RETROSYNTHETIC ANALYSIS SYSTEM, RETROSYNTHETIC ANALYSIS APPARATUS, RETROSYNTHETIC ANALYSIS METHOD, AND RETROSYNTHETIC ANALYSIS PROGRAM

Abstract

The present invention provides a retrosynthetic analysis system, a retrosynthetic analysis device, a retrosynthetic analysis method, and a retrosynthetic analysis program that can efficiently search for a synthetic route for a target compound. [Solution] A retrosynthetic analysis system comprising an input information receiving unit that receives a target compound to be analyzed, a preprocessing unit that uses the target compound and / or an intermediate compound of the target compound as a base compound, performs preprocessing to simplify part of the molecular structure of the base compound, and sets a tentative structural compound, and an analysis unit that performs retrosynthetic analysis on the tentative structural compound, and determines a synthetic route for the base compound.

Description

[Technical field]

[0001] The present disclosure relates to a retrosynthetic analysis system, a retrosynthetic analysis device, a retrosynthetic analysis method, and a retrosynthetic analysis program for analyzing synthetic routes of compounds. [Background technology]

[0002] As shown in Patent Document 1, a method for planning a synthetic route for a target compound by retrosynthetic analysis is known. In retrosynthetic analysis, a target compound is used as a starting point, and bonds contained in the molecular structure of the target compound are cut to decompose the target compound into precursors. By repeating such decomposition steps into precursors until the resulting precursors become available compounds such as over-the-counter drugs, a multi-step synthetic route for obtaining the target compound is planned. In an information processing system that performs such retrosynthetic analysis, a machine learning model such as a neural network that has been trained on known chemical reactions and compounds, and a route search algorithm using reinforcement learning are used. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Special Publication No. 2022-526697 Summary of the Invention [Problem to be solved by the invention]

[0004] Conventional retrosynthetic analysis techniques had the problem that they would output chemical reactions that were impossible to carry out or synthetic routes that were unnecessarily complicated during the step-by-step decomposition of the target compound.

[0005] For example, retrosynthetic analysis techniques generally employ one-step processing (a process in which, for one reaction point, a cycle of identifying the chemical reaction that produces the reaction point and the compound before the reaction is repeated), and are limited in that only one change in a reaction point can be detected per step. In actual chemical reactions, multiple reaction points may be generated in one chemical reaction (for example, multiple substituents with the same structure may be added to a precursor in one chemical reaction), but retrosynthetic analysis techniques using one-step processing are unable to detect such chemical reactions and may output a synthetic route that is more complicated than necessary.

[0006] Furthermore, when determining a synthetic route for a compound having a complex structure, such as a compound with a large molecular weight or a compound with a cyclic structure, it is necessary to consider various factors that may affect a chemical reaction, such as the reactivity of substituents, steric hindrance, competitive reactions, etc. However, conventional retrosynthetic analysis techniques tend to give priority to well-known chemical reactions or frequently performed chemical reactions, and may output chemical reactions that are not suitable for the structure of the target compound or that cannot be performed.

[0007] An object of the exemplary embodiments of the present disclosure is to provide a retrosynthetic analysis system, a retrosynthetic analysis device, a retrosynthetic analysis method, and a retrosynthetic analysis program that can efficiently search for a synthetic route for a target compound. [Means for solving the problem]

[0008] A retrosynthetic analysis system according to one embodiment of the present disclosure includes: an input information receiving unit that receives a target compound to be analyzed; a pre-processing unit that performs a pre-processing for simplifying a part of a molecular structure of the base compound using the target compound and / or an intermediate compound of the target compound as a base compound to set a tentative structure compound; and an analysis unit that performs a retrosynthetic analysis on the tentative structural compound and determines a synthetic route for the base compound.

[0009] Since the retrosynthetic analysis system has the above-mentioned features, it is possible to improve the efficiency of synthetic route searching in retrosynthetic analysis. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram illustrating an example of the configuration of a retrosynthetic analysis system according to an embodiment of the present disclosure. [Diagram 2] FIG. 2 is a block diagram illustrating a hardware configuration of the management server illustrated in FIG. [Diagram 3] FIG. 3 is a block diagram illustrating an example of a software configuration of the management server illustrated in FIG. [Figure 4] FIG. 4 is a diagram for explaining an embodiment of the preprocessing. [Diagram 5] FIG. 5 is a diagram for explaining another embodiment of the pretreatment. [Figure 6] FIG. 6 is a diagram for explaining another embodiment of the pretreatment. [Figure 7] FIG. 7 is a diagram showing a simulation of the analysis result (comparative example) in the case where retrosynthetic analysis was performed without pretreatment. [Figure 8] FIG. 8 is a diagram showing an example of a result of the retrosynthesis analysis. [Figure 9] FIG. 9 is a diagram showing a simulation of the analysis result (comparative example) when retrosynthetic analysis is performed without pretreatment. [Figure 10] FIG. 10 is a diagram showing another example of the results of the retrosynthesis analysis. [Figure 11] FIG. 11 is a diagram showing a simulation of the analysis result (comparative example) in the case where retrosynthetic analysis was performed without pretreatment. [Figure 12] FIG. 12 is a diagram showing another example of the results of the retrosynthesis analysis. [Figure 13] FIG. 13 is a flowchart illustrating a retrosynthetic analysis method according to this embodiment. [Figure 14] FIG. 14 is a flowchart illustrating a retrosynthetic analysis method according to this embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] A retrosynthetic analysis system according to an embodiment of the present disclosure will be described with reference to the drawings. In each of the attached drawings, identical or similar elements are given identical or similar reference symbols and names, and duplicate descriptions of identical or similar elements may be omitted in the description of the embodiment. Note that the contents shown in each drawing are merely examples for explaining the present embodiment, and are merely examples that are shown in a schematic manner to make the present embodiment easier to explain. The contents of each drawing may be modified or changed within the scope of no technical problems.

[0012] <System Overview> The retrosynthetic analysis system according to this embodiment is an information processing system that accepts input of a target compound, which is a final product, from a user, performs retrosynthetic analysis on the target compound, and plans a synthetic route from compounds available to the user, such as over-the-counter drugs, to the target compound.

[0013] <Outline of retrosynthetic analysis> The target compound to be analyzed in the retrosynthetic analysis system is mainly an organic compound. In this system, the retrosynthetic analysis is performed by a machine learning model constructed by a predetermined neural network such as a CNN (Convolutional Neural Network) or an RNN (Recurrent Neural Network), and / or a search algorithm using reinforcement learning such as a Monte Carlo tree search. The architecture of the neural network employed in this system is not particularly limited, and machine learning models other than neural networks, such as a linear model or a tree model, may be employed. In addition, the learning data for the machine learning model for retrosynthetic analysis is also not particularly limited, and for example, a compound database, a chemical reaction database, or learning data with teacher labels designed based on these databases may be used. The machine learning model employed in this system may be configured as a model that accepts structural information of a compound as an input and predicts a chemical reaction and a reactive compound (a compound used in a reaction). In addition, the machine learning model may be configured as a single model, or may be configured by combining multiple models in charge of each prediction. Note that the prediction of a chemical reaction and a reactive compound may be performed by classical arithmetic processing that does not use machine learning, but in this embodiment, it is assumed that a machine learning model is used. In addition, the search algorithm used in this system performs route search based on a reward function such as the number of synthesis steps using chemical reactions and reaction compounds predicted by the machine learning model. Note that the search algorithm may employ active learning such as reinforcement learning and Bayesian optimization other than Monte Carlo tree search, or may employ other classical search algorithms. In addition, the reward function for search is not particularly limited.

[0014] In the retrosynthetic analysis system of this embodiment, in retrosynthetic analysis, first, input of a target compound is accepted from a user. In retrosynthetic analysis, a chemical reaction and a reaction compound are determined by a machine learning model and a search algorithm as one step, and the step is repeated until all intermediate compounds used to generate the target compound are available compounds such as commercially available drugs, thereby planning a series of synthetic routes for synthesizing the target compound. The algorithm for retrosynthetic analysis is not particularly limited, and a known algorithm can be adopted.

[0015] For example, a machine learning model for retrosynthetic analysis refers to a compound database, a chemical reaction database, etc., and searches for retrons (the smallest partial structure that can be synthesized by a specified chemical reaction) contained in a target compound, thereby determining the bond cleavage site (i.e., chemical reaction) and compounds separated by bond cleavage (i.e., compounds before reaction such as raw material compounds and intermediate compounds (precursors)). Note that a molecular fragment generated by bond cleavage is called a synthon (synthetic equivalent). Since a synthon is merely a fragment and cannot actually exist, in retrosynthetic analysis, a reactive compound that will be a source of synthons during reaction is determined from the compound database. In this embodiment, a compound (reactant) corresponding to a synthon determined by retrosynthetic analysis may be called a "synthon compound".

[0016] From the viewpoint of outputting an optimal synthetic route, it is preferable to use a search algorithm constructed by reinforcement learning or the like together with a machine learning model. The search algorithm for retrosynthesis analysis refers to the output of the machine learning model to determine candidates for chemical reactions for generating the target compound. For example, the search algorithm determines candidates for chemical reactions for generating the reactive compounds (i.e., steps prior to the target step on the synthetic route) for each candidate chemical reaction and reactive compound of the target step output by the machine learning model based on the machine learning model. By repeating a series of processes multiple times, synthetic route candidates for the target compound are calculated. Then, the search algorithm evaluates those routes using a reward function, and determines the chemical reaction and reactive compound that can output a synthetic route leading to a higher reward as the chemical reaction at the target step. The search algorithm determines the entire synthetic route by repeating the above-mentioned processes.

[0017] After determining a chemical reaction of one step by the machine learning model and / or search algorithm as described above, if any one of the intermediate compounds (synthon compounds or reaction compounds) is difficult to obtain as a raw material, the intermediate compound is used as input data for the retrosynthetic analysis algorithm, and the next step of the retrosynthetic analysis is performed to search for a chemical reaction that produces the intermediate compound. In this embodiment, the target compound and the intermediate compound that are input data to the machine learning model at each step of the retrosynthetic analysis are referred to as "base compounds". In addition, the intermediate compound that is available as a raw material is referred to as "registered compound". The registered compound may be a compound registered as a commercially available drug (commercially available compound), or may be a compound registered as a compound independently managed or owned by a user even if it is not a commercially available compound (user-registered compound), or may be a compound (inventory compound) that is available in stock among the commercially available compounds and user-registered compounds.

[0018] In retrosynthetic analysis, if all intermediate compounds determined in a given step are registered compounds, it is judged that the "search for synthetic pathways was successful." Then, with the step being the final step of retrosynthesis (the step that is the first chemical reaction in synthesis), a "retrosynthetic tree" showing a series of chemical reactions from the target compound to the final step is output. If the final step (the step in which all intermediate compounds are registered compounds) as described above cannot be found, the steps of retrosynthesis will continue to be repeated, but if the number of steps increases too much, it cannot be considered a practical synthetic pathway. Therefore, in retrosynthetic analysis, a predetermined upper limit (search limiting condition) is set for the search time, number of search steps, cost, etc. as analysis conditions, and if the final step cannot be found before the condition is reached, it is judged that the search for synthetic pathways has failed and the calculation process of the retrosynthetic analysis is terminated.

[0019] In retrosynthetic analysis, it is necessary to plan an efficient synthetic route with fewer steps. However, in conventional retrosynthetic analysis systems, the pre-reaction structure is identified one by one for each reactive point contained in the structure of the target compound in a one-step process, and therefore a complex synthetic route may be output for a target compound that should be a simple synthetic route. In addition, when searching for chemical reactions, an algorithm is used that prioritizes well-known chemical reactions, chemical reactions that are frequently performed, etc. Although the operation of setting such prioritized chemical reactions can be useful in improving the efficiency of searching for chemical reactions, there is also a problem that a chemical reaction that does not match the structure of the target compound or a chemical reaction that cannot be performed may be output.

[0020] In the retrosynthetic analysis system of the present embodiment, a preprocessing is performed on the base compound to simplify the molecular structure of the base compound, thereby realizing the planning of a synthetic route more efficiently than in the past. The details of the retrosynthetic analysis system of the present embodiment will be described below.

[0021] <Components of the retrosynthetic analysis system> As shown in FIG. 1, the retrosynthesis analysis system of this embodiment includes a management server 1 and one or more user terminals 2. The management server 1 and the user terminals 2 are communicatively connected via a network NW. In this embodiment, the network NW is mainly assumed to be the Internet, but the network NW is not limited to the Internet and may be constructed, for example, by a public telephone line network, a mobile phone line network, a wireless communication network, Ethernet (registered trademark), or the like. Note that the configuration of the retrosynthesis analysis system shown in FIG. 1 is an example, and a certain configuration may be combined with other configurations or may include other configurations.

[0022] The management server 1 is an information processing device (retrosynthesis analysis device) that executes processing related to retrosynthesis analysis, and may be managed by a provider that provides a retrosynthesis analysis system. The management server 1 may be a general-purpose computer such as a workstation or a personal computer, or may be logically realized by cloud computing. As shown in FIG. 2, the management server 1 includes at least a processor 10, a memory 11, a storage 12, a transmission / reception unit 13, an input / output unit 14, etc., which are electrically connected to each other through a bus 15. Note that the illustrated configuration is an example, and the management server 1 may have other configurations.

[0023] The processor 10 is a computing device that controls the operation of the entire management server 1, controls the transmission and reception of data between each element, and performs information processing required for application execution and authentication processing. For example, the processor 10 is a CPU (Central Processing Unit) and / or a GPU (Graphics Processing Unit), and executes programs stored in the storage 12 and deployed in the memory 11 to perform each information processing.

[0024] The memory 11 includes a main memory configured with a volatile storage device such as a DRAM (Dynamic Random Access Memory), and an auxiliary memory configured with a non-volatile storage device such as a flash memory, an HDD (Hard Disc Drive), etc. The memory 11 is used as a work area for the processor 10, and also stores a BIOS (Basic Input / Output System) that is executed when the management server 1 is started, various setting information, and the like.

[0025] The storage 12 stores various programs such as application programs. A database storing data used in each process may be constructed in the storage 12. For example, a storage unit 120 described later may be provided in a part of the storage area of ​​the memory 11 and / or the storage 12.

[0026] The transmission / reception unit 13 is a communication interface for the management server 1 to communicate with the user terminal 2, etc., via a communication network. The transmission / reception unit 13 may further include a short-range communication interface such as Bluetooth (registered trademark) and BLE (Bluetooth Low Energy) and / or a USB (Universal Serial Bus) terminal, etc.

[0027] The input / output unit 14 includes information input devices such as a keyboard and a mouse, and output devices such as a display.

[0028] A bus 15 is commonly connected to each of the above elements, and transmits, for example, address signals, data signals, and various control signals.

[0029] The user terminal 2 is, for example, a terminal used by a user of an organization conducting research and development of a target substance. In the retrosynthetic analysis system, the user operates the user terminal 2 to perform various input operations such as specifying a target compound and setting conditions for retrosynthetic analysis. The user terminal 2 may be, for example, a general-purpose computer such as a workstation or a personal computer, or may be a mobile terminal such as a smartphone or a tablet.

[0030] The user terminal 2 also includes a processor, memory, storage, a transmission / reception unit, an input / output unit, etc., which are electrically connected to each other via a bus. The hardware configuration of the user terminal 2 can be configured in the same manner as the above-mentioned management server 1, and detailed description of each element in the hardware configuration of the user terminal 2 will be omitted.

[0031] <Management Server 1 Functions> 3 is a block diagram illustrating functions implemented in the management server 1. In this embodiment, the management server 1 may include an input information receiving unit 101, a preprocessing unit 102, an analysis unit 103, and a display control unit 104. The storage unit 120 of the management server 1 may include various databases such as a compound DB (database) 121, a chemical reaction DB 122, an analysis information storage unit 123, and a specific structure DB 124. Note that the various functional units illustrated in FIG. 3 are exemplified as functional units realized by the processor 10 of the management server 1, but some or all of the various functional units may be realized in the processor of the user terminal 2.

[0032] The compound DB121 is a database that consolidates information about compounds, which is referred to during retrosynthetic analysis. The data of each compound recorded in the compound DB121 is not necessarily limited. For example, the compound DB121 may include basic information such as the substance name, structural formula, rational formula, and CAS registration number of each compound, as well as physical property data (melting point, boiling point, solubility, etc.), safety data (toxicity, flammability, handling precautions, etc.), and information for obtaining (manufacturer, product number, price, stock status, etc.). The rational formula of each compound may be stored in a machine-readable format such as SMILES format, InChI format, or MOL file format.

[0033] Compounds registered in the compound DB121 may be managed in two categories. For example, the first category is compounds available as synthetic raw materials for over-the-counter drugs and the like (hereinafter, raw material compounds), and the second category is compounds manufactured by a predetermined synthesis such as intermediate compounds, final compounds, and the like. Raw material compounds registered in the first category are not limited to commercially available compounds, and compounds managed or owned by users may be registered. Information on raw material compounds registered in the first category may include information on the presence or absence of stock, stock quantity, price, storage location, source of acquisition, manufacturer, and the like, and information on compounds registered in the second category may include information on a synthesis method, production cost for synthesis, and the like.

[0034] The compound DB121 may be constructed using a publicly known database, may be constructed by adding and registering information on compounds managed and held by a user to a public database, or may adopt a database that the user independently sets (and appropriately updates) as the compound DB121. In addition, the compound DB121 may be expanded by adding new target compounds input by the user for retrosynthesis analysis, intermediate compounds output by retrosynthesis analysis, and the like from time to time (that is, the information on the compound DB121 may be updated based on past retrosynthesis analysis results). In addition, the compound DB121 may be updated based on a public database provided by a reagent manufacturer or the like. When updating the compound DB121, information required for updating, such as inventory status, may be automatically collected from the website of the reagent manufacturer or other public databases using a technology such as web scraping. Database updates, such as addition, change, and correction of compounds to the compound DB121, may be performed by the operation of the system provider.

[0035] The chemical reaction DB122 is a database that aggregates information on chemical reactions that is referred to during retrosynthesis analysis. The chemical reactions aggregated in the chemical reaction DB are not limited to known chemical reactions, and may include new chemical reactions generated by information technology such as machine learning models. The data of each chemical reaction recorded in the chemical reaction DB122 is not necessarily limited. For example, the chemical reaction DB122 may include information such as the name of the chemical reaction, the reaction formula, information identifying the reactants (raw material compounds) and products (information identifying the compound names, molecular structures, etc.), reaction conditions (temperature, pressure, type of solvent, reaction time, type and amount of catalyst, etc.), yield, by-products, cost of the chemical reaction, and other information related to the reaction (for example, information related to the manufacture such as the ratio of raw materials, the order of addition, and post-treatment). In the chemical reaction DB122, the reaction formula, the reactants, the products, etc. may also be stored in a machine-readable format.

[0036] The chemical reaction DB122 may be constructed using a public database, similar to the compound DB121, and may be constructed by registering additional chemical reactions known to the user in such a public database, or by performing proofreading such as changing, deleting, or correcting the contents of the database. The chemical reaction DB122 may be updated by collecting information on new chemical reactions from academic paper databases and other public databases, or may be updated based on past retrosynthetic analysis results. When collecting information on chemical reactions from public databases, etc., a technology such as web scraping may be used. The update of the chemical reaction DB122 may also be performed by the operation of the system provider.

[0037] The analysis information storage unit 123 stores information related to retrosynthetic analysis results performed in the past. The information related to the retrosynthetic analysis results may include, for example, the analysis execution date and time, information about the analysis requester (user) (user ID, organization, etc.), information indicating the target compound input by the user, information about the conditions set during the analysis (upper limit of search time, upper limit of the number of search steps, restrictions on usable raw materials, etc.), the contents of the preprocessing performed in the preprocessing unit 102 (simplified sites, masked substituents, etc.), the synthetic route output by the analysis unit 103, evaluation indexes of the synthetic route (yield, cost, number of steps, etc.), the type of reaction prediction model (machine learning model) performed in the retrosynthetic analysis and parameters of the reaction prediction model, the type of optimization algorithm (search algorithm) for route search and parameters of the optimization algorithm, etc. The retrosynthetic analysis results recorded in the analysis information storage unit 123 may be viewed with a viewing restriction set so that only the user who requested the analysis can view them. The information related to the retrosynthetic analysis results recorded in the analysis information storage unit 123 may be referred to by the analysis unit 103 when a new retrosynthetic analysis is performed.

[0038] The specific structure DB 124 stores information about molecular structures that can be the target of preprocessing, which will be described later. For example, the specific structure DB 124 may store a list of partial structures that are candidates for reaction-constraining substituents in a predetermined format (for example, SMILES format, MOL file format, etc.). The structures that are candidates for reaction-constraining substituents may be set by collecting information on substituents designated by a user as reaction-constraining substituents from past retrosynthetic analysis results registered in the analysis information storage unit 123, or may be set based on the knowledge of experts such as chemists.

[0039] The various data stored in the above-mentioned compound DB121, chemical reaction DB122, analysis information storage unit 123, and specific structure DB124 can be used as learning data for constructing or strengthening a machine learning model that performs retrosynthesis analysis. In addition, the various databases included in the storage unit 120 do not necessarily need to be constructed in the management server 1, and may be provided in an external database server. For example, the compound DB121 and the chemical reaction DB122 may be configured to directly refer to an external database provided by a reagent manufacturer or the like. In addition, the analysis information storage unit 123 may be configured to be managed by providing a dedicated storage server from the viewpoint of ensuring security. In addition to the above-mentioned databases, the storage unit 120 may store data related to users such as user identification numbers, data related to various authorities, and various other data used in retrosynthesis analysis (for example, a database related to the cost of chemical synthesis, etc.).

[0040] The input information receiving unit 101 receives the designation of the target compound to be analyzed input by the user. The input format of the target compound by the user is not particularly limited, and may receive the input of the target compound in a machine-readable format such as SMILES format or MOL file format, or may receive the input of the substance name or structural formula. When the input information receiving unit 101 receives the input of the target compound in a machine-readable format such as SMILES format, the input information receiving unit 101 may generate a structural formula of the compound input by the user and present it to the user to allow the user to check whether there is an error in the input content. In addition, the input information receiving unit 101 may receive an input regarding the conditions of the retrosynthetic analysis from the user. Items set as the conditions of the retrosynthetic analysis are not particularly limited, and may include, for example, conditions regarding the search time, the number of search steps, and costs. These conditions may be specified by the user, or conditions previously defined by the system provider may be used.

[0041] Furthermore, the input information receiving unit 101 may receive designation of a site (substituent) to be preprocessed, such as a reaction-constraining substituent, which will be described later. The designation of the site to be preprocessed may also be input in a machine-readable format, such as the SMILES format, as in the case of inputting the target compound, or may be input in the format of a substituent name, structural formula, or the like.

[0042] The preprocessing unit 102 performs preprocessing for simplifying a part of the molecular structure of a base compound, using a target compound and / or an intermediate compound of the target compound as a base compound, to set a tentative structure compound.

[0043] Here, "target compound and / or intermediate compound of target compound are used as base compound" means that pretreatment may be performed on target compound before retrosynthetic analysis starts (before performing initial route search), may be performed on intermediate compound at an intermediate step in the synthetic route, or may be performed before starting and then at any intermediate step. In addition, "simplifying a part of molecular structure" corresponds to, for example, "deleting" a part of molecular structure, "masking" a part of molecular structure, "converting" a part of molecular structure to a structure in which a part of molecular structure is deleted or a simple structure, etc. In pretreatment, any one of "deleting", "masking" and "converting" may be performed, or a combination of a plurality of these processes may be performed. In this embodiment, a compound having a structure after pretreatment is referred to as a "provisional structure compound". Hereinafter, pretreatment will be described in detail based on the examples shown in Figures 4 to 6.

[0044] Fig. 4 is a diagram illustrating a pretreatment for "deleting" a part of a molecular structure. Fig. 4(a) shows the molecular structure of 4,4'-(4-(tetrahydrofuran-2-yl)phenyl)azanediyl)diphenol, which is exemplified as Example 1. This compound of Example 1 has two substituents (substituent Ia and substituent Ib) of the same structure, as shown by the double-dashed line in Fig. 4(b).

[0045] These two substituents of the same structure are substituents that can be formed (added) simultaneously in one chemical reaction. The preprocessing unit 102 determines whether or not there are multiple substituents of the same structure in the molecular structure of the base compound, and when it is determined that there are multiple substituents of the same structure as shown in FIG. 4(b), it sets a tentative structure compound by deleting at least one of the multiple substituents of the same structure. FIG. 4(c) is an example of a tentative structure compound set by preprocessing, and the tentative structure compound is set by deleting the substituent Ia and leaving the substituent Ib out of the substituents Ia and Ib of the same structure (the substituent Ia may be left and the substituent Ib may be deleted). In this way, when at least one of the substituents of the same structure is deleted, the preprocessing unit 102 may set a tentative structure compound by replacing the deleted portion with a hydrogen atom, or may set a tentative structure compound by replacing the deleted portion with an atom other than a hydrogen atom (FIG. 4(c) illustrates the case where it is replaced with a hydrogen atom).

[0046] 4 shows an example in which there are two substituents with the same structure, but when there are three or more substituents with the same structure, the same pretreatment may be performed, and the selection of the substituents to be deleted is not particularly limited. More specifically, among a plurality of substituents with the same structure, any one of the substituents may be left and the remaining substituents may be deleted (i.e., at least one may be left).

[0047] In addition, in a molecular structure having a plurality of identical substituents, a symmetrical site may occur in the molecular structure. For example, in the case of the compound shown in FIG. 4, the site A surrounded by a dashed line in FIG. 4(b) can be said to be a symmetrical site. The pre-processing unit 102 may specify a symmetrical site from the molecular structure of the base compound, and delete at least one of the substituents included in the symmetrical site to set a tentative structural compound. More specifically, among the plurality of substituents that are elements that generate the symmetry in the symmetrical site, any one of the substituents may be left and the other substituents may be deleted. Even when the molecular structure of the base compound is simplified based on symmetry in this way, a tentative structural compound (FIG. 4(c)) similar to that in the case of searching for the above-mentioned identical structural substituent is set.

[0048] As a method for specifying a site to be deleted by generalizing the identity and symmetry of the structure, for example, a method of referring to the surrounding structure as shown below may be adopted. In this method, the surrounding structure up to a radius r from an atom to which a structure such as a substituent is added, such as the nitrogen atom (N) in the compound of FIG. 4(a), is taken as the reference range. The "radius r" shown here is a value indicating the range within the molecular structure by the nodes and edges, with the vertices of atoms, etc. in the molecular structure being nodes, and the vertices between the vertices of bonds, etc. being edges, and is expressed as an integer (it should be noted that the radius r does not define a circle expressed in units of length such as angstroms). In other words, the "surrounding structure from the base point to a radius r" means the range from the atom as the base point to the node away by the edge value indicated by the radius r. For example, in the compound of FIG. 4(a), when the radius r is set to "5" and the base point is the nitrogen atom, the range from the nitrogen atom along the three edges to the OH group of the substituent Ia, the OH group of the substituent Ib, and the tetrahydro-2-furyl group (tetrahydrofuranyl group) corresponds to the "peripheral structure from the base point to the radius r = 5."

[0049] After the reference range is determined as described above, when structures (substituents) attached to the base atom are compared, the peripheral structure determined by the radius r is referred to, and if the structures to be compared match including the peripheral structure, the compared structure is determined to be a target for deletion in preprocessing. When the radius r is set to "r=0", it is determined whether the structures attached to the base point are identical or not without considering the peripheral structure (i.e., r=0 is equivalent to deleting the same structure). On the other hand, when the radius r is set to "r=∞ (or a positive integer including the terminal node)", it is determined whether the peripheral structures are completely identical or not (i.e., r=∞ is equivalent to deleting a site having symmetry). When the above method is adopted, the radius r for determining the reference range may be set by receiving a designation of the radius r from the user in the input information receiving unit 101. In addition, the preprocessing unit 102 may propose (recommend) the value of the radius r to the user using a machine learning model and / or user log data (history data, etc.), or may automatically determine the value of the radius r.

[0050] FIG. 5 is a diagram illustrating a pretreatment for "masking" a part of a molecular structure. FIG. 5(a) shows the molecular structure of N-phenyl-2,7-dibutyl-10-(piperidin-1-yl)pyrene-4-carboxamide, which is exemplified as Example 2. The compound of Example 2 has butyl groups as shown by the substituents IIa and IIb, but these butyl groups are substituents with limited reactivity, and may not be added to the precursor at the end of the normal synthesis (in other words, at the beginning of the retrosynthesis). If the bond of the butyl group is cut at the beginning of the retrosynthesis analysis of Example 2, even if it is calculated, a synthetic route that is not reproducible (low feasibility) in reality will be output. In order to plan an appropriate synthetic route, it is necessary to consider the reactivity of the substituents.

[0051] Examples of substituents with restricted reactivity, such as the butyl group in Example 2, include substituents with restrictions due to bonding properties, such as substituents with strong covalent bonds and substituents with resonance-stabilized electronic structures (e.g., phenyl, cyclohexyl, trifluoromethyl, etc.), substituents with restrictions due to steric factors that cause access hindrance (steric hindrance) (e.g., isopropyl, tert-butyl, mesityl, etc.), substituents with restrictions due to electronic factors, such as substituents that are inactive against electrophilic attack and substituents that are inactive against nucleophilic attack (e.g., nitro, sulfonyl, trimethylsilyl, etc.), substituents that are difficult to solvate, and substituents with limited intermolecular interactions, and other substituents with restrictions due to physicochemical properties (e.g., perfluoroalkyl groups, crown ether rings, biphenyl groups, etc.).

[0052] In addition to the above-mentioned restrictions on the reactivity of the substituent itself, there are also cases where a specific substituent (functional group) is temporarily protected in the course of a synthetic route to restrict the reaction of the substituent for the purpose of protecting it from a specific chemical reaction or for the purpose of realizing a regioselective reaction (so-called a protecting group). In this embodiment, the above-mentioned substituents with restrictions on reactivity and the substituents to be protected are collectively referred to as "reaction-restricting substituents". In the retrosynthetic analysis of this embodiment, the reaction-restricting substituent is defined as "a substituent that temporarily restricts the cleavage of a bond".

[0053] The preprocessing unit 102 may set a tentative structural compound by simplifying a part of the molecular structure by masking reaction-restricting substituents that temporarily restrict bond cleavage among the substituents included in the molecular structure of the base compound. "Masking reaction-restricting substituents" means that a specific substituent is virtually covered with a mask, and the masked portion is temporarily excluded from the target of chemical reaction search in the retrosynthetic analysis so that the masked portion is not subject to bond cleavage.

[0054] The masked portion, i.e., the reaction-constraining substituent, may be specified by the user. In this case, the input information receiving unit 101 receives the specification of the substituent to be set in the reaction-constraining substituent from the user. Such specification of the reaction-constraining substituent by the user may be executed when the target compound is input, or may be executed during the synthetic route search. In the latter case, for example, when the candidate synthetic route branches into multiple routes, or when a candidate chemical reaction for the next synthetic step is not found, an alert may be notified to the user, and the specification of the reaction-constraining substituent may be accepted using the alert as a trigger. The preprocessing unit 102 identifies the reaction-constraining substituent specified by the user from the molecular structure of the base compound, and sets a tentative structure compound in which the reaction-constraining substituent is masked.

[0055] The preprocessing unit 102 may automatically specify the reaction-restricting substituent or its candidate without being limited to the method specified by the user. In this case, the preprocessing unit 102 may refer to the above-mentioned specific structure DB 124, and automatically specify the reaction-restricting substituent from the molecular structure of the base compound based on the database of partial structures that are candidates for the reaction-restricting substituent, and set a tentative structure compound in which the reaction-restricting substituent is masked. In addition, the preprocessing unit 102 may specify candidates for the substituent to be the reaction-restricting substituent from the molecular structure of the base compound based on the database of partial structures that are candidates for the reaction-restricting substituent, and present (suggest) them to the user. Then, the preprocessing unit 102 may set the tentative structure compound by having the user who has received the presentation select a substituent to be masked as the reaction-restricting substituent from the presented candidates.

[0056] As described above, the database of partial structures that are candidates for reaction-constraining substituents may be constructed based on past retrosynthetic analysis results (i.e., historical data indicating substituents designated by the user as reaction-constraining substituents in past analyses), or may be constructed based on expert knowledge, or may be a database that reflects both data.

[0057] In Example 2 of FIG. 5, the butyl groups of the substituents IIa and IIb shown in (a) are specified as reaction-restricting substituents, and a tentative structure compound in which the butyl groups are masked is set ((b) of FIG. 5). In the compound of Example 2, the substituents IIa and IIb also correspond to "symmetrical substituents (substituents at a site having symmetry)". In such a case, the preprocessing unit 102 may set a tentative structure compound by performing both a preprocessing for deleting a part of the substituents of the same structure and a preprocessing for masking the reaction-restricting substituents on the base compound. For example, after deleting at least one of the substituents of the same structure (deleting either the substituent IIa or the substituent IIb), the remaining substituent may be masked. In addition, a synthetic route search by retrosynthetic analysis may be carried out by masking both the substituents IIa and IIb, and a preprocessing for deleting at least one of the substituents of the same structure may be performed at the timing when the mask is released in a predetermined step.

[0058] FIG. 6 is a diagram illustrating a pretreatment for "masking" a part of a molecular structure, similar to FIG. 5. FIG. 6(a) shows the molecular structure of 2-propyl-6-(piperidin-1-yl)-4-phenylbenzoate-naphthalene, which is exemplified as Example 3. In the compound of Example 3, the butyl group shown as Substituent III is also specified as a reaction-restricting substituent, and a tentative structure compound in which the butyl group is masked (FIG. 6(b)) is set.

[0059] The analysis unit 103 is a functional unit that executes a retrosynthetic analysis to determine a synthetic route of a base compound. The analysis unit 103 may be realized using a machine learning model for reaction prediction and a search algorithm for reaction route search. The architecture of the retrosynthetic analysis executed by the analysis unit 103 may adopt a known algorithm and is not necessarily limited.

[0060] When preprocessing is performed, the analysis unit 103 performs a retrosynthetic analysis on the provisional structure compound using the provisional structure compound with a part of the molecular structure simplified as input information. Specifically, the analysis unit 103 searches for a chemical reaction that synthesizes the molecular structure of the provisional structure compound by referring to the compound DB 121 and / or the chemical reaction DB 122, targeting the molecular structure of the provisional structure compound excluding the structure simplified by "deletion", "mask", "conversion", etc. Then, based on various DBs and the architecture of the retrosynthetic analysis, the chemical reaction for generating the molecular structure of the provisional structure compound and the intermediate compound required for the chemical reaction are determined.

[0061] When searching for a chemical reaction, search conditions may be set such as giving priority to a reaction with a low synthesis cost, giving priority to a reaction in which a raw compound in stock is output as an intermediate compound, and so on (other conditions such as giving priority to a well-known chemical reaction, giving priority to a chemical reaction with a high frequency of execution, and so on). In addition, when a plurality of candidates for a chemical reaction are found by searching for a chemical reaction, the analysis unit 103 may branch the synthesis route in parallel and continue the retrosynthetic analysis for each of the branched routes. Alternatively, when a plurality of candidates for a chemical reaction are found, the candidates may be presented to the user to allow the user to select a chemical reaction to be adopted. In addition, when a plurality of candidates for a chemical reaction exist, the analysis unit 103 may adopt a reaction with the lowest synthesis cost, or may refer to information such as the stock status recorded in the compound DB 121 and give priority to a chemical reaction that can be handled by a compound in stock. In addition, when preprocessing of the base compound is not performed before the retrosynthetic analysis in a predetermined step, the analysis unit 103 may accept the molecular structure of the base compound as input information as it is and perform a normal retrosynthetic analysis.

[0062] After determining the chemical reaction in one step as described above, the analysis unit 103 continues the retrosynthetic analysis using an intermediate compound required in the determined chemical reaction as a starting compound. At this time, the analysis unit 103 may execute a verification as to whether or not to restore the partial structure simplified by the preprocessing. For example, the analysis unit 103 may execute a process of provisionally removing the mask from the intermediate compound determined in the retrosynthetic analysis of the masked provisional structure compound, and verify whether or not the next synthetic route can be output from the intermediate compound in the molecular structure in the unmasked state (whether or not a suitable chemical reaction for the next step can be found).

[0063] When the analysis unit 103 determines that the next synthetic route can be output, it performs a retrosynthetic analysis on the intermediate compound after the mask is removed, in a mask-removed state, and determines a synthetic reaction (chemical reaction) of the intermediate compound. On the other hand, when it determines that the next synthetic route cannot be output (or when it determines that the mask removal is not appropriate), the analysis unit 103 reapplies the mask to the intermediate compound, performs a retrosynthetic analysis, and determines a synthetic reaction of the intermediate compound. The location of the reapplied mask is the same as the location where the mask was applied in the previous step.

[0064] In addition, after the chemical reaction at each step is determined, the analysis unit 103 may determine whether or not a series of ongoing retrosynthetic analysis satisfies the limiting condition (i.e., whether or not the upper limit of the search time, the upper limit of the search steps, or the upper limit of the synthesis cost has been reached) and determine whether or not the retrosynthetic analysis can be continued. If the ongoing analysis is within the range of the limiting condition, the retrosynthetic analysis may be continued, and if the ongoing analysis is outside the range of the limiting condition (if the upper limit of the search time, the upper limit of the search steps, or the upper limit of the synthesis cost has been reached), the processing may be stopped as a search failure. If all intermediate compounds of the determined chemical reaction are registered compounds within the range of the limiting condition, the analysis unit 103 determines that the search of the analysis is successful and ends the analysis of the target compound. The analysis unit 103 may set registered compounds such as over-the-counter drugs and user-registered compounds as the goal of the synthetic route search, or may refer to the inventory status of the compound DB 121 and set only registered compounds in stock as the goal for ending the synthetic route search.

[0065] For example, Figures 8, 10, and 12 are examples of the retrosynthesis analysis results of Examples 1-3 executed by the analysis unit 103, and Figures 7, 9, and 11 are the corresponding comparative example data. The analysis results of each comparative example and example will be described below.

[0066] Comparative Example 1 FIG. 7 is a diagram showing the results of performing retrosynthetic analysis on the compound shown in FIG. 4(a) without performing preprocessing (Comparative Example 1). As shown in FIG. 7, when preprocessing is not performed on the compound in FIG. 4(a) having the same structural substituent (a site having symmetry), the bonds are broken finely at the end of the molecular structure, the synthetic route does not converge easily, the number of steps exceeds the limiting condition, and the search for the synthetic route fails. Note that, although FIG. 7 shows only the output results up to step 2, in reality, the search for chemical reactions continues even after step 2, and the search results after step 2 are omitted from the illustration.

[0067] Example 1 On the other hand, FIG. 8 is a diagram showing the results (Example 1) of performing a retrosynthetic analysis on the compound shown in FIG. 4(a) by performing preprocessing to delete some of the substituents (symmetrical sites) of the same structure. Specifically, in Example 1, one of the two hydroxyphenyl groups contained in the molecular structure was deleted to set a tentative structure compound. Then, when this tentative structure compound was input into a retrosynthetic analysis algorithm and analysis was performed, as shown in FIG. 8, in step 2, both intermediate compounds converged to the registered compound, and a synthetic route more efficient than that of Comparative Example 1 could be output.

[0068] Comparative Example 2 Figure 9 is a diagram showing the results of performing a retrosynthetic analysis on the compound shown in Figure 5(a) without performing pretreatment (Comparative Example 2). As shown in Figure 9, in Comparative Example 2, a synthetic route was searched for in which bonds were finely cut from the end of the molecular structure to avoid the butyl group, and the synthetic route did not converge easily, so the number of steps exceeded the limiting condition, resulting in a failure in the search for a synthetic route. Note that the search results from step 3 onwards are also omitted in Figure 9.

[0069] Example 2 On the other hand, FIG. 10 is a diagram showing the result of performing a retrosynthesis analysis (Example 2) for the compound shown in FIG. 5(a) by performing preprocessing to mask and simplify a part of the molecular structure. As shown in FIG. 10, in Example 2, two butyl groups included in the molecular structure were specified as reaction-constraining substituents, and a tentative structure compound in which these butyl groups were masked was set. This masked tentative structure compound was input to a machine learning model to perform a retrosynthesis analysis. In the case of Example 2, the analysis unit 103 determined the chemical reaction of step 1, temporarily released the mask, and verified whether the next synthetic reaction could be output from the intermediate compound of step 1 in the unmasked state. As a result, the mask was reapplied and the chemical reaction of step 2 was determined. After step 2, the same verification was performed, and the mask was reapplied to determine the chemical reaction of step 3.

[0070] After step 3, the mask is removed, and the preprocessing unit 102 performs preprocessing on the intermediate compound of step 3 to delete one of two identically structured substituents (butyl group), and the analysis unit 103 determines the chemical reaction of step 4 by performing a retrosynthetic analysis on the tentative structure compound after the preprocessing. Note that the deletion of the identically structured substituent may be performed at the stage of preprocessing on the initial target compound, not between step 3 and step 4. In Example 2, finally, in step 5, a synthetic reaction using two registered compounds could be output, and an efficient synthetic route could be output with fewer steps than in Comparative Example 2.

[0071] Comparative Example 3 Figure 11 shows the results of performing retrosynthetic analysis on the compound shown in Figure 6(a) without performing pretreatment (Comparative Example 3). In Comparative Example 3, as in Comparative Example 2, a synthetic route that finely breaks bonds was selected, and the synthetic route did not converge, the number of steps exceeded the limiting condition, and the search for the synthetic route failed. In Figure 11, the chemical reaction search from step 5 onwards is omitted.

[0072] Example 3 On the other hand, FIG. 12 is a diagram showing the results of performing a retrosynthetic analysis (Example 3) on the compound shown in FIG. 6(a) by performing preprocessing to mask and simplify a part of the molecular structure. As shown in FIG. 12, in Example 3, a butyl group included in the molecular structure was designated as a reaction-constraining substituent, and a tentative structural compound in which this butyl group was masked was set. In Example 3, the mask was released after step 5, and in step 6, a synthetic reaction converging to two registered compounds could be output. As with the other examples, in Example 3, a synthetic route with fewer steps could be output more efficiently than in Comparative Example 3.

[0073] As shown in Examples 1 to 3, in the retrosynthetic analysis system of the present embodiment, a part of the molecular structure of the base compound is simplified (deleted, masked, converted, etc.) as a preprocessing step of the retrosynthetic analysis, thereby making it possible to set a synthetic route that is more efficient than the route shown in the comparative example.

[0074] The display control unit 104 executes a process of outputting the synthetic route of the target compound determined by the analysis unit 103 to the display unit 22 of the user terminal 2. The synthetic route output by the analysis unit 103 may be output in the form of a phylogenetic tree (e.g., FIG. 8, FIG. 10, FIG. 12) showing chemical reactions of a series of steps determined by the retrosynthetic analysis. The display control unit 104 may output the synthetic route determined by the analysis to the display unit 22 after the retrosynthetic analysis is completed up to the final step (after the search has converged), or may update the phylogenetic tree and output the analysis results sequentially after the chemical reactions at each step are determined.

[0075] 8, 10, 12, etc. show analysis results without branching of the synthetic route, but when a plurality of suitable synthetic reaction candidates are searched for in the middle of a step, the synthetic route may be branched and the analysis results may be output. In such a case, the display control unit 104 may notify the user of a plurality of branched synthetic reaction candidates and accept a selection operation from the user on a UI (User Interface) presented on the display unit 22 regarding which route to continue the chemical reaction search.

[0076] When preprocessing is performed in the first stage of the retrosynthetic analysis and / or in the middle of the synthetic route, the display control unit 104 may present the site simplified by the preprocessing in a visually recognizable manner on the synthetic route output to the display unit 22. The method of presenting the preprocessed site is not particularly limited, and for example, a method of changing the color of the target site, highlighting the target site, showing the preprocessing step (transition from the base compound to the provisional structure compound), or the like may be adopted. In addition, since the preprocessing itself is a step that is not executed in the actual synthetic reaction, whether or not the target site (step) of the preprocessing is displayed on the synthetic route may be switchable according to a user's selection operation.

[0077] When the target compound is input or when the designation of a reaction-constraining substituent is received from the user in a step during the retrosynthetic analysis, the display control unit 104 may present the molecular structure (e.g., structural formula) of the base compound on the UI presented on the display unit 22, and allow the user to select a substituent to be set as a reaction-constraining substituent via the UI screen presenting the molecular structure. In addition, when the preprocessing unit 102 automatically sets a reaction-constraining substituent, the display control unit 104 may present the reaction-constraining substituent set by the preprocessing unit 102 together with the molecular structure of the base compound on the UI presented on the display unit 22, and may allow the user to select whether or not to accept the setting of the reaction-constraining substituent specified by the preprocessing unit 102 via the UI screen. Even when the preprocessing unit 102 automatically proposes (suggests) candidates for the substituent to be the reaction-constraining substituent, the display control unit 104 may present candidates for the reaction-constraining substituent on the UI screen of the display unit 22, and allow the user to select whether or not to set a reaction-constraining substituent (whether or not to apply a mask), which of the multiple candidates is to be the reaction-constraining substituent, and the like.

[0078] The management server 1 may have a function not illustrated in FIG. 3. For example, the management server 1 may have a functional unit (learning unit) that executes prediction and / or a search algorithm by a machine learning model for retrosynthesis analysis that realizes the function of the analysis unit 103. The learning unit may have a function of additionally learning the contents of various databases updated in the machine learning model along with updates of various databases such as the compound DB 121 and the chemical reaction DB 122. In addition, the learning unit may execute additional learning of the machine learning model and / or correction of the search algorithm by using the retrosynthesis analysis result data (data including information indicating the success / failure of the search) stored in the analysis information storage unit 123. The management server 1 may have a functional unit that constructs learning data to be used for machine learning in addition to the functional unit that executes machine learning.

[0079] A part or all of the functions of the management server 1 as described above may be realized by a processor of the user terminal 2.

[0080] <Functions of user terminal 2> As shown in FIG. 3, the user terminal 2 may include, for example, an input unit 21, a display unit 22, a transmission / reception unit 23, and the like. The input unit 21 is composed of information input devices such as a keyboard, mouse, touch panel, and the like, and directly accepts various input operations by the user, such as designation of a target compound, designation of a reaction constraint substituent, and designation of a limiting condition during a search. The display unit 22 is composed of information display devices such as a display and a touch panel, and displays a UI for accepting input operations, a UI for presenting analysis results, and other data transmitted from the management server 1 on the screen. The input unit 21 and the display unit 22 may be integrally configured (e.g., a touch panel). The transmission / reception unit 23 is a communication interface that communicates between the user terminal 2 and the management server 1. The transmission / reception unit 23 has a function of transmitting user operation information input via the user terminal 2 to the management server 1, and a function of receiving various processing result information, such as analysis results output by the management server 1, from the management server 1.

[0081] <An example of a retrosynthetic analysis method> Next, the retrosynthesis analysis method according to this embodiment will be described with reference to the flowcharts shown in FIGS.

[0082] When performing a retrosynthetic analysis, first, the input information receiving unit 101 receives an input of a target compound to be analyzed from a user via the user terminal 2 (step SQ101 in FIG. 13). At this time, the input information receiving unit 101 may receive from the user designation of restriction information for the search, such as an upper limit of the search time, an upper limit of the number of search steps, an upper limit of the synthesis cost, etc.

[0083] Next, the preprocessing unit 102 judges whether or not a partial structure to be preprocessed (e.g., a plurality of substituents of the same structure, a site having symmetry, or a reaction-constraining substituent, etc.) exists in the molecular structure of the target compound input by the user (step SQ102). If it is judged that a partial structure to be preprocessed exists in the molecular structure of the target compound ("Yes" in step SQ102), the preprocessing unit 102 executes preprocessing to simplify a part of the molecular structure of the target compound (base compound) and sets a tentative structure compound with a simplified part of the structure (step SQ103). "Simplification" in the preprocessing corresponds to, for example, "deleting" a part of the molecular structure, "masking" a part of the molecular structure, or "converting" a part of the molecular structure to a structure with a part of the molecular structure deleted or a simple structure. After the tentative structure compound is set, the analysis unit 103 performs a retrosynthetic analysis on the tentative structure compound using the tentative structure compound as input information and determines a chemical reaction to generate the target compound (step SQ104).

[0084] In addition, in step SQ102, if the preprocessing unit 102 determines that the substructure to be preprocessed does not exist in the molecular structure of the target compound ("No" in step SQ102), the process proceeds to step SQ104 without performing preprocessing, and a retrosynthetic analysis is performed using the molecular structure of the target compound as input information as is.

[0085] In the retrosynthetic analysis in step SQ104, the analysis unit 103 refers to various databases such as a compound DB 121 and a chemical reaction DB 122 to determine the positions at which bonds are cut in the molecular structure of the input information, intermediate compounds produced by the bond cutting, and chemical reactions. The function of the analysis unit 103 in step SQ104 may be executed as a function of a machine learning model or a search algorithm for retrosynthetic analysis, and the architecture of the retrosynthetic analysis is not particularly limited.

[0086] If any of the intermediate compounds determined by the retrosynthetic analysis in step SQ104 is not a registered compound ("Yes" in step SQ105), the analysis unit 103 continues the retrosynthetic analysis step with the intermediate compound as a base compound to be input to the machine learning model and the search algorithm. At this time, the analysis unit 103 determines whether the ongoing analysis is within a range of a predetermined limiting condition (whether the upper limit of the search time has been reached, whether the upper limit of the number of search steps has been reached, whether the upper limit of the synthesis cost has been reached, etc.) (step SQ106). If it is determined that the analysis can be continued within the range of the limiting condition, the analysis unit 103 determines whether to perform preprocessing on the intermediate compound determined in step SQ104 (step SQ107).

[0087] When the molecular structure of the intermediate compound includes a partial structure to be preprocessed and the analysis unit 103 judges that preprocessing should be performed to determine the next synthetic route ("Yes" in step SQ107), the process proceeds to step SQ103. Then, the preprocessing unit 102 sets a tentative structure compound that simplifies a part of the molecular structure of the intermediate compound, and performs retrosynthetic analysis on the tentative structure compound (step SQ104). On the other hand, when the analysis unit 103 judges that preprocessing is not to be performed on the intermediate compound ("No" in step SQ107), the process proceeds to step SQ104, and the analysis unit 103 performs retrosynthetic analysis on the molecular structure of the intermediate compound that has not been subjected to preprocessing.

[0088] The cycle of steps SQ103 to SQ107 as described above is repeated, and the one-step process of the retrosynthetic analysis is continued until all of the intermediate compounds in the analysis result converge to registered compounds that can be used as raw materials. In step SQ105, if all of the intermediate compounds determined in the retrosynthetic analysis are registered compounds, the process proceeds to step SQ108, and the analysis unit 103 completes the search for chemical reactions and outputs a series of synthetic routes, which are the analysis results, to the user terminal 2. On the other hand, in the repetition of the cycle of steps SQ103 to SQ107, if the analysis reaches a predetermined upper limit (upper limit of search time, upper limit of number of search steps, upper limit of synthesis cost, etc.) set as a limiting condition ("No" in step SQ106), the analysis unit 103 determines that the search for the synthetic route has failed, and ends the analysis.

[0089] Fig. 14 shows an example of a flowchart for carrying out a retrosynthetic analysis after receiving a reaction constraint substituent specification from a user. The flowchart in Fig. 14 embodies steps SQ101, SQ102, SQ103, and SQ107 in Fig. 13, and the other steps are the same as those in the flowchart shown in Fig. 13 and are denoted by the same reference numerals.

[0090] As shown in step SQ1011 of FIG. 14, the input information receiving unit 101 receives an input of a target compound from a user, and may also receive from the user a designation of a reaction-restricting substituent that temporarily restricts the cleavage of a bond, such as a substituent with a restriction on reactivity (a substituent with low reactivity). Next, the preprocessing unit 102 judges whether or not a plurality of substituents with the same structure are present in the molecular structure of the target compound input by the user (in other words, whether or not a site having symmetry is present) (step SQ1021 of FIG. 14). If a plurality of substituents with the same structure are present ("Yes" in step SQ1021), the preprocessing unit 102 deletes at least one of the plurality of substituents with the same structure to simplify a part of the molecular structure of the target compound (step SQ1031). On the other hand, if a substituent with the same structure is not present in the molecular structure of the target compound ("No" in step SQ1021), the process proceeds to step SQ1032.

[0091] In step SQ1032, the preprocessing unit 102 sets a tentative structure compound by masking the reaction constraint substituents designated by the user. Then, the analysis unit 103 executes a retrosynthetic analysis on the tentative structure compound set in steps SQ1031 to SQ1032 to determine the synthesis reaction of the base compound (step SQ104). After one chemical reaction is determined in the one-step process, if the analysis is to be continued in the same flow as in FIG. 13, in step SQ1071, the analysis unit 103 tentatively removes the mask from the intermediate compound determined in step SQ104 and verifies whether the next synthesis route can be output (step SQ1072).

[0092] If the next synthetic route can be output ("Yes" in step SQ1072), the analysis unit 103 removes the mask set in the preprocessing and proceeds to step SQ104, where it performs a retrosynthetic analysis on the intermediate compound after the mask is removed. On the other hand, if it is determined that the next synthetic route cannot be output, it proceeds to step SQ1032, where the preprocessing unit 102 reapplies the mask to the reaction constraint substituents in the molecular structure of the intermediate compound. Then, it performs a retrosynthetic analysis on the intermediate compound with the mask applied.

[0093] In the flow shown in FIG. 14 as well, the cycle from step SQ1031 to step SQ1072 is repeated, and when the synthetic route converges to a registered compound (step SQ105 “No”), the search for chemical reactions is terminated and a series of synthetic routes is output.

[0094] Note that the flow of the retrosynthetic analysis method shown in FIG. 13 and FIG. 14 is merely an example, and the retrosynthetic analysis method of this embodiment is not limited to the examples shown in FIG. 13-FIG. 14. For example, the preprocessing unit 102 may automatically set a reaction-constraining substituent from the molecular structure of the base compound based on a database (a database of partial structures that are candidates for reaction-constraining substituents) constructed based on past retrosynthetic analysis results and / or expert knowledge, or may suggest candidates for reaction-constraining substituents. In addition, the order of some of the steps shown in FIG. 13 and FIG. 14 may be changed. For example, the user may specify the reaction-constraining substituent in a step (SQ107) during the retrosynthetic analysis, rather than in the first step (step SQ1011). In addition, in FIG. 14, a preprocessing step is performed before the reaction-constraining substituent is masked, in which a part of the substituents of the same structure is deleted, but the part of the substituents of the same structure may be deleted at the stage where the mask is released from the intermediate compound (SQ107).

[0095] In the retrosynthetic analysis system according to the present embodiment, as described above, a target compound or an intermediate compound is used as a base compound, and a preprocessing is performed to simplify a part of the molecular structure of the base compound, and then a retrosynthetic analysis is performed, thereby making it possible to efficiently search for a synthetic route for the target compound. As a result, it is possible to plan a feasible and efficient synthetic route for a compound having a specific structure, such as a molecular structure having multiple substituents of the same structure, a molecular structure having symmetry, or a molecular structure including a reaction-constraining substituent.

[0096] The above-described embodiment is merely an example for facilitating understanding of the present disclosure, and is not intended to limit the present disclosure. The present disclosure can be modified or improved without departing from the spirit thereof, and it goes without saying that the present disclosure includes equivalents thereof.

[0097] <Modification> For example, the preprocessing of the retrosynthetic analysis is not limited to the case of simplifying a part of the molecular structure. The preprocessing unit 102 may execute a preprocessing to add a predetermined partial structure (substituent such as a protecting group) to a part of the molecular structure of the base compound. In the synthetic route of an organic compound, a specific substituent may be temporarily modified (so-called a protecting group) to control the reactivity of the substituent and protect it from a chemical reaction during the synthesis. Such a protecting group may be formed not only for the purpose of protection from a chemical reaction, but also for the purpose of realizing a regioselective reaction or for controlling the solubility in a predetermined solvent. After completing its purpose in the process of the synthetic route, the protecting group is removed (removed) by a deprotection reaction or the like.

[0098] The formation of a protecting group is an addition reaction, but since retrosynthetic analysis is a method of sequentially determining chemical reactions and reacting compounds at reaction points, such as bond cleavage, and searching for a synthetic route, it may be difficult to form and select an appropriate protecting group in normal retrosynthetic analysis. In the retrosynthetic analysis system of the present disclosure, after receiving from the user a designation of a protecting group to be added to a part of the molecular structure of a base compound, the substituent on which the protecting group is formed may be masked and retrosynthetic analysis may be performed. In this case, the preprocessing unit 102 may receive a designation of a protecting group during the retrosynthetic analysis, or may receive an instruction to remove the protecting group at a predetermined timing in the retrosynthetic analysis. When receiving a designation of a protecting group from a user, it is not limited to receiving a designation of one structure to be a protecting group, but may also receive a designation of multiple candidate structures and / or a selection from multiple candidates. For example, in addition to receiving the addition of an F group, the preprocessing unit 102 may receive a designation of a group of candidates including other halogen atoms such as Cl, Br, and I, and appropriately determine a structure to be added to the base compound from the group of candidates received by the preprocessing unit 102. In addition, instead of receiving a designation of a protecting group from a user, the pre-processing unit 102 may automatically set or suggest the formation and removal of a protecting group by referring to a database based on past retrosynthetic analysis results and / or expert knowledge.

[0099] In the present disclosure, the preprocessing unit 102 may execute a combination of multiple preprocessing processes on the base compound. For example, when combining the deletion of substituents with the same structure with masking, various combinations are possible, such as (1) executing masking after deleting the substituents with the same structure, (2) deleting the substituents with the same structure after executing masking, and (3) executing masking and deletion of the substituents with the same structure in parallel. The order of the combination of the preprocessing processes may be appropriately set according to the structure of the target compound, a user's specification, and the like.

[0100] The retrosynthetic analysis system described herein may be realized as a single device, or may be realized by a plurality of devices (e.g., cloud servers) partially or entirely connected via a network. For example, the functions of the control unit (processor 10) and storage 12 of the management server 1 may be realized by different servers connected to each other via a network.

[0101] In addition, the series of processes by the retrosynthesis analysis system described in this specification may be realized using any of software, hardware, and a combination of software and hardware. It is also possible to create a computer program for realizing each function of the management server 1 according to this embodiment and install it in a PC or the like. In addition, a computer-readable recording medium in which such a computer program is stored can also be provided. The recording medium is, for example, a magnetic disk, an optical disk, a magneto-optical disk, a flash memory, etc. In addition, the above computer program may be distributed, for example, via a network, without using a recording medium.

[0102] In addition, the effects described in this specification are merely descriptive or exemplary and are not limiting. In other words, the technology according to the present disclosure may achieve other effects that are apparent to a person skilled in the art from the description of this specification, in addition to or in place of the above effects.

[0103] The retrosynthetic analysis system, the retrosynthetic analysis device, the retrosynthetic analysis method, and the retrosynthetic analysis program according to the present disclosure may have the following configuration. [Item 1] an input information receiving unit that receives a target compound to be analyzed; a pre-processing unit that performs a pre-processing for simplifying a part of a molecular structure of the base compound using the target compound and / or an intermediate compound of the target compound as a base compound to set a tentative structure compound; and an analysis unit that performs a retrosynthetic analysis on the tentative structural compound and determines a synthetic route for the base compound. [Item 2] The pre-treatment unit includes: determining whether or not a plurality of substituents having the same structure are present in the molecular structure of the base compound; 2. The retrosynthetic analysis system according to item 1, wherein, when it is determined that a plurality of substituents having the same structure are present, at least one of the plurality of substituents having the same structure is deleted to set the tentative structure compound. [Item 3] The pre-treatment unit includes: Identifying a site having symmetry in the molecular structure of the base compound; 2. The retrosynthetic analysis system according to item 1, wherein the provisional structure compound is set by deleting at least one substituent contained in the symmetrical portion. [Item 4] The pre-treatment unit includes: 4. The retrosynthetic analysis system according to any one of items 1 to 3, wherein a reaction-constraining substituent that temporarily restricts bond cleavage is masked among substituents contained in a molecular structure of the base compound to simplify a part of the molecular structure and set the tentative structure compound. [Item 5] the input information receiving unit receives, from a user, designation of a substituent to be set as the reaction constraint substituent; 5. The retrosynthetic analysis system according to item 4, wherein the preprocessing unit identifies the reaction constraint substituent designated by the user from the molecular structure of the base compound, and sets the tentative structure compound in which the reaction constraint substituent is masked. [Item 6] 5. The retrosynthetic analysis system according to item 4, wherein the preprocessing unit automatically identifies the reaction restricting substituent from the molecular structure of the base compound based on a database of partial structures that are candidates for the reaction restricting substituent, and sets the tentative structure compound in which the reaction restricting substituent is masked. [Item 7] The analysis unit is performing a process of provisionally removing the mask from an intermediate compound determined by a retrosynthetic analysis of the masked provisional structure compound; 5. The retrosynthetic analysis system according to item 4, wherein it is verified whether or not a next synthetic route can be output from the intermediate compound in a state where the mask is removed. [Item 8] 8. The retrosynthetic analysis system according to item 7, wherein when the analysis unit determines that the next synthetic route can be output, the analysis unit performs a retrosynthetic analysis on the intermediate compound with the mask removed, and determines a synthetic reaction of the intermediate compound. [Item 9] 8. The retrosynthetic analysis system according to item 7, wherein when the analysis unit determines that the next synthetic route cannot be output, the analysis unit reapplies the mask to the intermediate compound, performs retrosynthetic analysis, and determines a synthetic reaction of the intermediate compound. [Item 10] A display control unit that outputs the synthetic route determined by the analysis unit to a display unit, 4. The retrosynthetic analysis system according to any one of items 1 to 3, wherein the display control unit presents, on the synthetic route output to the display unit, a portion simplified by the preprocessing. [Item 11] an input information receiving unit that receives a target compound to be analyzed; a pre-processing unit that performs a pre-processing for simplifying a part of a molecular structure of the base compound using the target compound and / or an intermediate compound of the target compound as a base compound to set a tentative structure compound; and an analysis unit that performs a retrosynthetic analysis on the tentative structural compound and determines a synthetic route for the base compound. [Item 12] receiving a target compound for analysis; performing a pre-processing for simplifying a part of a molecular structure of the base compound using the target compound and / or an intermediate compound of the target compound as a base compound, and performing a pre-processing for setting a tentative structure compound; and performing a retrosynthetic analysis on the tentative structure compound to determine a synthetic route for the base compound. [Item 13] receiving a target compound for analysis; performing a pre-processing for simplifying a part of a molecular structure of the base compound using the target compound and / or an intermediate compound of the target compound as a base compound, and performing a pre-processing for setting a tentative structure compound; and determining a synthetic route for the base compound by performing a retrosynthetic analysis on the tentative structural compound. [Explanation of symbols]

[0104] 1 Management Server 2. User terminal 101 Input information reception unit 102 Pretreatment section 103 Analysis Department 104 Display control unit

Claims

1. A retrosynthetic analysis system comprising one or more computers, The computer includes: an input information receiving unit that receives an input of a target compound to be analyzed; a preprocessing unit that performs preprocessing of simplifying a part of the molecular structure of the base compound by one or more processes of deleting a part of the molecular structure, masking the part of the molecular structure, converting the part of the molecular structure into a structure in which the part of the molecular structure is deleted, and converting the part of the molecular structure into a structure simpler than the part of the molecular structure, using the target compound and / or an intermediate compound of the target compound as a base compound, and sets a tentative structure compound; and an analysis unit that performs a retrosynthetic analysis on the tentative structural compound and determines a synthetic route for the base compound.

2. The pre-treatment unit includes: determining whether or not a plurality of substituents having the same structure are present in the molecular structure of the base compound; The retrosynthetic analysis system according to claim 1 , wherein, when it is determined that a plurality of substituents having the same structure are present, at least one of the plurality of substituents having the same structure is deleted to set the tentative structure compound.

3. The pre-treatment unit includes: Identifying a site having symmetry in the molecular structure of the base compound; The retrosynthetic analysis system according to claim 1 , wherein the provisional structural compound is set by deleting at least one substituent contained in the symmetrical portion.

4. The pre-treatment unit includes: The retrosynthetic analysis system according to any one of claims 1 to 3, wherein a reaction-constraining substituent that temporarily restricts bond cleavage is masked among the substituents contained in the molecular structure of the base compound to simplify a part of the molecular structure and set the provisional structure compound.

5. the input information receiving unit receives, from a user, designation of a substituent to be set as the reaction constraint substituent; 5. The retrosynthetic analysis system according to claim 4, wherein the preprocessing unit identifies the reaction-constraining substituent designated by the user from the molecular structure of the base compound, and sets the tentative structure compound in which the reaction-constraining substituent is masked.

6. 5. The retrosynthetic analysis system according to claim 4, wherein the preprocessing unit automatically identifies the reaction-constraining substituent from the molecular structure of the base compound based on a database of partial structures that are candidates for the reaction-constraining substituent, and sets the tentative structure compound in which the reaction-constraining substituent is masked.

7. The analysis unit is performing a process of provisionally removing the mask from an intermediate compound determined by a retrosynthetic analysis of the masked provisional structure compound; The retrosynthetic analysis system according to claim 4 , wherein it is verified whether or not a next synthetic route can be output from the intermediate compound in a state where the mask is removed.

8. 8. The retrosynthetic analysis system according to claim 7, wherein when the analysis unit determines that the next synthetic route can be output, the analysis unit performs a retrosynthetic analysis on the intermediate compound with the mask removed, and determines a synthetic reaction of the intermediate compound.

9. 8. The retrosynthetic analysis system according to claim 7, wherein when the analysis unit determines that the next synthetic route cannot be output, the analysis unit reapplies the mask to the intermediate compound, performs retrosynthetic analysis, and determines a synthetic reaction of the intermediate compound.

10. A display control unit that outputs the synthetic route determined by the analysis unit to a display unit, 4. The retrosynthetic analysis system according to claim 1, wherein the display control unit presents, on the synthetic route output to the display unit, a portion simplified by the preprocessing.

11. an input information receiving unit that receives an input of a target compound to be analyzed; a preprocessing unit that performs preprocessing of simplifying a part of the molecular structure of the base compound by one or more processes of deleting a part of the molecular structure, masking the part of the molecular structure, converting the part of the molecular structure into a structure in which the part of the molecular structure is deleted, and converting the part of the molecular structure into a structure simpler than the part of the molecular structure, using the target compound and / or an intermediate compound of the target compound as a base compound, and sets a tentative structure compound; and an analysis unit that performs a retrosynthetic analysis on the tentative structural compound and determines a synthetic route for the base compound.

12. Receiving input of a target compound to be analyzed; performing a preprocessing for simplifying a part of a molecular structure of the base compound by one or more processes of deleting a part of the molecular structure, masking the part of the molecular structure, converting the part of the molecular structure into a structure in which the part of the molecular structure is deleted, and converting the part of the molecular structure into a structure simpler than the part of the molecular structure, using the target compound and / or an intermediate compound of the target compound as a base compound, and performing a preprocessing for setting a tentative structure compound; and performing a retrosynthetic analysis on the tentative structure compound to determine a synthetic route for the base compound.

13. Receiving input of a target compound to be analyzed; performing a preprocessing for simplifying a part of a molecular structure of the base compound by one or more processes of deleting a part of the molecular structure, masking the part of the molecular structure, converting the part of the molecular structure into a structure in which the part of the molecular structure is deleted, and converting the part of the molecular structure into a structure simpler than the part of the molecular structure, using the target compound and / or an intermediate compound of the target compound as a base compound, and performing a preprocessing for setting a tentative structure compound; and determining a synthetic route for the base compound by performing a synthetic synthesis analysis on the tentative structural compound.

Images