WO2022045334A1 - 推論装置、推論方法、推論プログラム、モデル生成方法、推論サービス提供システム、推論サービス提供方法及び推論サービス提供プログラム - Google Patents

推論装置、推論方法、推論プログラム、モデル生成方法、推論サービス提供システム、推論サービス提供方法及び推論サービス提供プログラム Download PDF

Info

Publication number
WO2022045334A1
WO2022045334A1 PCT/JP2021/031711 JP2021031711W WO2022045334A1 WO 2022045334 A1 WO2022045334 A1 WO 2022045334A1 JP 2021031711 W JP2021031711 W JP 2021031711W WO 2022045334 A1 WO2022045334 A1 WO 2022045334A1
Authority
WO
WIPO (PCT)
Prior art keywords
inference
input data
lipid molecule
lipid
chemical structure
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2021/031711
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
友大 前田
臨 小川
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Takeda Pharmaceutical Co Ltd
Original Assignee
Takeda Pharmaceutical Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Takeda Pharmaceutical Co Ltd filed Critical Takeda Pharmaceutical Co Ltd
Priority to CN202180052991.6A priority Critical patent/CN115989511A/zh
Priority to EP21861755.3A priority patent/EP4207205A4/en
Priority to JP2022545756A priority patent/JP7735624B2/ja
Priority to US18/022,491 priority patent/US20240038340A1/en
Publication of WO2022045334A1 publication Critical patent/WO2022045334A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/30—Prediction of properties of chemical compounds, compositions or mixtures
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/20—Ensemble learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics

Definitions

  • the present disclosure relates to an inference device, an inference method, an inference program, a model generation method, an inference service providing system, an inference service providing method, and an inference service providing program.
  • a drug delivery system using particles containing lipid molecules is known for introducing active substances such as nucleic acids into cells with high efficiency.
  • particles of a complex are formed by including the active substance in particles containing a lipid molecule, and the active substance is introduced into the cell (transfection) through the particles of the complex. ) Is performed.
  • DDS drug Delivery System
  • Such DDS is used not only for transfection into in vivo cells by administration to a living body, but also for transfection into cells in vitro (in vitro, in situ or ex vivo).
  • the design and selection of the chemical structure of the lipid molecules constituting the particles containing the active substance are generally performed manually, the design and selection of appropriate lipid molecules according to the purpose is based on the experience of the expert and the selection. It depends largely on know-how.
  • lipid molecules with a designed or selected chemical structure are evaluated experimentally, and the work of redesigning and selecting according to the evaluation results is repeated, so it is necessary to search for a more suitable chemical structure of the lipid molecule. take time.
  • the present disclosure is intended to support the work of designing or selecting the chemical structure of lipid molecules that make up particles containing active substances.
  • the present inventors have generated a learning model based on data such as chemical structure information of lipid molecules, transfection efficiency and / or cell viability, and chemistry of lipid molecules using this. It has been found that structural information, transfection efficiency and / or cell viability can be inferred.
  • the present disclosure provides: [1] An acquisition unit that acquires input data including at least chemical structure information of lipid molecules, and For a learning model that correlates at least the input data containing the chemical structural information of the lipid molecule with the transfection efficiency and / or cell viability of the active substance contained in the particles containing the lipid molecule into cells. It has a trained model generated by the training process, and has The trained model is an inference device that infers the transfection efficiency and / or the cell viability associated with the input data newly acquired by the acquisition unit. [2] The transfection efficiency and / or cell viability used when the learning process is performed is a particle containing a lipid molecule having chemical structural information used when the learning process is performed.
  • the inference device which is calculated from the measurement results measured by introducing the contained active substance into cells.
  • the output when input data including at least the chemical structure information of the lipid molecule is input to the learning model is the transfection efficiency and / or the cell viability calculated from the measurement result.
  • the inference device which is generated by updating the model parameters of the learning model so as to approach. "4"
  • the acquisition unit performs a predetermined pretreatment on the newly acquired input data, and the trained model is a transfection efficiency and / or cell associated with the input data after the pretreatment.
  • the inference device which infers the survival rate.
  • It is an inference program for making a computer execute an execution process that executes a trained model generated by performing a training process.
  • the execution step is an inference program that infers the transfection efficiency and / or the cell viability associated with the input data newly acquired in the acquisition step by executing the trained model.
  • a learning model that correlates at least input data including chemical structural information of lipid molecules with transfection efficiency and / or cell viability of active substances contained in particles containing the lipid molecules.
  • a model generation method that generates a trained model by performing training processing.
  • the trained model is an inference device that infers the chemical structure information of the lipid molecule associated with the input data newly acquired by the acquisition unit.
  • the input data used when the learning process is performed is a measurement measured by introducing an active substance contained in particles containing a designed or selected lipid molecule into cells.
  • the inference apparatus according to [8] which comprises the transfection efficiency and / or cell viability of the active substance contained in the particles containing the lipid molecule, which is calculated from the result, into cells.
  • the output when the input data including the preconditions is input to the learning model is close to the chemical structure information of the lipid molecule used when the learning process is performed.
  • the inference device according to [8] which is generated by updating the model parameters of the learning model.
  • the acquisition unit performs a predetermined pretreatment on the newly acquired input data, and the trained model obtains the chemical structure information of the lipid molecule associated with the input data after the pretreatment.
  • the inference device according to [8] for inferring.
  • a reasoning method that has an execution process that executes a trained model that has been trained.
  • the execution step is an inference method that infers the chemical structure information of the lipid molecule associated with the input data newly acquired in the acquisition step by executing the trained model.
  • It is an inference program for making a computer execute an execution process that executes a trained model.
  • the execution step is an inference program that infers the chemical structure information of the lipid molecule associated with the input data newly acquired in the acquisition step by executing the trained model.
  • Trained model By inputting input data including preconditions newly acquired by the user by the acquisition unit, an inference having a providing unit that provides the user with the chemical structure information of the lipid molecule inferred by the learned model.
  • Service provision system By inputting input data including preconditions newly acquired by the user by the acquisition unit, the trained model charges the user when the chemical structure information of the lipid molecule is inferred.
  • the inference service providing system according to [15], further comprising a billing unit.
  • the lipid molecule is contained, which is calculated from the measurement result measured by introducing the active substance contained in the particles containing the lipid molecule having the chemical structure information inferred by the trained model into the cell.
  • the charging unit changes the charging content to the user [16].
  • the inference service providing system described in. [18] The acquisition process of acquiring from the user the prerequisites for designing or selecting the lipid molecules that make up the particles containing the active substance, Generated by performing a learning process on a learning model that associates input data including prerequisites for designing or selecting lipid molecules that make up particles containing active substances with chemical structural information of the lipid molecules.
  • the execution process to execute the trained model, and By inputting input data including preconditions newly acquired by the user in the acquisition step, the user has a providing step of providing the chemical structure information of the lipid molecule inferred by the learned model to the user.
  • Inference service provision method [19] The acquisition process of acquiring from the user the prerequisites for designing or selecting the lipid molecules that make up the particles containing the active substance, Generated by performing a learning process on a learning model that associates input data including prerequisites for designing or selecting lipid molecules that make up particles containing active substances with chemical structural information of the lipid molecules. The execution process to execute the trained model, and A computer provides the user with the chemical structure information of the lipid molecule inferred by the trained model by inputting the input data including the preconditions newly acquired by the user in the acquisition process. A program that provides an inference service to be executed by a computer. [20] An acquisition unit that acquires input data including prerequisites for designing or selecting lipid molecules that make up particles containing active substances.
  • a reinforcement learning model that infers chemical structural information of lipid molecules by inputting input data including preconditions acquired by the acquisition unit, and The lipid molecule is contained, which is calculated from the measurement result measured by introducing the active substance contained in the particles containing the lipid molecule having the chemical structure information inferred by the enhanced learning model into the cell. It has a calculation unit that calculates the reward based on the transfection efficiency and / or the cell viability of the active substance contained in the particles.
  • the reinforcement learning model is an inference device that performs learning processing based on a reward calculated by the calculation unit. [21] The inference device according to [20], wherein the calculation unit calculates the reward so as to be maximized by increasing the transfection efficiency and / or the cell survival rate.
  • the acquisition unit performs a predetermined pretreatment on the input data, and the reinforcement learning model infers the chemical structure information of the lipid molecule by inputting the input data after the pretreatment.
  • the inference device according to [20].
  • the lipid molecule is contained, which is calculated from the measurement result measured by introducing the active substance contained in the particles containing the lipid molecule having the chemical structure information inferred by the enhanced learning model into the cell.
  • the reinforcement learning model is an inference method that performs learning processing based on the reward calculated in the calculation process.
  • the lipid molecule is contained, which is calculated from the measurement result measured by introducing the active substance contained in the particles containing the lipid molecule having the chemical structure information inferred by the enhanced learning model into the cell.
  • the reinforcement learning model is an inference program that performs learning processing based on the reward calculated in the calculation process.
  • a reinforcement learning model that infers chemical structure information of lipid molecules by inputting input data including preconditions acquired by the user by the acquisition unit.
  • a providing unit that provides the user with the chemical structure information of the lipid molecule inferred by the reinforcement learning model.
  • the lipid molecule is contained, which is calculated from the measurement result measured by introducing the active substance contained in the particles containing the lipid molecule having the chemical structure information inferred by the enhanced learning model into the cell. It has a calculation unit that calculates the reward based on the transfection efficiency and / or the cell viability of the active substance contained in the particles.
  • the reinforcement learning model is an inference service providing system that performs learning processing based on a reward calculated by the calculation unit. [26]
  • the active substance contained in the particles containing the lipid molecule having the chemical structure information inferred by the enhanced learning model is calculated from the measurement result measured by being introduced into the cell.
  • the charging unit changes the charging content to the user.
  • the inference service providing system [28] An acquisition step of acquiring a prerequisite for designing or selecting a lipid molecule constituting a particle containing an active substance from a user, and An execution step of executing a reinforcement learning model for inferring chemical structure information of lipid molecules by inputting input data including preconditions acquired by the user in the acquisition step, and an execution step.
  • a step of providing the chemical structure information of the lipid molecule inferred by the reinforcement learning model to the user, and The lipid molecule calculated from the measurement result measured by introducing the active substance contained in the particles containing the lipid molecule having the chemical structure information inferred by the enhanced learning model into the cell is used.
  • a method for providing an inference service comprising a calculation step of calculating a reward based on the efficiency of transfection of an active substance contained in the contained particles into cells and / or the cell viability.
  • the reinforcement learning model is an inference service providing method that performs learning processing based on the reward calculated in the calculation process.
  • a step of providing the chemical structure information of the lipid molecule inferred by the reinforcement learning model to the user, and The lipid molecule calculated from the measurement result measured by introducing the active substance contained in the particles containing the lipid molecule having the chemical structure information inferred by the enhanced learning model into the cell is used.
  • a inference service providing program for causing a computer to perform a calculation step of calculating a reward based on the efficiency of transfection of an active substance contained in the contained particles into cells and / or the cell viability.
  • the reinforcement learning model is an inference service providing program that performs learning processing based on the reward calculated in the calculation process.
  • a learning model that correlates at least input data containing chemical structural information of a lipid molecule with transfection efficiency and / or cell viability of an active substance contained in particles containing the lipid molecule.
  • the generation unit is Based on the inference result, one of the search spaces is selected from a plurality of search spaces according to the combination of the molecular fragment of the formable hydrocarbon and the chemical skeleton of the lipid molecule, and the characteristics of the selected search space are used.
  • the inference apparatus which generates chemical structure information of the next new lipid molecule.
  • the reasoning apparatus according to [31], wherein the plurality of search spaces differ from each other in the combination of the length, saturation, and number of branches of the molecular fragment and the type of the chemical skeleton of the lipid molecule.
  • the inference device wherein the generation unit generates chemical structure information of the next new lipid molecule under a predetermined constraint condition.
  • the inference device wherein the generation unit generates chemical structure information of the next new lipid molecule with the acquired precondition as the predetermined constraint.
  • a learning model that correlates at least input data containing chemical structural information of a lipid molecule with transfection efficiency and / or cell viability of an active substance contained in particles containing the lipid molecule.
  • the execution process that executes the trained model generated by the training process, and If the transfection efficiency and / or cell viability associated with the input data, including the chemical structural information of the newly generated lipid molecule, was inferred by the trained model, then based on the inference results, the next new A transfection method comprising a generation step of repeating a generation process for generating chemical structural information of a lipid molecule until a predetermined termination condition is satisfied.
  • a learning model that correlates at least input data containing chemical structural information of a lipid molecule with transfection efficiency and / or cell viability of an active substance contained in particles containing the lipid molecule.
  • the execution process that executes the trained model generated by the training process, and If the transfection efficiency and / or cell viability associated with the input data, including the chemical structural information of the newly generated lipid molecule, was inferred by the trained model, then based on the inference results, the next new An inference program for causing a computer to perform a generation process that repeats the generation process for generating chemical structural information of lipid molecules until certain termination conditions are met.
  • FIG. 1 is a diagram showing an example of accumulating various data in the drug delivery system review process.
  • FIG. 2 is a diagram showing an application example of the inference device according to the first embodiment.
  • FIG. 3 is a diagram showing an example of the hardware configuration of the inference device.
  • FIG. 4 is a diagram showing an example of the functional configuration of the learning device according to the first embodiment.
  • FIG. 5 is a diagram showing an example of the functional configuration of the inference device according to the first embodiment.
  • FIG. 6 is an example of a flowchart showing the flow of transfection efficiency and / or cell viability inference processing.
  • FIG. 7A is a diagram showing an embodiment of the learning device.
  • FIG. 7B is a diagram showing an embodiment of the inference device.
  • FIG. 7A is a diagram showing an embodiment of the learning device.
  • FIG. 7B is a diagram showing an embodiment of the inference device.
  • FIG. 8 is a diagram showing an application example of the inference device according to the second embodiment.
  • FIG. 9 is a diagram showing an example of the functional configuration of the learning device according to the second embodiment.
  • FIG. 10 is a diagram showing an example of the functional configuration of the inference device according to the second embodiment.
  • FIG. 11 is an example of a flowchart showing the flow of the chemical structure information inference processing of the lipid molecule.
  • FIG. 12 is a diagram showing an application example of the inference service providing system according to the third embodiment.
  • FIG. 13 is an example of a flowchart showing the flow of the inference service provision process.
  • FIG. 14 is a diagram showing an application example of the inference service providing system according to the fourth embodiment.
  • FIG. 14 is a diagram showing an application example of the inference service providing system according to the fourth embodiment.
  • FIG. 15 is a diagram showing an example of the functional configuration of the inference device according to the fourth embodiment.
  • FIG. 16 is another example of a flowchart showing the flow of the inference service provision process.
  • FIG. 17 is a diagram showing an example of the functional configuration of the inference device according to the fifth embodiment.
  • FIG. 18 is an example of a flowchart showing the flow of the generation process.
  • FIG. 19 is a diagram showing an example of the functional configuration of the inference device according to the sixth embodiment.
  • FIG. 1 is a diagram showing an example of accumulating various data in the drug delivery system review process.
  • the particles containing the active substance are at least, -A case where lipid molecules are mixed with an active substance (for example, nucleic acid 140) to form complex particles.
  • a lipid molecule forms an outer shell and a complex substance is formed by containing an active substance (for example, nucleic acid 140) in the outer shell.
  • Examples of the lipid molecule treated in one aspect of the present disclosure include a cationic lipid molecule.
  • Cationic lipid means a lipid having a net positive charge at a selected pH, such as physiological pH.
  • Examples of the method for producing particles containing a lipid molecule and an active substance include the methods described in International Publication No. 2016/021683, International Publication No. 2019/131389, and International Publication No. 2020/032184.
  • Examples of the lipid molecule treated in another aspect of the present disclosure include an anionic lipid molecule, a cholesterol derivative molecule, an amphiphilic lipid molecule and the like.
  • the nucleic acid treated in one aspect of the present disclosure may be any molecule as long as it is a polymer of a nucleotide and a molecule having a function equivalent to the nucleotide, for example, RNA which is a polymer of ribonucleotide, weight of deoxyribonucleotide.
  • RNA which is a polymer of ribonucleotide, weight of deoxyribonucleotide.
  • examples thereof include a polymer in which combined DNA, ribonucleotides and deoxyribonucleotides are mixed, and a nucleotide polymer containing a nucleotide analog, and may be a nucleotide polymer containing a nucleic acid derivative.
  • the nucleic acid may be a single-stranded nucleic acid or a double-stranded nucleic acid.
  • the double-stranded nucleic acid also includes a double-stranded nucleic acid in which one strand is hybridized with the other strand under stringent conditions.
  • the nucleic acid handled in the present embodiment is not particularly limited, and is, for example, improvement of a disease, symptom, disorder, or pathological condition, reduction of the disease, symptom, disorder, or pathological condition, or prevention of its onset (the present invention).
  • nucleic acid for the purpose of "treatment of a disease, etc.”), and it may be used to regulate the expression of a desired protein useful for research, although it does not contribute to the treatment of a disease, etc. It may be a nucleic acid of.
  • Specific examples of the nucleic acids handled in the present embodiment include siRNA, miRNA, miRNA mimic, antisense nucleic acid, ribozai ⁇ , mRNA, decoy nucleic acid, aptamer, DNA, and analogs or derivatives obtained by artificially modifying these. Can be mentioned.
  • the designer 110 designs or selects the chemical structure of the lipid molecule from the design prerequisite 101 based on the experience and know-how so far.
  • the lipid molecule 111 having the chemical structure designed or selected by the designer 110 is subjected to experimental processing and evaluation processing by the experimenter and the evaluator 120, and the evaluation data 121 is notified to the designer 110.
  • the designer 110 redesigns or selects the chemical structure of the lipid molecule based on the notified evaluation data 121.
  • the lipid molecule 111'(not shown) having the chemical structure redesigned or selected by the designer 110 is subjected to the experimental treatment and the evaluation treatment again by the experimenter and the evaluator 120, and the evaluation data 121'(not shown) is obtained.
  • the designer 110 is notified.
  • the design or selection by the designer 110 and the experimental processing and evaluation processing by the experimenter and the evaluator 120 are repeated a plurality of times. This allows the designer 110 to search for a more suitable chemical structure of the lipid molecule that constitutes the particle containing the nucleic acid 140.
  • the nucleic acid 140 is contained by the particles containing the generated lipid molecule 130, and the nucleic acid 140 is complexed. Particles of body 150 are formed.
  • the particles of the complex 150 may contain components other than the lipid molecule and the nucleic acid, if necessary, in addition to the nucleic acid 140. Examples of such components include appropriate amounts of stabilizers, antioxidants and the like. These ingredients can be pharmaceutically acceptable ingredients.
  • the formed complex 150 particles are applied to 160 target animals and the like, and in vitro (in vitro, in situ or ex vivo) cells and / or tissues 160 and the like. Changes (or in vitro (in vitro, in situ or ex vivo)) caused by the introduction of the particles of the complex 150 into specific cells in the body (in vivo) of 160 such as target animals (Changes caused by the introduction) are measured by various measuring methods and measuring instruments, and are output as effect data 161.
  • a method for applying the particles of the complex 150 to the target animal or the like 160 a method known to those skilled in the art may be used.
  • Specific examples of the method of application to the target animals 160 include intravenous, intramuscular, intraperitoneal, intracerebral intrathecal (intracerobrospinal), subcutaneous, intra-articular, and synovial fluid as a bolus or by continuous injection over a certain period of time.
  • Intramuscular, intrathecal, oral, topical or inhalation routes may be mentioned. Further, those skilled in the art can appropriately determine the number of administrations, the dose, and the administration interval.
  • particles of the complex 150 are placed in a container in which the cells of interest are cultured. Addition and culturing for a certain period may be mentioned. A person skilled in the art can appropriately determine the number of additions, the amount of addition, the interval of addition, the culture conditions, the culture period, and the like.
  • the effect data 161 includes the transfection efficiency and / or the cell viability of the nucleic acid 140 contained in the particles containing the lipid molecule 130, which is calculated from the measurement results.
  • the transfection efficiency can be appropriately evaluated using a known method based on the attributes of the nucleic acid as an active substance and the like.
  • siRNA when siRNA is used as a nucleic acid, it can be evaluated based on the knockdown rate of the expression of the gene targeted by the siRNA. More specifically, the expression level of the gene (for example, mRNA) in the group (administered group) to which the particles containing the siRNA were administered, and the control group (for example, the group to which nothing was administered, siRNA-free or particles).
  • the cell viability when transfected can also be appropriately evaluated using a known method. For example, it can be evaluated by comparing the number of cells before application and the number of cells after application and measuring the ratio of the number of cells after application to the number of cells before application. A person skilled in the art can appropriately select a method for measuring the number of cells and a measuring instrument.
  • the evaluation data and effect data exemplify the transfection efficiency and cell viability into cells, and further, the pharmacokinetics (absorption) of nucleic acid 140 in vivo when the particles of the complex 150 are applied to 160 such as a target animal. , Distribution, metabolism) and toxicity data may also be included, but not limited to these.
  • various data acquired in these series of flows in the drug delivery system review process 100 are stored in the data storage unit 170 regarding the drug delivery system.
  • various data stored in the data storage unit 170 regarding the drug delivery system include, for example, -Design prerequisites 101, ⁇ Evaluation data 121, -Chemical structure information of lipid molecule 130, -Type of nucleic acid 140, chemical structure information, ⁇ Chemical structure information of complex 150, Effect data 161 (including transfection efficiency and / or cell viability), Etc. are included.
  • the chemical structure information of the lipid molecule is not particularly limited, and examples thereof include a chemical formula, a three-dimensional structure, a molecular weight, a number of carbon atoms, a number of nitrogen atoms, a number of oxygen atoms, and an electric charge.
  • the chemical structure information of the nucleic acid is not particularly limited, and examples thereof include the number of bases constituting the nucleic acid, the chemical formula, the three-dimensional structure, the molecular weight, and the electric charge.
  • Examples of the chemical structure information of the complex include the particle size and membrane potential of the complex.
  • the various data stored in the data storage unit 170 regarding the drug delivery system may further include various public information (for example, patent gazettes, papers) and data that can be obtained from a database.
  • FIG. 2 is a diagram showing an application example of the inference device according to the first embodiment.
  • 260 or in vitro (in vitro, in situ or ex vivo) cells and / or tissue 260), ⁇ Types of 260 diseases such as target animals, Attributes of the included active substance (eg, nucleic acid 240) (eg, type of active substance (eg, nucleic acid 240), chemical structure information, etc.), A target into which particles of the complex 250 are introduced (specific cells in vivo (in vivo) such as a target animal or specific cells 260 in vitro (in vitro, in situ or ex vivo)). Etc. are input to the designer 110.
  • active substance eg, nucleic acid 240
  • a target into which particles of the complex 250 are introduced specific cells in vivo (in vivo) such as a target animal or specific cells 260 in vitro (in vitro, in situ or ex vivo)
  • Etc. are input to the designer 110.
  • the designer 110 designs or selects the chemical structure of the lipid molecule from the design prerequisite 201 based on the experience and know-how so far.
  • the chemical structure information of the lipid molecule 211 having the chemical structure designed or selected by the designer 110 is input to the inference device 220.
  • the inference device 220 has a trained model generated by the learning device 210.
  • the learning device 210 generates a trained model by performing a learning process on the learning model using a learning data set generated based on various data stored in the data storage unit 170 related to the drug delivery system. do.
  • the inference device 220 uses the trained model generated by the learning device 210 to generate evaluation data 221 for the chemical structure information of the lipid molecule 211.
  • the evaluation data 221 generated by the inference device 220 is notified to the designer 110.
  • the design or selection by the designer 110 and the generation of the evaluation data by the inference device 220 are repeated a plurality of times. This allows the designer 110 to search for a more suitable chemical structure of the lipid molecule that constitutes the particle containing the nucleic acid 240.
  • the nucleic acid 240 is contained by the particles containing the generated lipid molecule 230, and the nucleic acid 240 is complexed. Particles of body 250 are formed.
  • the particles of the complex 250 may contain components other than the lipid molecule and the nucleic acid, if necessary, in addition to the nucleic acid 240. Examples of such components include appropriate amounts of stabilizers, antioxidants and the like. These ingredients can be pharmaceutically acceptable ingredients.
  • the formed complex 250 particles are applied to 260 target animals and the like, and in vitro (in vitro, in situ or ex vivo) cells and / or tissues 260 and the like. Changes caused by the introduction of the particles of the complex 250 into specific cells in vivo (in vivo) such as the target animal (or in vitro, in situ or ex vivo) to specific cells 260 (Changes caused by the introduction) are measured by various measuring methods and measuring instruments, and are output as effect data 261.
  • a method for applying the particles of the complex 250 to the target animal or the like 260 a method known to those skilled in the art may be used.
  • Specific examples of the method of application to the target animals such as 260 include intravenous, intramuscular, intraperitoneal, intracerebral intrathecal (intracerobrospinal), subcutaneous, intra-articular, and synovial fluid as a bolus or by continuous injection over a certain period of time.
  • Intramuscular, intrathecal, oral, topical or inhalation routes may be mentioned.
  • a person skilled in the art can appropriately determine and design the number of administrations, the dose, and the administration interval.
  • particles of the complex 250 are placed in a container in which the cells of interest are cultured. Addition and culturing for a certain period may be mentioned. A person skilled in the art can appropriately determine the number of additions, the amount of addition, the interval of addition, the culture conditions, the culture period, and the like.
  • the change caused by the introduction of the particles of the complex 250 into a specific cell or the like in vivo of the target animal or the like 260 is the target animal or the like 260 into which the particles of the complex 250 are introduced, or It can be grasped by measuring a sample containing the specific cell collected from the target animal or the like 260.
  • the sample is not particularly limited as long as the specific cell is contained, and is not limited to whole blood, plasma, urine, serum, lymph, saliva, anal and vaginal secretions, sweat and semen, body fluids not limited thereto, and organs. , Tissue samples and cells obtained from tissue biopsy.
  • the collected sample may be labeled by any known method for use in measurement.
  • target animals include, but are not limited to, animals such as mice, rats, guinea pigs, dogs, cats, rabbits, cows, horses, sheep, goats, and pigs, in addition to humans.
  • the target animal or the like may be a healthy human or animal, or may be a human (patient) or animal suffering from some kind of disease.
  • any in vitro or in vivo method known to those skilled in the art may be used. Specific examples thereof include flow cytometry, immunological assay, mRNA transcript analysis, PCR method and hybridization method. Alternatively, sequencing methods, RFLP methods, Western blots, ELISAs, radioimmunoassays, immunoprecipitation, FACS, HPLC, surface plasmon resonance, optical spectroscopy, mass spectrometry and the like can be mentioned.
  • the transfection efficiency to cells and the cell viability are exemplified as measurement results, but the measurement items are not limited thereto.
  • the effect data 261 includes the transfection efficiency and / or the cell viability of the nucleic acid 240 contained in the particles containing the lipid molecule 230, which is calculated from the measurement results.
  • the effect data 261 may further include data on the pharmacokinetics (absorption, distribution, metabolism) and toxicity of the nucleic acid 240 in vivo when the particles of the complex 250 are applied to 260 of the target animal, etc., and are limited thereto. Not done.
  • the evaluation data 121 is generated by performing the experimental processing and the evaluation processing by the experimenter and the evaluator 120, but by applying the inference device 220, the experimental processing and the evaluation processing can be performed. Evaluation data 221 can be generated without doing this.
  • FIG. 3 is a diagram showing an example of the hardware configuration of the inference device.
  • the inference device 220 includes a processor 301, a memory 302, an auxiliary storage device 303, an I / F (Interface) device 304, a communication device 305, and a drive device 306.
  • the hardware of the inference device 220 is connected to each other via the bus 307.
  • the processor 301 has various arithmetic devices such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit).
  • the processor 301 reads various programs installed in the auxiliary storage device 303 onto the memory 302 and executes them.
  • the memory 302 has a main storage device such as a ROM (ReadOnlyMemory) and a RAM (RandomAccessMemory).
  • the processor 301 and the memory 302 form a so-called computer, and the processor 301 executes various programs read on the memory 302, so that the computer realizes various functions.
  • the auxiliary storage device 303 stores various programs and various data used when various programs are executed by the processor 301.
  • the I / F device 304 is a connection device that connects the operation device 310, the display device 311 and the inference device 220.
  • the I / F device 304 receives various instructions to the inference device 220 via the operation device 310. Further, the I / F device 304 outputs the processing result of the inference device 220 via the display device 311.
  • the communication device 305 is a communication device for communicating with other devices via a network.
  • the drive device 306 is a device for setting the recording medium 312.
  • the recording medium 312 referred to here includes a medium such as a CD-ROM, a flexible disk, a magneto-optical disk, or the like that optically, electrically, or magnetically records information. Further, the recording medium 312 may include a semiconductor memory or the like for electrically recording information such as a ROM or a flash memory.
  • the various programs installed in the auxiliary storage device 303 are installed, for example, by setting the distributed recording medium 312 in the drive device 306 and reading the various programs recorded in the recording medium 312 by the drive device 306. Will be done.
  • various programs installed in the auxiliary storage device 303 may be installed by being downloaded from the network via the communication device 305.
  • FIG. 4 is a diagram showing an example of the functional configuration of the learning device according to the first embodiment.
  • the learning data set 400 is an example of a learning data set generated based on various data stored in the data storage unit 170 related to the drug delivery system.
  • the items of each information of the learning data are illustrated below, but the present invention is not limited to these, and a part or all of them can be appropriately used as the learning data. ..
  • the learning data set 400 has input data and correct answer data, and the input data contains, for example, “disease type”, “nucleic acid type”, and “introduction” as information items. Includes “target”, “chemical structure information of lipid molecules”, and “attribute information of target animals, etc.”
  • the "disease type” includes, for example, when the target animal or the like 160 is a patient (hereinafter, may be referred to as a "target patient” in the present specification), the type of the disease, "disease A1". Stored.
  • nucleic acid type for example, "nucleic acid X 1 ", “nucleic acid X 2 " and the like are stored as the type of nucleic acid corresponding to "disease A 1 ".
  • Examples of the information indicating the target cell referred to here include the origin or the organ or tissue in which the cell is present in the living body, the type of cell (for example, nerve cell, parenchymal cell, stromal cell, etc.) and the like. However, it is not limited to these.
  • “Chemical structure information of lipid molecules” is more appropriate than previously designed or selected by the designer 110 based on design prerequisites 101 such as “type of disease”, “type of nucleic acid”, “target of introduction”. Information indicating the chemical structure of various lipid molecules is stored.
  • the attribute information of the target animal or the like 160 (for example, the target patient) is stored.
  • the correct answer data includes "transfection efficiency and / or cell viability" as an information item.
  • the "transfection efficiency and / or cell viability” stores the transfection efficiency and / or cell viability of the nucleic acid contained in the particles containing the lipid molecule into the cells.
  • the particles of the corresponding complex are applied to the corresponding target animal such as 260 or in vitro (in vitro, in situ or ex vivo) cells and / or tissue 260, and the applied target animal is applied.
  • Etc. and / or cell viability calculated from the measurement results obtained by measuring the sample isolated from the transfection efficiency and / or the cell viability are stored.
  • These learning data sets may be data obtained from various public information (for example, patent gazettes, papers) or databases.
  • a learning program is installed in the learning device 210, and when the learning program is executed, the learning device 210 functions as a preprocessing unit 410, a learning model 420, and a comparison / change unit 430.
  • the pre-processing unit 410 acquires "input data" of the learning data set 400 and performs various pre-processing to generate pre-processing and post-processing data suitable for input to the learning model 420.
  • the various pre-processing performed by the pre-processing unit 410 includes a process of normalizing the input data, a process of vectorizing the input data, and the like.
  • the learning model 420 is a model that associates input data with correct answer data (transfection efficiency and / or cell viability). Specifically, the learning model 420 takes the post-treatment data notified by the pre-treatment unit 410 as an input, and outputs the transfection efficiency and / or the cell survival rate.
  • a learning process for updating the model parameters is performed by back-propagating the error calculated by the comparison / change unit 430.
  • This will generate a trained model. That is, the trained model is generated by updating the model parameters of the training model 420 so that the output of the training model 420 approaches the correct data (transfection efficiency and / or cell viability).
  • the comparison / change unit 430 compares the transfection efficiency and / or the cell survival rate output from the learning model 420 with the correct answer data (transfection efficiency and / or the cell survival rate) of the learning data set 400. Calculate the error. Further, the comparison / change unit 430 updates the model parameters of the learning model 420 by back-propagating the calculated error.
  • FIG. 5 is a diagram showing an example of the functional configuration of the inference device according to the first embodiment.
  • the input data 500_1, 500_2, 500_3, ... -Information included in the design prerequisite 201 type of disease, type of nucleic acid, target of introduction, attribute information of target animal, etc.
  • the input data may be various public information (for example, patent gazettes, papers) or data obtained from a database.
  • Input data 500_1, 500_2, 500_3, ... Containing chemical structure information of different lipid molecules are input to the inference device 220.
  • An inference program is installed in the inference device 220, and when the inference program is executed, the inference device 220 functions as a preprocessing unit 510, a trained model 520, and an evaluation data generation unit 530.
  • the pre-processing unit 510 is an example of the acquisition unit, and has the same function as the pre-processing unit 410 of the learning device 210. Specifically, the preprocessing unit 510 acquires the input data 500_1, 500_2, 500_3, ..., And performs various preprocessing on the acquired input data 500_1, 500_2, 500_3, .... , Generate post-processing data.
  • the trained model 520 is a trained model generated by performing a training process by the learning device 210, and inputs preprocessed data notified from the preprocessing unit 510 as input to transfection efficiency and / or cell survival. Infer the rate.
  • the evaluation data generation unit 530 generates evaluation data 540_1, 540_2, 540_3, ... Based on the transfection efficiency and / or the cell survival rate inferred by the trained model 520.
  • the transfection efficiency and / or cell viability is inferred for the lipid molecule having the chemical structure designed or selected by the designer 110 without performing experimental treatment and evaluation treatment. Evaluation data can be generated.
  • FIG. 6 is an example of a flowchart showing the flow of transfection efficiency and / or cell viability inference processing.
  • step S601 the learning device 210 acquires various data from the data storage unit 170 related to the drug delivery system.
  • step S602 the learning device 210 generates a learning data set 400 using various acquired data.
  • step S603 the learning device 210 performs learning processing on the learning model 420 using the learning data set 400, and generates a trained model 520.
  • step S604 the inference device 220 acquires input data (for example, input data 500_1) including chemical structure information of the lipid molecule newly designed or selected by the designer 110.
  • step S605 the inference device 220 executes the trained model 520 by inputting the acquired input data (for example, input data 500_1) into the trained model 520, and infers the transfection efficiency and / or the cell survival rate. do.
  • the inference device 220 also generates evaluation data (eg, evaluation data 540_1) based on the inferred transfection efficiency and / or cell viability.
  • step S606 the inference device 220 determines whether or not there is the next input data (for example, input data 500_2, 500_3, ).
  • step S606 If it is determined in step S606 that there is the next input data (YES in step S606), the process returns to step S604.
  • step S606 determines whether there is no next input data (NO in step S606). If it is determined in step S606 that there is no next input data (NO in step S606), the transfection efficiency and / or cell viability inference process is terminated.
  • the inference device 220 is -Obtain input data including at least chemical structure information of lipid molecules. -Generated by performing a learning process on a learning model that correlates input data with cell transfection efficiency and / or cell viability of nucleic acids contained in particles containing lipid molecules. Have a trained model. The trained model infers transfection efficiency and / or cell viability associated with newly acquired input data.
  • the inference device 220 according to the first embodiment it is possible to shorten the time required to search for a more suitable chemical structure of the lipid molecule constituting the particles containing the nucleic acid. That is, according to the inference device 220 according to the first embodiment, it is possible to support the work for designing or selecting the chemical structure of the lipid molecule constituting the particle containing the nucleic acid.
  • Example 1 ⁇ Example 1 >> Hereinafter, specific examples of the learning device 210 and the inference device 220 according to the first embodiment will be described. The following examples are merely examples, and the learning device 210 and the inference device 220 according to the first embodiment are not limited to the following examples.
  • the training data set is divided into a training data set and a verification data set, and the training model is trained using the training data set. Generate a model. Further, in this embodiment, the inference accuracy of the inference device 220 was evaluated using the data set for verification.
  • FIG. 7A is a diagram showing an embodiment of the learning device.
  • the preprocessing unit 410 has a conversion unit 710, and the conversion unit 710 reads input data from the data set for learning and converts it into a molecular descriptor.
  • the example of FIG. 7A shows a state in which the chemical structure information of 75 lipid molecules is read out from the chemical structure information 0001 of the lipid molecule to the chemical structure information 0200 of the lipid molecule and converted into 200 molecular descriptors.
  • reference numeral 701 indicates the chemical structure information of one of the 75 lipid molecules read out by the conversion unit 710.
  • the pretreatment unit 410 has an extraction unit 720, and deletes a molecular descriptor that is not suitable for use in the learning process from the 200 molecular descriptors.
  • the extraction unit 720 deletes the molecular descriptor having a small dispersion among the 200 molecular descriptors, and further deletes the molecular descriptor in which colinearity is recognized.
  • the extraction unit 720 extracts a molecular descriptor suitable for use in the learning process.
  • reference numeral 711 is an example of the molecular descriptor extracted by the extraction unit 720, and is input to the learning model 420 as preprocessed data.
  • the learning model 420 uses "Gradient Boosting Decision Tree” as a machine learning algorithm, and it is assumed that hyperparameters are optimized by K-validation cross-validation.
  • the learning model 420 performs the learning process with the transfection efficiency as the correct answer data.
  • the training model 420 is trained so as to associate the pre-processed data (molecular descriptor) with the transfection efficiency, and the trained model 520 (see FIG. 7B) is generated.
  • FIG. 7B is a diagram showing an embodiment of the inference device.
  • the preprocessing unit 510 has a conversion unit 710, and the conversion unit 710 reads input data from the data set for verification and converts it into a molecular descriptor.
  • the conversion unit 710 reads out the chemical structure information of 16 lipid molecules from 0201 to 0216 of the chemical structure information of the lipid molecule and converts it into 200 molecular descriptors. Is shown.
  • reference numeral 801 indicates the chemical structure information of one of the 16 lipid molecules read out by the conversion unit 710.
  • the molecular descriptor converted by the conversion unit 710 is input to the trained model 520 as preprocessed data, and the trained model 520 infers the transfection efficiency.
  • reference numeral 802 is the result of evaluating the inference accuracy of the trained model 520 with respect to the chemical structure information of 16 lipid molecules included in the data set for verification. Specifically, reference numeral 802 indicates a state in which the inference value of the transfection efficiency is compared with the actually measured value, and the inference accuracy is evaluated using the correlation coefficient and the mean absolute error as indexes. As shown by reference numeral 802, in this example, it was confirmed that the trained model 520 can infer the transfection efficiency of the chemical structural information of 16 lipid molecules with high inference accuracy.
  • the learning device 210 is based on various data stored in the data storage unit 170 regarding the drug delivery system.
  • ⁇ Input data including chemical structure information of lipid molecules and • Transfection efficiency and / or cell viability, Described the case of generating a trained model that associates with.
  • the learning device is based on various data stored in the data storage unit 170 regarding the drug delivery system.
  • the second embodiment will be described focusing on the differences from the first embodiment.
  • FIG. 8 is a diagram showing an application example of the inference device according to the second embodiment.
  • design prerequisite 201 (specifically, design) in inferring more appropriate chemical structural information of lipid molecules constituting particles containing nucleic acid 240.
  • Input data including the information included in the precondition 201) of the above is input to the inference device 820.
  • the inference device 820 has a trained model generated by the learning device 810.
  • the learning device 810 generates a trained model by performing a learning process on the learning model using a learning data set generated based on various data stored in the data storage unit 170 related to the drug delivery system. do.
  • the inference device 820 executes the trained model generated by the learning device 810, and infers the chemical structure information 280 of the lipid molecule from the information (including input data) included in the precondition 201 of the design.
  • the lipid molecule 230 is generated based on the chemical structure information 280 inferred by the inference device 820, and the nucleic acid 240 is contained by the particles containing the generated lipid molecule 230. By doing so, particles of the complex 250 are formed.
  • the formed complex 250 particles are applied to 260 target animals and the like, and in vitro (in vitro, in situ or ex vivo) cells and / or tissues 260 and the like. Changes caused by the introduction of the particles of the complex 250 into specific cells in vivo (in vivo) such as the target animal (or in vitro, in situ or ex vivo) to specific cells 260 (Changes caused by the introduction) are measured by various measuring methods and measuring instruments, and are output as effect data 261.
  • the method of applying the particles of the complex 250 to the target animal or the like 260 or to cells and / or tissues in vitro (in vitro, in situ or ex vivo) is the same as that of the above-mentioned [first embodiment]. Can be used.
  • the effect data 261 includes the transfection efficiency and / or the cell viability of the nucleic acid 240 contained in the particles containing the lipid molecule 230, which is calculated from the measurement results. Transfection efficiency and cell viability can be evaluated in the same manner as in [First Embodiment] described above.
  • the effect data 261 exemplifies the transfection efficiency to cells and the cell survival rate, and further, the pharmacokinetics (absorption, distribution) of nucleic acid 240 in vivo when the particles of the complex 250 are applied to 260 of a target animal or the like. , Metabolism) and toxicity data may also be included, but not limited to these.
  • the design or selection of the chemical structure of the lipid molecule depends on the experience and know-how of the designer 110, but according to the inference device 820, from the information contained in the precondition 201 of the design, The chemical structure information 280 of the lipid molecule can be directly inferred.
  • a more suitable chemical structure of the lipid molecule constituting the particle containing the nucleic acid 240 can be designed or selected without depending on the experience and know-how of the designer 110. That is, according to the inference device 820 according to the second embodiment, it is possible to support the work for designing or selecting the chemical structure of the lipid molecule constituting the particle containing the nucleic acid.
  • FIG. 9 is a diagram showing an example of the functional configuration of the learning device according to the second embodiment.
  • the learning data set 900 is an example of a learning data set generated based on various data stored in the data storage unit 170 related to the drug delivery system.
  • the items of each information of the learning data are illustrated below, but the present invention is not limited to these, and a part or all of them can be appropriately used as the learning data.
  • the learning data set 900 has input data and correct answer data, and the input data includes "disease type”, “nucleic acid type”, and “introduction target” as information items. , “Transfection efficiency and / or cell viability”, “attribute information such as target animals” are included.
  • disease A 1 which is the type of the disease
  • nucleic acid type for example, "nucleic acid X 1 ", “nucleic acid X 2 " and the like are stored as the type of nucleic acid corresponding to "disease A 1 ".
  • Examples of the information indicating the target cell referred to here include the origin or the organ or tissue in which the cell is present in the living body, the type of cell (for example, nerve cell, parenchymal cell, stromal cell, etc.) and the like. However, it is not limited to these.
  • the "transfection efficiency and / or cell viability” stores the transfection efficiency and / or cell viability of the nucleic acid contained in the particles containing the lipid molecule into the cells. Specifically, the particles of the corresponding complex are applied to the corresponding target animal, etc. 160 or the cell and / or tissue 160 in vitro (in vitro, in situ or ex vivo), and the applied target animal, etc. The transfection efficiency and / or cell viability calculated from the measurement results obtained by measuring the sample isolated from the sample is stored.
  • These learning data sets may be data obtained from various public information (for example, patent gazettes, papers) or databases.
  • the attribute information of the target animal or the like 160 (for example, the target patient) is stored.
  • the correct answer data includes "chemical structure information of lipid molecules” as an information item.
  • “Chemical structure information of lipid molecule” is designed by the designer 110 in the past based on the information contained in the design prerequisite 101 such as "type of disease”, “type of nucleic acid”, and “target of introduction”. Information indicating the chemical structure of the selected more suitable lipid molecule is stored.
  • a learning program is installed in the learning device 810, and when the learning program is executed, the learning device 810 functions as a preprocessing unit 910, a learning model 920, and a comparison / change unit 930.
  • the pre-processing unit 910 acquires "input data" of the learning data set 900 and performs various pre-processing to generate pre-processing and post-processing data suitable for input to the learning model 920.
  • the various pre-processing performed by the pre-processing unit 910 includes a process of normalizing the input data, a process of vectorizing the input data, and the like.
  • the learning model 920 is a model that associates input data with correct answer data (chemical structure information of lipid molecules). Specifically, the learning model 920 takes the pretreatment and post-treatment data notified from the pretreatment unit 910 as an input, and outputs the chemical structure information of the lipid molecule.
  • a learning process for updating the model parameters is performed by back-propagating the error calculated by the comparison / change unit 930. This will generate a trained model. That is, the trained model is generated by updating the model parameters of the training model 920 so that the output of the training model 920 approaches the correct answer data (chemical structure information of the lipid molecule).
  • the comparison / change unit 930 calculates an error by comparing the chemical structure information of the lipid molecule output from the learning model 920 with the correct answer data (chemical structure information of the lipid molecule) of the learning data set 900. Further, the comparison / change unit 930 updates the model parameters of the learning model 920 by back-propagating the calculated error.
  • FIG. 10 is a diagram showing an example of the functional configuration of the inference device according to the second embodiment.
  • the input data 1000 includes information included in the design prerequisite 201 (disease type, nucleic acid type, introduction target, attribute information of target animal, etc.) and target transfection efficiency. And / or cell viability.
  • the input data may be various public information (for example, patent gazettes, papers) or data obtained from a database.
  • the input data 1000 is input to the inference device 820.
  • An inference program is installed in the inference device 820, and when the inference program is executed, the inference device 820 functions as a preprocessing unit 1010 and a trained model 1020.
  • the pre-processing unit 1010 is another example of the acquisition unit, and has the same function as the pre-processing unit 910 of the learning device 810. Specifically, the pre-processing unit 1010 acquires the input data 1000 and performs various pre-processing on the acquired input data 1000 to generate pre-processing and post-processing data.
  • the trained model 1020 is a trained model generated by performing a learning process by the learning device 810, and receives the pre-processed data notified from the pre-processing unit 1010 as input, and obtains the chemical structure information 280 of the lipid molecule. Infer.
  • the chemical structure information of the lipid molecule can be inferred directly from the information included in the preconditions of the design.
  • a more suitable chemical structure of the lipid molecule constituting the particle containing the nucleic acid 240 can be designed or selected without depending on the experience and know-how of the designer 110. That is, according to the inference device 820 according to the second embodiment, it is possible to support the work for designing or selecting the chemical structure of the lipid molecule constituting the particle containing the nucleic acid.
  • FIG. 11 is an example of a flowchart showing the flow of the chemical structure information inference processing of the lipid molecule.
  • step S1101 the learning device 810 acquires various data from the data storage unit 170 related to the drug delivery system.
  • step S1102 the learning device 810 generates a learning data set 900 using various acquired data.
  • step S1103 the learning device 810 performs learning processing on the learning model 920 using the learning data set 900, and generates a trained model 1020.
  • step S1104 the inference device 820 acquires input data (eg, input data 1000) including the information contained in the design prerequisite 201 and the target transfection efficiency and / or cell viability.
  • input data eg, input data 1000
  • step S1105 the inference device 820 executes the trained model 1020 by inputting the acquired input data (for example, input data 1000) into the trained model 1020, and infers the chemical structure information of the lipid molecule.
  • the acquired input data for example, input data 1000
  • the inference device 820 is -Obtain input data including prerequisites for designing or selecting lipid molecules that make up particles containing nucleic acids. -Generated by performing a learning process on a learning model that associates input data, including prerequisites for designing or selecting lipid molecules that make up particles containing nucleic acids, with chemical structural information of the lipid molecules. Has a trained model. The trained model infers the chemical structural information of the lipid molecule associated with the newly acquired input data.
  • a more suitable chemical structure of the lipid molecule constituting the particles containing nucleic acid can be designed or selected without depending on the experience and know-how of the designer. be able to. That is, according to the inference device 820 according to the second embodiment, it is possible to support the work for designing or selecting the chemical structure of the lipid molecule constituting the particle containing the nucleic acid.
  • -A learning data set is generated based on various data stored in the data storage unit 170 related to the drug delivery system.
  • a trained model is generated by performing training processing using the generated training data set.
  • the chemical structure information of lipid molecules can be inferred.
  • -By encapsulating nucleic acid with particles containing lipid molecules generated based on the inferred chemical structure information particles of the complex are formed.
  • the chemical structure information of the lipid molecule is inferred.
  • the user generates a lipid molecule, and by including the nucleic acid with the particles containing the generated lipid molecule, the user forms the particles of the complex.
  • the third embodiment it becomes possible to search for more suitable lipid molecules constituting particles containing each of the various types of nucleic acids possessed by the user, and to form particles containing nucleic acids. It is possible to accumulate a large amount of know-how for searching for more suitable lipid molecules.
  • the third embodiment will be described focusing on the differences from the first and second embodiments.
  • FIG. 12 is a diagram showing an application example of the inference service providing system according to the third embodiment.
  • the inference service providing system 1210 has an inference device 820 and an information providing device 1211.
  • the inference device 820 is the same as the inference device 820 described with reference to FIGS. 8 and 10 in the second embodiment. Specifically, the inference device 820 executes the trained model generated by the learning device 810 and infers the chemical structure information of the lipid molecule from the input data including the preconditions of the design.
  • the information providing device 1211 generates input data including the acquired design precondition 1221 and the design precondition 1231, respectively, and notifies the inference device 820.
  • the inference device 820 executes the trained model and infers the chemical structure information of the lipid molecule as in the second embodiment.
  • the information providing device 1211 also functions as a charging unit, and charges each user when providing information on the chemical structure of the lipid molecule to each user.
  • the inference service providing system 1210 can receive compensation according to the inference service of the chemical structure information of the lipid molecule.
  • the billing refers to a process of recording the amount of money each user should pay to the inference service providing system 1210.
  • the user 1220 provides the attribute information of the target animal or the like 1225, the type of the disease of the target animal or the like 1225, the attribute of the nucleic acid 1223, the precondition 1221 of the design such as the target for introducing the particles of the complex 1224, and the terminal (not shown). Is transmitted to the inference service providing system 1210 via.
  • the user 1220 is provided with the chemical structure information 1212_1 of the lipid molecule corresponding to the precondition 1221 of the design from the inference service providing system 1210 via a terminal (not shown).
  • the effect data 1226 includes transfection efficiency and / or cell viability of nucleic acid 1223 contained in particles containing the lipid molecule 1222 into cells, which is calculated from the measurement results.
  • the user 1220 registers various data acquired in the series of flow of the drug delivery system review process 1200 in the data storage unit 170 regarding the drug delivery system.
  • ⁇ "Design prerequisite 1” ⁇ "Chemical structure information of lipid molecules 1”, ⁇ Chemical structure information of "nucleic acid 1", ⁇ Chemical structure information of "complex 1", ⁇ "Effect data 1", Etc. are included.
  • the data 1227 regarding the drug delivery system registered by the user 1220 may be data obtained by the user 1220 from various public information (for example, patent gazettes, papers) or databases.
  • the inference service providing system 1210 refunds a part of the consideration received by providing the chemical structure information 1212_1 of the lipid molecule to the user 1220. .. That is, the inference service providing system 1210 changes the billing content for the user 1220.
  • the user 1230 can use the attribute information of the target animal or the like 1235, the type of the disease when the target animal or the like is a patient, the attributes of the nucleic acid 1233 (for example, the type of the nucleic acid 1233, the chemical structure information), and the complex 1234.
  • Prerequisites 1231 for the design of the object to which the particles are introduced are transmitted to the inference service providing system 1210 via a terminal (not shown).
  • the user 1230 is provided with the chemical structure information 1212_2 of the lipid molecule corresponding to the precondition 1231 of the design from the inference service providing system 1210 via a terminal (not shown).
  • the user 1230 applies the formed particles of the complex 1234 to the target animal or the like 1235, or to cells and / or tissues 1235 in vitro (in vitro, in situ or ex vivo).
  • the effect data 1236 includes the transfection efficiency and / or the cell viability of the nucleic acid 1233 contained in the particles containing the lipid molecule 1232, which is calculated from the measurement results.
  • the user 1230 registers various data acquired in the series of flow of the drug delivery system review process 1200 in the data storage unit 170 regarding the drug delivery system.
  • ⁇ "Design prerequisite 2” ⁇ "Chemical structure information of lipid molecules 2”, ⁇ Chemical structure information of "nucleic acid 2”, ⁇ Chemical structure information of "complex 2”, ⁇ "Effect data 2", Etc. are included.
  • the inference service providing system 1210 refunds a part of the consideration received by providing the chemical structure information 1212_2 of the lipid molecule to the user 1230. be able to. That is, in the inference service providing system 1210, the billing content for the user 1230 can be changed.
  • the inference service providing system 1210 provides chemical structure information of the lipid molecule in response to the request from each user, so that more appropriate lipid molecules constituting particles containing each of various types of nucleic acids can be obtained. It will be possible to explore.
  • the data storage unit 170 regarding the drug delivery system can store data regarding the drug delivery system for various types of nucleic acids.
  • the data stored in the data storage unit 170 regarding the drug delivery system may include various public information (for example, patent gazettes, papers) and data available from a database.
  • the trained model can be updated by updating the learning data set using the newly accumulated data regarding the drug delivery system and performing the learning process again.
  • the inference service providing system 1210 according to the third embodiment it is possible to accumulate a large amount of know-how for searching for a more appropriate lipid molecule constituting particles containing nucleic acid. That is, according to the inference service providing system 1210 according to the third embodiment, it is possible to support the work for designing or selecting the chemical structure of the lipid molecule constituting the particles containing the nucleic acid.
  • FIG. 13 is an example of a flowchart showing the flow of the inference service provision process.
  • step S1301 the learning device 810 acquires various data from the data storage unit 170 related to the drug delivery system.
  • step S1302 the learning device 810 generates a learning data set 900 using various acquired data.
  • step S1303 the learning device 810 performs learning processing on the learning model 920 using the learning data set 900, and generates a trained model 1020.
  • step S1304 the inference device 820 of the inference service providing system 1210 acquires the input data generated by the information providing device 1211 based on the preconditions of the design transmitted from the user.
  • step S1305 the inference device 820 of the inference service providing system 1210 executes the trained model 1020 by inputting the acquired input data into the trained model 1020, and infers the chemical structure information of the lipid molecule.
  • step S1306 the information providing device 1211 of the inference service providing system 1210 provides the chemical structure information of the lipid molecule inferred by the inference device 820 to the user who has transmitted the preconditions of the design, and charges the user. conduct.
  • step S1307 when the inference service providing system 1210 collects data on the drug delivery system from the user in response to the information providing device 1211 providing the chemical structure information of the lipid molecule to the user, the user. Part of the consideration can be refunded.
  • step S1308 the inference service providing system 1210 determines whether or not a predetermined amount of data has been collected from the user regarding the data related to the drug delivery system.
  • step S1308 If it is determined in step S1308 that a predetermined amount of data has been collected (YES in step S1308), the process returns to step S1302. In this case, a training data set is generated based on the newly registered predetermined amount of data, and the learning process is performed again.
  • step S1308 determines whether a predetermined amount of data has been collected (NO in step S1308). If it is determined in step S1308 that a predetermined amount of data has not been collected (NO in step S1308), the process proceeds to step S1309.
  • step S1309 the information providing device 1211 of the inference service providing system 1210 determines whether or not to terminate the inference service providing process. If it is determined in step 1309 that the inference service provision process is to be continued (NO in step S1309), the process returns to step S1304.
  • step S1309 if it is determined in step S1309 that the inference service provision process is terminated (YES in step S1309), the inference service provision process is terminated.
  • the inference service providing system 1210 is -Obtain from the user the prerequisites for designing or selecting the lipid molecules that make up the particles containing nucleic acids. -Generated by performing a learning process on a learning model that associates input data, including prerequisites for designing or selecting lipid molecules that make up particles containing nucleic acids, with chemical structural information of the lipid molecules. Has a trained model. -By inputting input data including the preconditions acquired from the user, the chemical structure information of the lipid molecule inferred by the trained model is provided to the user who transmitted the preconditions.
  • the inference service providing system 1210 is -Collect data on drug delivery systems from users in response to providing information on the chemical structure of lipid molecules.
  • the inference service providing system 1210 according to the third embodiment it is possible to accumulate a large amount of know-how for searching for a more appropriate lipid molecule constituting particles containing nucleic acid. That is, according to the inference service providing system 1210 according to the third embodiment, it is possible to support the work for designing or selecting the chemical structure of the lipid molecule constituting the particles containing the nucleic acid.
  • the transfection efficiency and / or the cell viability is obtained from the user in response to the provision of the chemical structure information of the lipid molecule.
  • the reinforcement learning process is performed on the reinforcement learning model by using the reward calculated based on the acquired transfection efficiency and / or the cell survival rate.
  • FIG. 14 is a diagram showing an application example of the inference service providing system according to the fourth embodiment.
  • the inference service providing system 1410 provides each user with information on the chemical structure of the lipid molecule in response to a request from the user 1220, the user 1230, ... Shows the case.
  • the inference service providing system 1410 has an inference device 1420 and an information providing device 1211.
  • the information providing device 1211 has the same function as the information providing device 1211 described with reference to FIG. 12 in the third embodiment. Specifically, when the design preconditions are transmitted from each user, the information providing device 1211 generates input data and notifies the inference device 1420. Further, the information providing device 1211 acquires the chemical structure information of the lipid molecule inferred by the inference device 1420 in response to the notification of the input data, and provides it to the corresponding user.
  • the information providing device 1211 charges each user for providing the chemical structure information of the lipid molecule.
  • the inference service providing system 1410 can receive compensation according to the inference service of the chemical structure information of the lipid molecule.
  • the inference device 1420 acquires effect data from the user in response to the fact that the information providing device 1211 provides the chemical structure information of the lipid molecule to the user. In addition, the inference device 1420 calculates a reward based on the transfection efficiency and / or the cell survival rate included in the acquired effect data, and updates the model parameters of the reinforcement learning model based on the calculated reward.
  • the inference device 1420 executes the reinforcement learning model by inputting the input data newly notified from the information providing device 1211 into the reinforcement learning model with updated model parameters, and newly performs the chemical structure information of the lipid molecule. Infer.
  • the processing performed by the user 1220 and the user 1230 in FIG. 14 is the same as the processing described with reference to FIG. 12 in the third embodiment, and therefore the description thereof will be omitted here.
  • the user 1220 transmits the effect data 1226 to the inference service providing system 1410.
  • the effect data 1226 to be transmitted includes the transfection efficiency and / or the cell viability of the nucleic acid 1223 contained in the particles containing the lipid molecule 1222 into cells.
  • the user 1230 transmits the effect data 1236 to the inference service providing system 1410.
  • the effect data 1236 to be transmitted includes the transfection efficiency and / or the cell viability of the nucleic acid 1233 contained in the particles containing the lipid molecule 1232 into cells.
  • the data transmitted to the inference service providing system may be various public information (for example, patent gazettes, papers) or data that can be obtained from a database.
  • the inference service providing system 1410 can refund a part of the consideration for providing the chemical structure information 1212_1 of the lipid molecule to the user 1220. That is, in the inference service providing system 1410, the billing content for the user 1220 can be changed.
  • the inference service providing system 1410 can refund the user 1230 a part of the consideration for providing the chemical structure information 1212_2 of the lipid molecule. .. That is, in the inference service providing system 1410, the billing content for the user 1230 can be changed.
  • the inference service providing system 1410 performs reinforcement learning processing for the reinforcement learning model by acquiring effect data from each user each time it provides chemical structure information of the lipid molecule in response to a request from each user. conduct.
  • the inference service providing system 1410 According to the fourth embodiment, it becomes possible to perform reinforcement learning processing on a reinforcement learning model for a wide variety of nucleic acids, and with the provision of the inference service. Inference accuracy can be improved.
  • the inference service providing system 1410 according to the fourth embodiment it is possible to accumulate a large amount of know-how for searching for a more appropriate lipid molecule constituting particles containing nucleic acid. That is, according to the inference service providing system 1410 according to the fourth embodiment, it is possible to support the work for designing or selecting a more suitable chemical structure of a lipid molecule constituting particles containing nucleic acid.
  • FIG. 15 is a diagram showing an example of the functional configuration of the inference device.
  • An inference program is installed in the inference device 1420, and when the inference program is executed, the inference device 1420 functions as a preprocessing unit 1510, a reinforcement learning model 1520, and a reward calculation unit 1530.
  • the pre-processing unit 1510 has the same function as the pre-processing unit 910 of the learning device 810, and generates pre-processing and post-processing data by performing various pre-processing on the input data 1501, 1502 and the like.
  • the input data 1501 includes the design prerequisite 1221 transmitted by the user 1220. Further, the input data 1502 includes the precondition 1231 of the design transmitted from the user 1230.
  • the reinforcement learning model 1520 takes the post-treatment data notified from the pre-treatment unit 1510 as an input, and infers the chemical structure information 1212_1, 1212_2, etc. of the lipid molecule.
  • the reward calculation unit 1530 functions as a calculation unit, and calculates the reward based on the transfection efficiency and / or the cell survival rate included in the effect data 1226, 1236, etc. transmitted from the users 1220 and 1230.
  • the reward calculation unit 1530 calculates the reward so that the reward is maximized by increasing the transfection efficiency and / or the cell survival rate.
  • the reward calculation unit 1530 performs reinforcement learning processing for updating the model parameters of the reinforcement learning model 1520 based on the calculated reward.
  • FIG. 16 is another example of a flowchart showing the flow of the inference service provision process.
  • step S1601 the inference device 1420 of the inference service providing system 1410 acquires the input data generated by the information providing device 1211 based on the preconditions of the design transmitted by the user.
  • step S1602 the inference device 1420 of the inference service providing system 1410 executes the reinforcement learning model 1520 by inputting the acquired input data into the reinforcement learning model 1520, and infers the chemical structure information of the lipid molecule.
  • step S1603 the information providing device 1211 of the inference service providing system 1410 provides the chemical structure information of the lipid molecule inferred by the inference device 1420 to the user who has transmitted the preconditions of the design, and charges the user. conduct.
  • step S1604 the inference device 1420 of the inference service providing system 1410 acquires effect data from the user in response to the information providing device 1211 providing the chemical structure information of the lipid molecule to the user. Further, the inference device 1420 of the inference service providing system 1410 refunds a part of the consideration to the user who has transmitted the effect data.
  • step S1605 the inference device 1420 of the inference service providing system 1410 calculates the reward based on the transfection efficiency and / or the cell survival rate included in the acquired effect data.
  • step S1606 the inference device 1420 of the inference service providing system 1410 performs reinforcement learning processing for updating model parameters for the reinforcement learning model based on the calculated reward.
  • step S1607 the information providing device 1211 of the inference service providing system 1410 determines whether or not to terminate the inference service providing process. If it is determined in step 1607 that the inference service provision process is to be continued (NO in step S1607), the process returns to step S1601.
  • step S1607 determines whether the inference service provision process is terminated. If it is determined in step S1607 that the inference service provision process is terminated (YES in step S1607), the inference service provision process is terminated.
  • the inference service providing system 1410 is -Obtain from the user the prerequisites for designing or selecting the lipid molecules that make up the particles containing nucleic acids.
  • -It has a reinforcement learning model that infers the chemical structure information of lipid molecules by inputting input data including the acquired preconditions.
  • do. Perform reinforcement learning processing for the reinforcement learning model based on the calculated reward.
  • the inference accuracy can be improved with the provision of the inference service.
  • a large amount of know-how for searching for a more suitable lipid molecule constituting particles containing nucleic acid can be accumulated. That is, according to the inference service providing system 1410 according to the fourth embodiment, it is possible to support the work for designing or selecting a more suitable chemical structure of a lipid molecule constituting particles containing nucleic acid.
  • the method of applying the trained model 520 is not limited to this, and for example, the inference device is configured to search for the chemical structural information of the lipid molecule satisfying the target transfection efficiency and / or the cell viability. May be good.
  • the fifth embodiment will be described focusing on the differences from the first embodiment.
  • FIG. 17 is a diagram showing an example of the functional configuration of the inference device according to the fifth embodiment.
  • the inference device 1700 functions as a preprocessing unit 510, a trained model 520, and a generation unit 1710.
  • the preprocessing unit 510 and the trained model 520 have already been described with reference to FIG. 5 in the first embodiment, and therefore the description thereof will be omitted here.
  • the generation unit 1710 has, for example, a reinforcement learning function by Thompson Sample. Specifically, the generator 1710 satisfies the predetermined termination condition (eg, the transfection efficiency and / or the transfection efficiency and / or the cell targeted by the transfection efficiency and / or the cell viability inferred by the trained model 520). Whether or not the survival rate is satisfied) is determined.
  • the predetermined termination condition eg, the transfection efficiency and / or the transfection efficiency and / or the cell targeted by the transfection efficiency and / or the cell viability inferred by the trained model 520.
  • the generation unit 1710 determines that the predetermined termination condition is not satisfied, the generation unit 1710 generates the chemical structure information of the lipid molecule based on the transfection efficiency and / or the cell viability inferred by the trained model 520. In addition, the generation unit 1710 notifies the pretreatment unit 510 of the chemical structure information of the generated lipid molecule.
  • the generation unit 1710 determines that the predetermined termination condition is satisfied, the chemical structure information of the previously generated lipid molecule is used as the chemical structure information of the lipid molecule satisfying the target transfection efficiency and / or the cell viability. Output.
  • reference numeral 1730 is an example of chemical structure information of the lipid molecule output from the generation unit 1710.
  • the example of FIG. 17 shows how a specific lipid molecule was generated as chemical structural information of the lipid molecule satisfying the target transfection efficiency and / or cell viability.
  • FIG. 18 is an example of a flowchart showing the flow of the generation process.
  • the generation process shown in FIG. 18 is started by the inference device 1700, it is assumed that the target transfection efficiency and / or the cell survival rate is set in the inference device 1700 in advance.
  • step S1801 the generation unit 1710 generates a group of molecular fragments from chemically formable hydrocarbons in which the maximum values of length, saturation, and number of branches are set.
  • step S1802 the generation unit 1710 combines the generated molecular fragment with the chemical skeleton group of the lipid selected by the designer 110 to generate the chemical structure group of the lipid molecule.
  • the generation unit 1710 has a number of the generated chemical structure group of the lipid molecule specified by the designer 110 according to the combination of the length of the molecular fragment, the degree of saturation, the number of branches, the type of the chemical skeleton, and the like. Divide into multiple search spaces.
  • step S1803 the generation unit 1710 selects one of the search spaces from the plurality of search spaces divided in step S1802 by using Thumbson Sample.
  • selecting one of the search spaces is nothing but selecting the characteristics of the search space (combination of the length, saturation, number of branches, type of chemical skeleton, etc. of the molecular fragment of the lipid molecule). ..
  • step S1804 the generation unit 1710 generates a chemical structure group of the lipid molecule using a combination of the length, saturation, number of branches, type of chemical skeleton, etc. of the selected molecular fragment of the lipid molecule.
  • the generation unit 1710 randomly acquires a plurality of molecular fragments from a search space other than the selected search space with a certain probability, and generates a chemical structure group of lipid molecules together with the selected molecular fragments.
  • step S1805 the generation unit 1710 notifies the pretreatment unit 510 of each chemical structure information of the chemical structure group of the generated lipid molecule.
  • the pretreatment unit 510 performs various pretreatments on each chemical structure information of the lipid molecule notified from the generation unit 1710, so that the pretreatment data group suitable for input to the trained model 520 can be obtained.
  • the trained model 520 infers the transfection efficiency and / or the cell viability for the chemical structure group of the lipid molecule generated by the generator 1710 in step S1804.
  • step S1806 the generation unit 1710 The maximum of multiple transfection efficiencies and / or cell viability inferred by the trained model 520 (inference results).
  • step S1807 the generation unit 1710 determines whether or not a predetermined end condition is satisfied. If it is determined in step S1807 that the predetermined end condition is not satisfied (NO in step S1807), the process returns to step S1803.
  • step S1807 determines whether the predetermined end condition is satisfied (YES in step S1807). If it is determined in step S1807 that the predetermined end condition is satisfied (YES in step S1807), the process proceeds to step S1808.
  • step S1808 the generation unit 1710 uses the previously generated chemical structure information of the lipid molecule (chemical structure information of the lipid molecule whose maximum value is inferred) to satisfy the target transfection efficiency and / or cell viability. It is output as the chemical structure information of.
  • the inference device 1700 is -For a learning model that correlates at least input data containing chemical structural information of lipid molecules with transfection efficiency and / or cell viability of active substances contained in particles containing the lipid molecules. , Has a trained model in which the training process has been performed. • Based on the inference results, the following new transfection efficiencies and / or cell viability associated with input data, including chemical structural information of newly generated lipid molecules, are inferred by the trained model. It has a transfection unit that produces chemical structural information of lipid molecules. -The generation unit repeats the generation process of generating the chemical structure of the next new lipid molecule based on the inference result until the predetermined termination condition is satisfied.
  • the inference device 1700 according to the fifth embodiment for example, it is possible to generate chemical structural information of a lipid molecule that satisfies the target transfection efficiency and / or cell viability. That is, according to the reasoning apparatus 1700 according to the fifth embodiment, it is possible to support the work for designing or selecting the chemical structural information of the lipid molecule constituting the particle containing the nucleic acid.
  • the structure is such that the chemical structure information of the lipid molecule is generated according to the transfection efficiency and / or the cell viability.
  • the information included in the design precondition 201 (disease type, nucleic acid type). , Attribute information such as introduction target, target animal, etc.). Then, in the sixth embodiment, when generating the chemical structure information of the lipid molecule, the information included in the precondition 201 of the design (type of disease, type of nucleic acid, introduction target, attribute information of target animal, etc.) is used. Generates chemical structure information of the corresponding lipid molecule.
  • the sixth embodiment will be described focusing on the differences from the fifth embodiment.
  • FIG. 19 is a diagram showing an example of the functional configuration of the inference device according to the sixth embodiment.
  • the inference device 1900 functions as a preprocessing unit 510, a trained model 520, a generation unit 1910, and an acquisition unit 1930.
  • the preprocessing unit 510 and the trained model 520 have already been described with reference to FIG. 5 in the first embodiment, and therefore the description thereof will be omitted here.
  • the generation unit 1910 has a reinforcement learning function by Thompson Sample, similar to the generation unit 1710 in FIG. Specifically, the generator 1910 obtains the transfection efficiency and / or cell viability output from the trained model 520. In addition, the generation unit 1910 determines whether or not a predetermined termination condition is satisfied (for example, whether or not the acquired transfection efficiency and / or cell viability satisfies the target transfection efficiency and / or cell viability). Is determined. Further, when the generation unit 1910 determines that the predetermined termination condition is not satisfied, the generation unit 1910 generates chemical structure information of the lipid molecule based on the acquired transfection efficiency and / or cell viability, and the chemistry of the generated lipid molecule. Notify the preprocessing unit 510 of the structural information.
  • a predetermined termination condition for example, whether or not the acquired transfection efficiency and / or cell viability satisfies the target transfection efficiency and / or cell viability.
  • the generation unit 1910 the information included in the design precondition 201 (disease type, nucleic acid type, introduction target, target animal, etc.) notified by the acquisition unit 1930 when generating the chemical structure information of the lipid molecule. (Attribute information, etc.) is acquired as a predetermined constraint condition. Then, when the generation unit 1910 generates the chemical structure information of the lipid molecule, the chemical structure information is generated with these acquired information as a predetermined constraint condition.
  • the generation unit 1910 determines that the predetermined termination condition is satisfied, the previously generated chemical structure information of the lipid molecule is used as the chemical structure information of the lipid molecule satisfying the target transfection efficiency and / or the cell viability. Output.
  • the acquisition unit 1930 acquires the input data 1000. Further, the acquisition unit 1930 notifies the generation unit 1910 of the acquired input data 1000.
  • the inference device 1900 is -For a learning model that correlates at least input data containing chemical structural information of lipid molecules with transfection efficiency and / or cell viability of active substances contained in particles containing the lipid molecules. , Has a trained model in which the training process has been performed. • Infer transfection efficiency and / or cell viability associated with input data, including chemical structural information of newly generated lipid molecules, using a trained model. • Generate the following new lipid molecule chemical structural information based on the inferred transfection efficiency and / or cell viability.
  • the information included in the preconditions of the design is set as a predetermined constraint, and the following new chemical structure information of the lipid molecule is generated. Generates the next new lipid molecule chemical structure information based on the inferred transfection efficiency and / or cell viability and the information contained in the design prerequisites until certain termination conditions are met. Repeat the generation process.
  • the chemical structure information of the lipid molecule satisfying the target transfection efficiency and / or the cell viability under a predetermined constraint condition is generated. Can be done. That is, according to the inference device 1900 according to the sixth embodiment, it is possible to support the work for designing or selecting the chemical structure of the lipid molecule constituting the particle containing the nucleic acid.
  • the learning device and the inference device have been described as being configured as separate devices. However, the learning device and the inference device may be configured as an integrated device.
  • the inference service providing system 1210 has been described as having the inference device 820 and the information providing device 1211 configured as separate devices.
  • the inference device 820 and the information providing device 1211 may be configured as an integrated device.
  • the function of the inference device 820 and the function of the information providing device 1211 may be realized by, for example, executing the inference service providing program in the integrated device.
  • the inference service providing system 1410 has been described as having the inference device 1420 and the information providing device 1211 configured as separate devices.
  • the inference device 1420 and the information providing device 1211 may be configured as an integrated device.
  • the function of the inference device 1420 and the function of the information providing device 1211 may be realized by, for example, executing the inference service providing program in the integrated device.
  • the information providing device generates input data.
  • the generation of input data may be configured to be performed by, for example, an inference device.
  • the information providing device has been described as functioning as an acquisition unit, a provision unit, and a billing unit. However, some of the functions realized by the information providing device may be realized on the user's terminal (or on the cloud).
  • the details of the method of providing the chemical structure information of the lipid molecule to the user by the information providing device are not mentioned, but the method of providing the information by the information providing device is arbitrary.
  • the information providing device may be configured to directly transmit the chemical structure information of the lipid molecule to the user, or the storage location where the chemical structure information of the lipid molecule can be accessed by the user by inputting a password or the like. It may be configured to be stored in.
  • the generation unit uses the transfection efficiency and / or the cell viability inferred by the trained model 520 as the target transfection efficiency and / or the target transfection efficiency as a predetermined termination condition. Alternatively, it was determined whether or not the cell viability was satisfied.
  • the predetermined termination condition is not limited to this, and for example, it may be determined whether or not the chemical structure information of the lipid molecule to be produced has been updated. In this case, the generation unit determines that a predetermined termination condition is satisfied when it is determined that the chemical structure information of the lipid molecule to be produced has not been updated.
  • information items such as FIGS. 5, 10, and 15 are exemplified as input data information items, but the input data information items are not limited to these.
  • Drug delivery system review process 170 Data storage unit for drug delivery system 200: Drug delivery system review process 210: Learning device 220: Inference device 400: Learning data set 520: Trained model 800: Drug delivery system review process 810 : Learning device 820: Inference device 900: Learning data set 1000: Input data 1020: Trained model 1200: Drug delivery system review process 1211: Information provision device 1400: Drug delivery system review process 1410: Inference service provision system 1420: Inference Device 1520: Reinforcement learning model 1530: Reward calculation unit 1700: Inference device 1710: Generation unit 1900: Inference device 1910: Generation unit 1930: Acquisition unit

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Chemical & Material Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Databases & Information Systems (AREA)
  • Molecular Biology (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
  • Investigating Or Analysing Biological Materials (AREA)
PCT/JP2021/031711 2020-08-31 2021-08-30 推論装置、推論方法、推論プログラム、モデル生成方法、推論サービス提供システム、推論サービス提供方法及び推論サービス提供プログラム Ceased WO2022045334A1 (ja)

Priority Applications (4)

Application Number Priority Date Filing Date Title
CN202180052991.6A CN115989511A (zh) 2020-08-31 2021-08-30 推理装置、推理方法、推理程序、模型生成方法、推理服务提供系统、推理服务提供方法以及推理服务提供程序
EP21861755.3A EP4207205A4 (en) 2020-08-31 2021-08-30 Inference device, inference method, inference program, model generating method, inference service providing system, inference service providing method, and inference service providing program
JP2022545756A JP7735624B2 (ja) 2020-08-31 2021-08-30 推論装置、推論方法、推論プログラム、モデル生成方法、推論サービス提供システム、推論サービス提供方法及び推論サービス提供プログラム
US18/022,491 US20240038340A1 (en) 2020-08-31 2021-08-30 Inference device, inference method, inference program, model generating method, inference service providing system, inference service providing method, and inference service providing program

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2020-146402 2020-08-31
JP2020146402 2020-08-31

Publications (1)

Publication Number Publication Date
WO2022045334A1 true WO2022045334A1 (ja) 2022-03-03

Family

ID=80353466

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2021/031711 Ceased WO2022045334A1 (ja) 2020-08-31 2021-08-30 推論装置、推論方法、推論プログラム、モデル生成方法、推論サービス提供システム、推論サービス提供方法及び推論サービス提供プログラム

Country Status (5)

Country Link
US (1) US20240038340A1 (https=)
EP (1) EP4207205A4 (https=)
JP (1) JP7735624B2 (https=)
CN (1) CN115989511A (https=)
WO (1) WO2022045334A1 (https=)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023190974A1 (ja) * 2022-03-31 2023-10-05 武田薬品工業株式会社 遺伝子導入率予測方法

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US12587274B2 (en) 2023-03-28 2026-03-24 Quantum Generative Materials Llc Satellite optimization management system based on natural language input and artificial intelligence
KR102763227B1 (ko) * 2023-09-26 2025-02-07 주식회사 그래디언트바이오컨버전스 최적의 줄기 세포 배양배지 조합 결정 장치, 배양배지 조합들에 대한 분화 성공률을 산출하여 최적의 조합을 결정하는 방법, 및 컴퓨터 프로그램
KR102763225B1 (ko) * 2023-09-26 2025-02-07 주식회사 그래디언트바이오컨버전스 최적의 오가노이드 배양배지 조합 결정 장치, 배양배지 조합들에 대한 생존율을 산출하여 최적의 조합을 결정하는 방법, 및 컴퓨터 프로그램
KR102753534B1 (ko) * 2023-10-24 2025-01-14 주식회사 그래디언트바이오컨버전스 최적의 줄기 세포 배양배지 조합 결정 장치, 배양배지 조합들에 대한 유전자 발현량을 산출하여 최적의 조합을 결정하는 방법, 및 컴퓨터 프로그램
US12368503B2 (en) 2023-12-27 2025-07-22 Quantum Generative Materials Llc Intent-based satellite transmit management based on preexisting historical location and machine learning
US12603701B2 (en) 2023-12-27 2026-04-14 Quantum Generative Materials Llc Distributed satellite constellation management and control system

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005530771A (ja) * 2002-05-08 2005-10-13 ユニバーシティ カレッジ ロンドン 生物学的活性物質の細胞への運搬に適した複合体
WO2016021683A1 (ja) 2014-08-07 2016-02-11 武田薬品工業株式会社 カチオン性脂質
JP2019515913A (ja) * 2016-04-22 2019-06-13 科済生物医薬（上海）有限公司Ｃａｒｓｇｅｎ Ｔｈｅｒａｐｅｕｔｉｃｓ Ｌｉｍｉｔｅｄ 細胞免疫療法の組成物および方法
WO2019131839A1 (ja) 2017-12-28 2019-07-04 武田薬品工業株式会社 カチオン性脂質
WO2020032184A1 (ja) 2018-08-10 2020-02-13 武田薬品工業株式会社 カチオン性脂質
WO2020031671A1 (ja) * 2018-08-08 2020-02-13 パナソニックＩｐマネジメント株式会社 材料記述子生成方法、材料記述子生成装置、材料記述子生成プログラム、予測モデル構築方法、予測モデル構築装置及び予測モデル構築プログラム
JP2020146402A (ja) 2019-03-15 2020-09-17 トヨタホーム株式会社 宅配ボックスシステム

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4370591B2 (ja) * 2001-07-31 2009-11-25 独立行政法人産業技術総合研究所 リポソームおよびそれからなる細胞への物質導入担体
US7598421B2 (en) * 2002-05-08 2009-10-06 Ucl Biomedica Plc Materials for the delivery of biologically-active material to cells
WO2012119129A1 (en) * 2011-03-02 2012-09-07 Berg Biosystems, Llc Interrogatory cell-based assays and uses thereof
MX375673B (es) * 2012-04-02 2025-03-06 Berg Llc Ensayos basados en interrogatorios celulares y usos de los mismos.
WO2015138309A1 (en) * 2014-03-13 2015-09-17 Dana-Farber Cancer Institute, Inc. Compositions and methods for regulating pancreatic beta cell function using adipsin
CN104561067A (zh) * 2014-12-24 2015-04-29 广东省人民医院 一种安全高效的蛋白质转染进入细胞的载体
CN110418636A (zh) * 2016-08-23 2019-11-05 杰罗米.J.申塔格 将肽、蛋白质和核酸负载至乳糜微粒和体细胞中的胆固醇酯囊泡
WO2019010384A1 (en) * 2017-07-07 2019-01-10 The Broad Institute, Inc. METHODS FOR DESIGNING GUIDE SEQUENCES FOR GUIDED NUCLEASES

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005530771A (ja) * 2002-05-08 2005-10-13 ユニバーシティ カレッジ ロンドン 生物学的活性物質の細胞への運搬に適した複合体
WO2016021683A1 (ja) 2014-08-07 2016-02-11 武田薬品工業株式会社 カチオン性脂質
JP2019515913A (ja) * 2016-04-22 2019-06-13 科済生物医薬（上海）有限公司Ｃａｒｓｇｅｎ Ｔｈｅｒａｐｅｕｔｉｃｓ Ｌｉｍｉｔｅｄ 細胞免疫療法の組成物および方法
WO2019131839A1 (ja) 2017-12-28 2019-07-04 武田薬品工業株式会社 カチオン性脂質
WO2020031671A1 (ja) * 2018-08-08 2020-02-13 パナソニックＩｐマネジメント株式会社 材料記述子生成方法、材料記述子生成装置、材料記述子生成プログラム、予測モデル構築方法、予測モデル構築装置及び予測モデル構築プログラム
WO2020032184A1 (ja) 2018-08-10 2020-02-13 武田薬品工業株式会社 カチオン性脂質
JP2020146402A (ja) 2019-03-15 2020-09-17 トヨタホーム株式会社 宅配ボックスシステム

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4207205A4

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023190974A1 (ja) * 2022-03-31 2023-10-05 武田薬品工業株式会社 遺伝子導入率予測方法

Also Published As

Publication number Publication date
JPWO2022045334A1 (https=) 2022-03-03
CN115989511A (zh) 2023-04-18
JP7735624B2 (ja) 2025-09-09
EP4207205A4 (en) 2025-02-26
EP4207205A1 (en) 2023-07-05
US20240038340A1 (en) 2024-02-01

Similar Documents

Publication Publication Date Title
WO2022045334A1 (ja) 推論装置、推論方法、推論プログラム、モデル生成方法、推論サービス提供システム、推論サービス提供方法及び推論サービス提供プログラム
Kalmbach et al. Signature morpho-electric, transcriptomic, and dendritic properties of human layer 5 neocortical pyramidal neurons
Han et al. Integrating spatial and single-cell transcriptomics to characterize the molecular and cellular architecture of the ischemic mouse brain
BRAIN Initiative Cell Census Network (BICCN) BRAIN Initiative Cell Census Network (BICCN) Corresponding authors Callaway Edward M. CALLAWAY@ SALK. EDU 1 b Dong Hong-Wei HongWeiD@ mednet. ucla. edu 2 93 c Ecker Joseph R. ecker@ salk. edu 3 4 d Hawrylycz Michael J. MikeH@ alleninstitute. org 5 e Huang Z. Josh josh. huang@ duke. edu 6 7 f Lein Ed S. EdL@ alleninstitute. org 5 g Ngai John john. ngai@ nih. gov 8 9 94 h 94 Osten Pavel osten@ cshl. org 6 j Ren Bing biren@ ucsd. edu 10 11 k Tolias Andreas Savas astolias@ bcm. edu 12 13 m White Owen Owhite@ som. umaryland. edu 14 n http://orcid. org/0000-0002-0326-5878 Zeng Hongkui hongkuiz@ alleninstitute. org 5 p Zhuang Xiaowei zhuang@ chemistry. harvard. edu 15 q et al. A multimodal cell census and atlas of the mammalian primary motor cortex
Sun et al. Spatial transcriptomic clocks reveal cell proximity effects in brain ageing
Wang et al. Hippocampus–prefrontal coupling regulates recognition memory for novelty discrimination
Mollova et al. Cardiomyocyte proliferation contributes to heart growth in young humans
Lennon et al. Dysferlin interacts with annexins A1 and A2 and mediates sarcolemmal wound-healing
Zapala et al. Adult mouse brain gene expression patterns bear an embryologic imprint
Nakamura et al. The rostral raphe pallidus nucleus mediates pyrogenic transmission from the preoptic area
Li et al. Fetal tissue containing the suprachiasmatic nucleus restores multiple circadian rhythms in old rats
JP2022520009A (ja) Ｎｍｄａ、グリシンおよびａｍｐａ受容体のモジュレーターの薬理ゲノミクス決定支援
Ellegood et al. Neuroanatomical assessment of the integrin β3 mouse model related to autism and the serotonin system using high resolution MRI
Bouaziz et al. How artificial intelligence can improve our understanding of the genes associated with endometriosis: natural language processing of the PubMed Database
Almehmadi et al. Increased expression of miR‐155p5 in amygdala of children with autism spectrum disorder
Sharma et al. Temporal profile of clinical signs and histopathologic changes in an F-344 rat model of kainic acid–induced mesial temporal lobe epilepsy
Ahmed et al. Endothelial eNAMPT drives EndMT and preclinical PH: rescue by an eNAMPT-neutralizing mAb
Biliouris et al. A semi‐mechanistic population pharmacokinetic model of nusinersen: an antisense oligonucleotide for the treatment of spinal muscular atrophy
CN107194203A (zh) 基于miRNA数据和组织特异性网络的药物重定位方法
US20230238086A1 (en) Machine-learned pharmacology optimization
CN108271422A (zh) 基于器官间串扰系统的预测装置及预测程序
Qu et al. Scientific review of the proarrhythmic risks of oligonucleotide therapeutics: are dedicated ICH S7B/E14 studies needed for low‐risk modalities?
JP6737519B1 (ja) プログラム、学習モデル、情報処理装置、情報処理方法および学習モデルの生成方法
Eder et al. Exercise-induced CITED4 expression is necessary for regional remodeling of cardiac microstructural tissue helicity
Hamati et al. 65 years of research on dopamine's role in classical fear conditioning and extinction: A systematic review

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: 21861755

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022545756

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2021861755

Country of ref document: EP

Effective date: 20230331

WWW Wipo information: withdrawn in national office

Ref document number: 2021861755

Country of ref document: EP