specified limitation means specifying the use of only some of the n mother liquid culture media based on the performance of the mother liquid culture media; full combination exhaustive limitation means limiting the number of mother liquid culture media used to m, but not limiting which mother liquid culture media are used. Therefore, m mother liquid culture media are selected from n mother liquid culture media, and the number of combinations is shown in the following formula. When the number of mother liquids used is limited, its input characteristics will also be limited.

[0060]

[0061] Wherein, m is a positive integer less than or equal to n. Only the unrestricted input features are selected for optimization, and the restricted input features can be restricted to a certain fixed value, and the mother liquor addition ratio corresponding to the fixed value is 0%. Based on this method, the input ratio under which the model predicts the optimal culture effect of the culture medium and the use of part of the culture medium mother liquor is restricted can be obtained.

[0062] As one optional embodiment, according to the second input variable corresponding to each of the culture medium mother solutions, the calculation formula for obtaining the target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions is:

[0063]

[0064] but:

[0065]

[0066] in, In order to optimize the culture effect of each of the culture medium mother solutions, the target mixing ratio is: is the second input variable corresponding to each of the culture medium mother solutions, n is the number of culture medium mother solutions, and ε is an infinitesimal value greater than 0.

[0067] Specifically, the input variables are reversed to obtain the mixing ratio of the mother liquid culture medium, which is the mixing ratio of the mother liquid culture medium with the best culture medium effect. Then, based on the mixing ratios of the various mother liquid culture media optimized by simulation, the recommended mixed culture medium is biologically verified. The cultured cells are inoculated into the mixed culture medium, and the quantity and quality of the target product are tested. If the requirements are met, it means that the culture medium can be optimized according to the machine learning algorithm. If the requirements are not met, the model algorithm is used again to continue to recommend the formula.

[0068] In order to further reflect the technical effect achieved by the method for optimizing the mixing ratio of a culture medium provided by the present invention, the present invention is further described below in combination with application examples of the inventors of the present invention during the research and development process:

[0069] 1. Select high-quality commercial culture medium as the mother liquid, taking seven commercial culture media as the mother liquid as an example. The maximum addition value of the seven mother liquid culture media is 100%, and the total value of the proportions is 100%. Design the culture medium of the mixed mother liquid at different proportions. After the above design is completed, culture CHO cells separately for 7 days, take samples and count them on the third, fifth and seventh days, and detect the yield and quality of the target product, that is, the culture effect of CHO cells in different mixed culture media can be obtained.

[0070] 2. Record the mixing ratio of the mother liquid culture medium, as well as the target product yield and quality and other related information in the culture database for use by the algorithm model. The number of mother liquid culture media in the database is 7, and the total number of sample mixing formula databases is more than 50.

[0071] 3. Data cleaning is performed on the culture effect of the culture medium. The following methods are mainly used for data cleaning: missing value processing, outlier processing, data set balancing, data noise processing, data deduplication, and data format normalization. Read the culture data.

[0072] In order to improve the accuracy of the model and facilitate the recommendation of a mixing ratio with a total ratio of 100%, the culture medium mixing ratio data is preprocessed. Each mother liquid culture medium mixing ratio is converted into an n-1-dimensional model input feature with a dimension minus one, and the formula is as follows.

[0073]

[0074] Taking ε as 1e-3 and taking the mixing ratio shown in Table 1 as an example, the first input variable after data dimensionality reduction as shown in Table 2 is obtained.

[0075] Table 1 Mixing ratios of mixed commercial culture media

[0076] Serial number Mother liquor 1 Mother liquor 2 Mother liquor 3 Mother liquor 4 Mother liquor 5 Mother liquor 6 Mother liquor 7 1 50％ 50％ 0％ 0％ 0％ 0％ 0％ 2 50％ 0％ 50％ 0％ 0％ 0％ 0％ 3 0％ 50％ 50％ 0％ 0％ 0％ 0％ 4 0％ 0％ 50％ 50％ 0％ 0％ 0％

[0077] Table 2 Input variables after data dimensionality reduction

[0078] Serial number X2 X3 X4 X5 X6 X7 1 1 0.002 0.002 0.002 0.002 0.002 2 0.002 1 0.002 0.002 0.002 0.002 3 1 1 0.002 0.002 0.002 0.002 4 0.002 1 1 0.002 0.002 0.002

[0079] Next, data preprocessing is performed. Preprocessing methods include but are not limited to: standardization, zero mean centering, min-max normalization, and standardization is preferred in this paper.

[0080] It should be noted that the culture effect of the culture medium includes but is not limited to cell viability, cell density, protein expression, sugar spectrum, and charge heterogeneity, which can be set according to actual needs. It is only used as an example here and is not limited.

[0081] 4. Based on the input features of dimensionality reduction and pretreatment as model input values, and the changes in cell growth and the culture effect on the culture medium as output values, a variety of machine learning algorithm models are constructed. Machine learning models include but are not limited to: support vector machine regression model, K nearest neighbor model, XGBoost, ridge regression, LightGBM, random forest, GBDT, or deep learning model; the deep learning model includes but is not limited to: fully connected neural network, convolutional neural network, recurrent neural network or attention model; preferably a support vector machine regression model. When the amount of data is sufficient, the model can accurately predict the culture effect of the culture medium under different input variables. If the model accuracy is not enough and the data is insufficient, continue with the mixed culture medium design.

[0082] 5. Since the number of 7 mother liquid culture media is large, for commercial purposes, we hope to limit the number of mother liquids used to 3 mother liquids. We can limit the use of 3 of the mother liquid culture media, and the usage ratio of the remaining mother liquid culture media is 0%. The use of all combinations of mother liquids is limited to an exhaustive limit. Therefore, the use of 7 mother liquids is limited to 3 mother liquids, and there are a total of 35 combinations.

[0083] 6. Based on the machine learning model constructed above, a deep mining numerical simulation is performed, and a global optimization algorithm or a heuristic algorithm is used to search for input variables in the search space for regression prediction of the culture effect; the heuristic algorithm includes but is not limited to: genetic algorithm, greedy algorithm, annealing algorithm, ant colony algorithm, particle swarm algorithm, artificial bee colony algorithm, artificial fish swarm algorithm, shuffle frog leaping algorithm, fireworks algorithm, bacterial foraging optimization algorithm, and firefly algorithm; the global optimization algorithm includes but is not limited to: Newton method, quasi-Newton method, conjugate gradient method, and gradient descent method commonly used in deep learning; preferably gradient descent method, the gradient descent method is preferably SGD, Momentum, Adagrad, RMSprop, Adam, Nadam. Optimize the numerical value of the input variable to obtain the input variable with the best culture effect of the model simulation culture medium. The above optimization algorithm only optimizes the 3 unrestricted culture medium input features to obtain the best input variables under 35 combinations.

[0084] Table 3 Optimal input variables corresponding to artificial intelligence culture medium

[0085]

[0086] 7. The mixing ratio of the mother culture medium is calculated by adjusting the different three input variables under the 35 combinations, using the following formula. This mixing ratio is the recommended medium mixing ratio. For the four mother culture medium formulas that are restricted for use, the mixing ratio is 0%.

[0087] 8. Based on the mixing ratio of each mother culture medium that has been simulated and optimized, the recommended mixed culture medium is verified biologically. Cells are inoculated into the mixed culture medium for culture, and the protein expression level, one of the culture effects, is detected. The results are shown in Table 4 below.

[0088] Table 4 Optimization results of culture medium mixing ratio using artificial intelligence

[0089]

[0090] The data in Table 4 show that the protein expression of CHO cells cultured in the mixed culture medium optimized by artificial intelligence methods was significantly improved compared with the single culture medium before optimization.

[0091] Through the above method, the optimal mixing ratio of the optimal percentage under the condition of limiting the number of mother liquors in 35 combinations can be obtained. For example, assuming the optimal mixing ratio under the combination of mother liquors 1, 2, and 4, or the optimal mixing ratio under the combination of mother liquors 1, 2, and 5, etc.

[0092] The embodiment of the present invention provides a medium mixing ratio optimization device, see Figure 2 , is a schematic diagram of the structure of a medium mixing ratio optimization device provided in an embodiment of the present invention, comprising:

[0093] The machine learning model training module 21 is used to train the preset machine learning model using the first input variable corresponding to each culture medium mother solution as input data and the culture effect of the culture medium corresponding to each culture medium mother solution as output data to obtain a trained machine learning model; wherein the first input variable corresponding to each culture medium mother solution is obtained by converting the initial mixing ratio of each culture medium mother solution;

[0094] A second input variable acquisition module 22 is used to optimize the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother solution;

[0095] The target mixing ratio acquisition module 23 is used to obtain the target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions according to the second input variable corresponding to each of the culture medium mother solutions.

[0096] Compared with the prior art, the culture medium mixing ratio optimization device provided in the embodiment of the present invention can greatly shorten the formula development time for different cell lines by performing data dimensionality reduction processing on the data input into the model and developing the mixing ratio based on machine learning technology, select the mixed culture medium and provide one or more optimal mixing ratios for each cell line, thereby further improving the accuracy of the mixing experiment, reducing the learning cost of talents, and avoiding ineffective waste.

[0097] As an optional embodiment, the initial mixing ratio of each culture medium mother solution is converted into the first input variable corresponding to each culture medium mother solution by the following formula:

[0098]

[0099] Among them, x i is the first input variable corresponding to each of the culture medium mother solutions, i is the corresponding culture medium, the value range is 2~, n is the number of culture medium mother solutions, p i is the mixing ratio of the culture medium, and ε is an infinitesimal value greater than 0.

[0100] As an optional embodiment, the culture effect of the culture medium includes but is not limited to: cell viability, cell density, protein expression, sugar spectrum, and charge heterogeneity.

[0101] As one optional embodiment, the preset deep learning model includes but is not limited to: support vector machine regression model, K nearest neighbor model, XGBoost, ridge regression, LightGBM, random forest, GBDT, or deep learning model;

[0102] The deep learning model includes but is not limited to: a fully connected neural network, a convolutional neural network, a recurrent neural network or an attention model.

[0103] As one optional embodiment, the method of optimizing the trained deep learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother solution includes:

[0104] Based on a preset global optimization algorithm or a preset heuristic algorithm, a regression prediction of the culture effect is performed in the search space to obtain a second input variable corresponding to the optimal culture effect of each of the culture medium mother solutions.

[0105] As one optional embodiment, the global optimization algorithm includes but is not limited to: genetic algorithm, greedy algorithm, annealing algorithm, ant colony algorithm, particle swarm algorithm, artificial bee colony algorithm, artificial fish swarm algorithm, shuffle frog leaping algorithm, fireworks algorithm, bacterial foraging optimization algorithm, firefly algorithm;

[0106] The heuristic algorithm includes but is not limited to: Newton's method, quasi-Newton's method, conjugate gradient method, and gradient descent method.

[0107] As one of the optional embodiments, illustratively, when there are a large number of n mother liquid culture media, in order to reduce the number of mother liquids used, the use of m of them can be limited, and the remaining mother liquid culture media can be used at a ratio of 0%. The methods of limiting the use of mother liquids include but are not limited to: random limitation, specified limitation, and full combination exhaustive limitation. Random limitation means randomly selecting and limiting the use of m of the n mother liquid culture media; specified limitation means specifying the use of only some of the n mother liquid culture media based on the performance of the mother liquid culture media; full combination exhaustive limitation means limiting the number of mother liquid culture media used to m, but not limiting which mother liquid culture media to use. Therefore, m mother liquid culture media are selected from n mother liquid culture media, and the number of combinations is shown in the following formula. When the number of mother liquids used is limited, its input characteristics will also be limited.

[0108]

[0109] Wherein, m is a positive integer less than or equal to n. Only the unrestricted input features are selected for optimization, and the restricted input features can be restricted to a certain fixed value, and the mother liquor addition ratio corresponding to the fixed value is 0%. Based on this method, the input ratio under which the model predicts the optimal culture effect of the culture medium and the use of part of the culture medium mother liquor is restricted can be obtained.

[0110] As one optional embodiment, according to the second input variable corresponding to each of the culture medium mother solutions, the calculation formula for obtaining the target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions is:

[0111]

[0112] but:

[0113]

[0114] in, In order to optimize the culture effect of each of the culture medium mother solutions, the target mixing ratio is: is the second input variable corresponding to each of the culture medium mother solutions, n is the number of culture medium mother solutions, and ε is an infinitesimal value greater than 0.

[0115] In addition, it should be noted that the specific implementation schemes and beneficial effects of each embodiment of a culture medium mixing ratio optimization device provided in an embodiment of the present invention are the same as the specific implementation schemes and beneficial effects of each embodiment of a culture medium mixing ratio optimization method provided in an embodiment of the present invention, and will not be repeated here.

[0116] The embodiment of the present invention provides a terminal device, see Figure 3, is a schematic diagram of the structure of a terminal device provided in an embodiment of the present invention. The terminal device 3 of this embodiment includes: a processor 30, a memory 31, and a computer program stored in the memory 31 and executable on the processor 30. When the processor 30 executes the computer program, the method for optimizing the culture medium mixing ratio described in the above embodiment is implemented. Alternatively, when the processor 30 executes the computer program, the functions of each module in the above-mentioned device embodiments are implemented.

[0117] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory 31 and executed by the processor 30 to implement the present invention. The one or more modules may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program in the terminal device 3.

[0118] The terminal device 3 may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The terminal device 3 may include, but is not limited to, a processor 30 and a memory 31. Those skilled in the art will appreciate that the schematic diagram is merely an example of a terminal device and does not constitute a limitation on the terminal device. The terminal device 3 may include more or fewer components than shown in the diagram, or may combine certain components, or different components. For example, the terminal device 3 may also include an input / output device, a network access device, a bus, etc.

[0119] The processor 30 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor 30 is the control center of the terminal device 3, and uses various interfaces and lines to connect various parts of the entire terminal device 3.

[0120] The memory 31 can be used to store the computer program and / or module. The processor 30 realizes various functions of the terminal device 3 by running or executing the computer program and / or module stored in the memory 31 and calling the data stored in the memory 31. The memory 31 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 31 can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0121] Wherein, if the module integrated in the terminal device 3 is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor 30, the steps of the above-mentioned various method embodiments can be implemented. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0122] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0123] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the culture medium mixing ratio optimization method as described above.

[0124] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units in the embodiments may be combined into one module or unit, and in addition they may be divided into a plurality of submodules or subunits. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0125] It should be noted that the above embodiments illustrate rather than limit the present invention, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements.

[0126] In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present invention and form different embodiments. For example, in the claims of the present invention, any one of the claimed embodiments may be used in any combination.

[0127] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for optimizing the mixing ratio of multiple mother liquid culture media, characterized in that: The new formula is formed by mixing mother liquid culture medium in different proportions, and the method comprises: The first input variable corresponding to each culture medium mother solution is used as input data, and the culture effect of the culture medium corresponding to each culture medium mother solution is used as output data to train the preset machine learning model to obtain a trained machine learning model; wherein, the first input variable corresponding to each culture medium mother solution is obtained by conversion according to the initial mixing ratio of each culture medium mother solution, and the conversion formula is as follows: In the formula, x i is the first input variable corresponding to each of the culture medium mother solutions, i is the corresponding culture medium, the value range is 2 to n, n is the number of culture medium mother solutions, p i is the mixing ratio of the culture medium, ε is an infinitesimal value greater than 0; Optimizing the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each of the culture medium mother solutions; According to the second input variable corresponding to each of the culture medium mother solutions, a target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions is obtained; Among them, the maximum addition value of the mixing ratio of multiple mother liquid culture media is set to 100%, and the total value of the ratio is 100%.

2. The method for optimizing the mixing ratio of multiple mother liquid culture media according to claim 1, characterized in that: The culture effects of the culture medium include: cell viability, cell density, protein expression, sugar spectrum, and charge heterogeneity.

3. The method for optimizing the mixing ratio of multiple mother liquid culture media according to claim 1, characterized in that: The preset machine learning models include: support vector machine regression model, K nearest neighbor model, XGBoost, ridge regression, LightGBM, random forest, GBDT, or deep learning model; The deep learning model includes: a fully connected neural network, a convolutional neural network, a recurrent neural network or an attention model.

4. The method for optimizing the mixing ratio of multiple mother liquid culture media according to claim 1, characterized in that: The method of optimizing the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother solution includes: Based on a preset global optimization algorithm or a preset heuristic algorithm, a regression prediction of the culture effect is performed in the search space to obtain a second input variable corresponding to the optimal culture effect of each of the culture medium mother solutions.

5. The method for optimizing the mixing ratio of multiple mother liquid culture media according to claim 4, characterized in that: The global optimization algorithms include: genetic algorithm, greedy algorithm, annealing algorithm, ant colony algorithm, particle swarm algorithm, artificial bee colony algorithm, artificial fish swarm algorithm, shuffle frog leaping algorithm, fireworks algorithm, bacterial foraging optimization algorithm, and firefly algorithm; The heuristic algorithms include: Newton's method, quasi-Newton's method, conjugate gradient method, and gradient descent method.

6. The method for optimizing the mixing ratio of multiple mother liquid culture media according to claim 1, characterized in that: According to the second input variable corresponding to each of the culture medium mother solutions, the calculation formula for obtaining the target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions is: in, In order to optimize the culture effect of each of the culture medium mother solutions, the target mixing ratio is: is the second input variable corresponding to each of the culture medium mother solutions, n is the number of culture medium mother solutions, and ε is an infinitesimal value greater than 0.

7. A device for optimizing the mixing ratio of multiple mother liquid culture media, characterized in that: A new formula is formed by mixing mother liquid culture medium in different proportions, and the device comprises: The machine learning model training module is used to train the preset machine learning model using the first input variable corresponding to each culture medium mother solution as input data and the culture effect of the culture medium corresponding to each culture medium mother solution as output data to obtain a trained machine learning model; wherein the first input variable corresponding to each culture medium mother solution is obtained by conversion based on the initial mixing ratio of each culture medium mother solution, and the conversion formula is as follows: In the formula, x i is the first input variable corresponding to each of the culture medium mother solutions, i is the corresponding culture medium, the value range is 2 to n, n is the number of culture medium mother solutions, p i is the mixing ratio of the culture medium, ε is an infinitesimal value greater than 0; A second input variable acquisition module is used to optimize the trained machine learning model based on a preset optimization algorithm to obtain a second input variable corresponding to the optimal culture effect of each culture medium mother solution; A target mixing ratio acquisition module, used to obtain a target mixing ratio corresponding to the optimal culture effect of each of the culture medium mother solutions according to the second input variable corresponding to each of the culture medium mother solutions; Among them, the maximum addition value of the mixing ratio of multiple mother liquid culture media is set to 100%, and the total value of the ratio is 100%.

8. A terminal device, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, it implements the method for optimizing the mixing ratio of multiple mother liquid culture media as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for optimizing the mixing ratio of multiple mother liquid culture media as described in any one of claims 1 to 6.

Images