Molecular analysis system, program, and molecular analysis method

Abstract

A first acquisition unit 16 acquires information on a three-dimensional structure of a molecule obtained by quantum chemical calculation. A setting unit 18 sets a plurality of information acquisition points in the inside or around the three-dimensional structure of the molecule. A second acquisition unit 20 acquires data on one or a plurality of parameters set in advance at each information acquisition point set by the setting unit 18.

Description

Technical Field

[0001] This invention relates to a molecular analysis system, program, and method. Background Technology

[0002] Previously, methods for molecular analysis included: methods for obtaining one-dimensional structural information such as molecular formula, molecular weight, number of atoms, number of functional groups, and number of hydrogen bonds; methods for obtaining two-dimensional structural information such as chemical bonds and adjacent functional groups; and methods for obtaining three-dimensional structural information such as molecular geometry, surface area, electron density, and molecular orbitals. Additionally, Patent Document 1 discloses a method that uses the chemical characteristics of a compound as input information, employs a training set containing the compound and related biological or chemical information, and utilizes machine learning methods to generate a generative model.

[0003] Existing technical documents Patent documents Patent Document 1: Japanese Patent Publication No. 2019-502988 Summary of the Invention The problem the invention aims to solve The molecular analysis methods described above, which obtain one-dimensional, two-dimensional, and three-dimensional structural information, are only meaningful from a chemical perspective and have problems such as not being able to be linked to behavioral predictions within organisms.

[0004] This disclosure was made with consideration of the above points in mind, and its purpose is to provide a new molecular analysis system, program, and information processing method for extracting spatial information of molecules.

[0005] means for solving problems The molecular analysis system disclosed herein is characterized by having: a first acquisition unit for acquiring information on the three-dimensional structure of a molecule obtained through quantum chemical calculations; a setting unit for setting multiple information acquisition points inside or around the three-dimensional structure of the molecule; and a second acquisition unit for acquiring data at each of the information acquisition points set by the setting unit regarding one or more pre-set parameters.

[0006] In addition, in the molecular analysis system disclosed herein, the setting unit can also divide the three-dimensional structure of the molecule obtained by the first acquisition unit into a defined cube, and set information acquisition points for each cube.

[0007] In addition, in the molecular analysis system disclosed herein, the setting unit may also set an axis connecting the start point and the end point of the three-dimensional structure of the molecule obtained by the first acquisition unit. When arranging information acquisition points around the axis, a virtual three-dimensional curve from the set start point to the set end point is set, and multiple information acquisition points are set at intervals on the set virtual three-dimensional curve.

[0008] In addition, in the molecular analysis system disclosed herein, the setting unit may also set a length direction in the three-dimensional structure of the molecule obtained by the first acquisition unit, and set the starting point and the ending point near the end along the length direction.

[0009] In addition, in the molecular analysis system disclosed herein, the setting unit can also be set in a manner that maximizes the straight-line distance between the starting point and the ending point.

[0010] In addition, in the molecular analysis system disclosed herein, the setting unit may also set the virtual three-dimensional curve in a manner that covers part or all of the three-dimensional structure of the molecule obtained by the first acquisition unit.

[0011] In addition, in the molecular analysis system disclosed herein, the setting unit may also set multiple information acquisition points at equal intervals on the virtual three-dimensional curve.

[0012] In addition, the molecular analysis system disclosed herein may also include: a receiving unit for receiving information about the chemical formula of a molecule; and a generating unit for generating the three-dimensional structure of a molecule by quantum chemical calculations based on the chemical formula of the molecule received by the receiving unit.

[0013] In addition, the molecular analysis system disclosed herein may also include: a determining unit that uses a first model to determine the location of an information acquisition point based on the three-dimensional structure of the molecule obtained by the first acquiring unit, wherein the first model is generated by machine learning using training data that includes the three-dimensional structure of the molecule, the location of the information acquisition point, and a score of the analysis results of the molecule; the setting unit may also set the location of the information acquisition point determined by the determining unit.

[0014] In addition, the molecular analysis system disclosed herein may also include: a parameter type setting unit, which uses a second model to set the parameter type based on the three-dimensional structure of the molecule obtained by the first acquisition unit or the position of each information acquisition point set by the setting unit; the second model is generated by machine learning using training data that includes the three-dimensional structure of the molecule, the position of the information acquisition point, the parameter type, and a score of the analysis result of the molecule; the second acquisition unit may also acquire data on one or more parameters set by the parameter type setting unit at each information acquisition point.

[0015] The program disclosed herein is characterized by being executed by a computer as a first acquisition unit, a setting unit, and a second acquisition unit. The first acquisition unit acquires information about the three-dimensional structure of a molecule obtained through quantum chemical calculations. The setting unit sets multiple information acquisition points inside or around the three-dimensional structure of the molecule. The second acquisition unit acquires data about one or more pre-set parameters at each of the information acquisition points set by the setting unit.

[0016] The molecular analysis method disclosed herein is characterized by being a computer-executed molecular analysis method, comprising the following steps: acquiring information on the three-dimensional structure of a molecule obtained through quantum chemical calculations; setting multiple information acquisition points inside or around the three-dimensional structure of the molecule; and acquiring data at each of the set information acquisition points regarding one or more pre-set parameters.

[0017] The effects of the invention According to this disclosure, a new molecular analysis system, program, and information processing method for extracting spatial information of molecules can be provided. Attached Figure Description

[0018] Figure 1 This is a diagram that schematically illustrates the structure of the molecular analysis system involved in the embodiments of this disclosure.

[0019] Figure 2 This diagram illustrates an example of the information processing flow in the molecular analysis system and information processing method according to embodiments of this disclosure, and also illustrates an example of the acquisition of information acquisition points and parameters by the first model generation unit and the second model generation unit of the molecular analysis system.

[0020] Figure 3 This is a diagram illustrating the method for setting information acquisition points in a molecular analysis system and information processing method according to an embodiment of this disclosure.

[0021] Figure 4 This is a diagram illustrating another method for setting information acquisition points in a molecular analysis system and information processing method according to another embodiment of this disclosure.

[0022] Figure 5 This is a diagram illustrating an exemplary flow of information processing in an information processing method according to an embodiment of the present disclosure.

[0023] Figure 6 This is a diagram illustrating another exemplary flow of information processing in an information processing method according to an embodiment of the present disclosure. Detailed Implementation

[0024] Hereinafter, exemplary embodiments of the present disclosure will be described with reference to the accompanying drawings, but the invention involved in the present disclosure is not limited thereto. Figures 1 to 6 This is a diagram illustrating the molecular analysis system 1 and information processing method involved in this disclosure.

[0025] [Molecular Analysis System 1] Figure 1 The molecular analysis system 1 of this embodiment, as shown, can extract spatial information of molecules using a novel method not previously known. The molecular analysis system 1 is a computer capable of connecting to a communication network (not shown) such as an internet connection, and can be composed of an industrial computer, a personal computer, a tablet terminal, etc. Alternatively, the molecular analysis system 1 can also be a virtual server within a physical server accessible via the internet. Furthermore, there can be multiple such hardware and virtual servers, and they can be arbitrarily combined. For example... Figure 1 As shown, the molecular analysis system 1 has a control unit 10, a storage unit 28, a communication unit 30, a display unit 32, and an operation unit 34.

[0026] (Control Department 10) The control unit 10, composed of a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit), controls various processes and actions within the molecular analysis system 1. Specifically, the control unit 10 executes programs stored in the storage unit 28 (described later) to function as the receiving unit 12, generating unit 14, first acquisition unit 16, setting unit 18, second acquisition unit 20, determining unit 22, parameter type setting unit 24, scoring generation unit 26, first model generation unit 200, and second model generation unit 300. These functions can also be implemented by executing one or more independent programs or applications. Furthermore, these programs or applications can be installed on a single terminal (molecular analysis system 1) or distributed across multiple terminals (including molecular analysis system 1). In the latter case, they can be interconnected via wired cables or communication networks. The details of each of these units will be described later.

[0027] (Storage Department 28) The storage unit 28 may be composed of, for example, an HDD (Hard Disk Drive), RAM (Random Access Memory), ROM (Read Only Memory), or SSD (Solid State Drive). Furthermore, the storage unit 28 is not limited to a storage unit built into the molecular analysis system 1; it may also be a storage medium (e.g., a USB storage device) that can be freely installed and removed from the molecular analysis system 1. Alternatively, instead of a storage unit 28, various information may be stored in other storage units (such as cloud servers).

[0028] The storage unit 28 stores the program executed by the control unit 10, the first learned model 220, the second learned model 320, the three-dimensional structure of each molecule, information acquisition points, parameters and other information.

[0029] (Ministry of Communications 30) The communication unit 30 enables the control unit 10 to communicate with external devices (e.g., a computing unit not shown) via wireless or wired connection. The control unit 10 transmits and receives information from the external devices through the communication unit 30.

[0030] (Display unit 32, Operation unit 34) The display unit 32, such as a monitor, displays (outputs) various images by receiving display command signals from the control unit 10. The operation unit 34, such as a keyboard, provides (inputs) various commands to the control unit 10. Alternatively, in one embodiment, a display operation unit such as a touch panel may be provided, integrating the display unit 32 and the operation unit 34.

[0031] (Detailed description of control unit 10) (Receiver Unit 12) The receiving unit 12 receives information about the chemical formula of a molecule, for example, via the communication unit 30 and the operation unit 34. In this specification, "molecule" is used in its common sense and may include, for example, a pharmaceutical agent, the material itself, a structural isomer, a stereoisomer, an optical isomer (enantiomer), etc., without any particular limitation. Furthermore, in the analysis involved in this disclosure, the stability and flexibility of the molecule are not limited during analysis.

[0032] (Generation Unit 14) The generating unit 14 generates the three-dimensional structure of the molecule through quantum chemical calculations based on the chemical formula of the molecule received by the receiving unit 12. Figure 2 100 in (refer to) Figure 3 , Figure 4It can utilize HF (Hartree-Fock), DFT (Density Functional Theory), CI (Configuration Interaction Method), CC (Coupled Cluster), MBPT (Many Body Perturbation Theory) and other methods as "quantum chemical calculations" to obtain the three-dimensional structure of molecules.

[0033] (First Acquisition Unit 16) The first acquisition unit 16 acquires information about the three-dimensional structure of the molecule obtained through quantum chemical calculations. In addition to acquiring information about the three-dimensional structure of the molecule from the receiving unit 12 and the generating unit 14 mentioned above, the first acquisition unit 16 can also acquire information about the three-dimensional structure of the molecule from external sources.

[0034] (Setting Unit 18) The setting unit 18 sets multiple information acquisition points inside or around the three-dimensional structure of the molecule. Figure 2 (110 in the document). In this specification, "information acquisition point" refers to any spatial coordinate arbitrarily set inside or around a three-dimensional structure. As described below, for each information acquisition point (coordinate), information such as electron density (hereinafter referred to as "parameter (value)") is acquired and set. In order to find the optimal information acquisition point for each molecule, the position and number of information acquisition points can be appropriately set and changed. If the optimal information acquisition point corresponding to its use is found for each molecule, the same information acquisition point can be applied to other molecules with similar structures. The term "optimal" here will vary depending on the use and can be arbitrarily determined. If the expected effect is obtained through the actual application of the molecule (e.g., if the molecular compound binds to a specified receptor), it can also be evaluated as "optimal".

[0035] Although there are no particular limitations on the method of setting information acquisition points, the setting unit 18 can also set an axis connecting the start point and the end point of the three-dimensional structure of the molecule acquired by the first acquisition unit 16. When arranging information acquisition points around the axis (i.e., around the three-dimensional structure), a virtual three-dimensional curve from the set start point to the end point is set, and multiple information acquisition points are set at intervals on the set virtual three-dimensional curve. Figure 3 These starting and ending points can also serve as information acquisition points.

[0036] In this specification, "axis connecting the start and end points" refers to a line passing through the object's molecules, with the "start point" and "end point" being its two ends. For example, in Figure 3The upper left side of the section is depicted with a double-dotted line, representing the "axis connecting the start and end points." Furthermore, similar to the information acquisition point, the "start point" and "end point" do not necessarily need to be set on the three-dimensional structure; they can be set inside or around the three-dimensional structure. The setting unit 18 can be set in a way that maximizes the straight-line distance between the start and end points (and the axis), or it can set the length direction within the three-dimensional structure of the molecule acquired by the first acquisition unit 16, and set the start and end points near the ends along that length direction. In cases where multiple length directions can be set, such as with water molecules, any one of them can be chosen.

[0037] Additionally, the setting unit 18 can also set a virtual three-dimensional curve to cover part or all of the three-dimensional structure of the molecule acquired by the first acquisition unit 16. The "virtual three-dimensional curve" is a mathematical curve that configures information acquisition points. Information acquisition points can be set as 0 to multiple intermediate points other than the "start point" and "end point." Furthermore, since the binding ability, interactions, and side effects of the molecule and its functional groups are predicted based on the parameter values ​​at the information acquisition points, more information is better. However, depending on the molecule, not all molecules require information acquisition points (parameter values) sufficient to cover their entire three-dimensional structure. For example, efficacy and side effects can be estimated by focusing on highly reactive parts such as functional groups. More specifically, even among molecules (compounds) with common functional groups, it is possible to predict changes in the efficacy and side effects of each compound due to differences in structures other than those functional groups. That is, it is also possible to focus on identifying specific structures.

[0038] Furthermore, as a "shape" of a virtual three-dimensional curve, spiral shapes can be listed (e.g., Figure 3 The virtual 3D curve can include, but is not limited to, the upper right side of the curve (e.g., a parabolic shape). As mentioned above, if multiple information acquisition points are set, a portion of the virtual 3D curve can also contain straight lines.

[0039] As an "interval" for setting multiple information acquisition points on a virtual 3D curve, multiple information acquisition points can be set at equal intervals, and the distance between each interval can be arbitrarily set and changed. In addition, information acquisition points do not necessarily have to be configured on the virtual 3D curve; they can also be set along the virtual 3D curve or in its vicinity.

[0040] As another method for setting information acquisition points, the setting unit 18 can also divide the three-dimensional structure of the molecule acquired by the first acquisition unit 16 into a defined cube, and set information acquisition points for each cube. The "defined cube" is equivalent to a voxel. A voxel can be treated as an information acquisition point as a whole (one voxel = one information acquisition point). In this case, the information acquisition point can be the center point of the voxel, or multiple coordinates within a voxel can be integrated into a single information acquisition point by generalizing statistical measures such as the mean, median, and mode. The size of the voxel is not particularly limited and can be set and changed arbitrarily. In this voxel method, a collection of voxels of the same size is formed in the coordinate system, and the three-dimensional structure of the molecule is represented by the collection of unit voxels. As a method for setting information acquisition points in the voxel method, similar to the case of setting information acquisition points on a virtual three-dimensional curve, after arranging voxels inside and around the three-dimensional structure of the molecule, the information acquisition points (coordinates) in the arranged voxels are set. Alternatively, information acquisition points can be appropriately selected from the outermost voxel of an assembly of voxels corresponding to the molecule's three-dimensional structure, or from a portion of the molecule's three-dimensional structure (e.g., highly reactive parts such as functional groups). Or, the molecule's three-dimensional structure can be divided into a lattice (equally spaced on the XY, YZ, and ZX planes; i.e., voxels are defined by the lattice and spacing), and the points where the three-dimensional structure intersects with the lattice can be set as information acquisition points. In these methods, since it is not necessary to set start and end points, information acquisition points can be easily set.

[0041] (Second Acquisition Unit 20) The second acquisition unit 20 acquires data about one or more preset parameters at each information acquisition point set by the setting unit 18. Figure 2(120 in the original text). Here, "parameters" can encompass multidimensional information related to the binding capacity of molecules, including stereoscopic, electronic, and chemical information. Based on stereoscopic information, the arrangement of the molecule's three-dimensional structure can be oriented, distinguishing stereoisomers that have been almost entirely ignored until now, and even resolving flexible molecules. Furthermore, based on electronic and stereoscopic information, calculations of intermolecular influences (interactions) become possible. Examples of parameters include charge states, electron density, density of states (DOS), molecular orbitals, electrostatic potential, and space-filling ratio. Additionally, "pre-defined" refers to the ability to select data for parameters of existing molecules, since data on the positions of information acquisition points set within or around their three-dimensional structure can be obtained. Specifically, the electron wavefunction of the molecule can be calculated non-empirically using a quantum chemical calculation program to obtain parameters such as electron density. These quantum chemical calculations can also be independent of the quantum chemical calculations used to obtain the molecule's three-dimensional structure. Examples of programs and software for performing quantum chemical calculations to obtain parameter values ​​include Gaussian, ADF (Amsterdam Density Functional), and Psi4.

[0042] The storage unit 28 can map the three-dimensional structure, information acquisition points, and parameters of each molecule to a "dataset" for storage. Additionally, axes, virtual three-dimensional curves, voxels, etc., can also be mapped and stored. Figure 3 , Figure 4 A heatmap is also a type of dataset.

[0043] In addition, information acquisition points and parameters can be obtained, for example, through... Figure 3 , Figure 4 The table shown is in the form of a heatmap (as an example, the horizontal axis of the table represents the location of the information acquisition point, and the vertical axis represents the content of the corresponding numerical parameter). For example, the electron density, as a parameter, can be represented by a color indicating a specific value. Alternatively, the information acquisition point and parameter can be represented as a matrix.

[0044] (Determine Unit 22) Unit 22 uses the first model ( Figure 2The first learned model 220 determines the location of information acquisition points based on the three-dimensional structure of the molecule acquired by the first acquisition unit 16. This first model is generated through machine learning using training data 210, which includes the acquired three-dimensional structure of the molecule, the location of the information acquisition points, and a score (which may also include various parameters) of the analysis results of the molecule (described later). Specifically, the determining unit 22 can "input" the three-dimensional structure of the molecule acquired by the first acquisition unit 16 into the first learned model 220, causing the first learned model 220 to "output" the location of the information acquisition points with high scores. As a result, the setting unit 18 can set the location of the information acquisition points determined by the determining unit 22.

[0045] (Parameter type setting unit 24) Parameter type setting unit 24 uses the second model ( Figure 2 The second learned model 320 sets the types of parameters based on the three-dimensional structure of the molecule acquired by the first acquisition unit 16 or the positions of each information acquisition point set by the setting unit 18 (or determined by the determining unit 22). The second model is generated through machine learning using training data 310 that includes the acquired three-dimensional structure of the molecule, the positions of the information acquisition points, the types of parameters, and the scores of the analysis results of the molecule (described later). Specifically, the parameter type setting unit 24 can "input" the three-dimensional structure of the molecule acquired by the first acquisition unit 16 or the positions of each information acquisition point set by the setting unit 18 into the second learned model 320, causing the second learned model 320 to "output" the types of parameters with high scores. As a result, the second acquisition unit 20 can acquire data about one or more parameters set by the parameter type setting unit 24 at each information acquisition point.

[0046] (Score generation unit 26) To perform the aforementioned learning, the scoring generation unit 26 uses a dataset containing the three-dimensional structure, information acquisition points, and parameters of a molecule to generate scores as analysis results for different molecules. For example, suppose that the three-dimensional structures of molecule A and molecule B are similar. In this case, in a specific application verification, it is highly likely that the information acquisition points and parameters of one (molecule A) can be used as the information acquisition points and parameters of the other (molecule B). Specifically, the information acquisition points and parameter values ​​of molecule A are compared with the corresponding information acquisition points and parameter values ​​in the three-dimensional structure of molecule B. During the comparison, it is preferable that the positions of the information acquisition points are the same. As a result of the comparison, if the difference in parameter values ​​is small, a high score is generated for both the information acquisition points and parameter values. Conversely, if the difference in parameter values ​​is large, a low score is generated for both the information acquisition points and parameter values. In the case of a low score, the position of the information acquisition points is changed (by shifting the virtual three-dimensional curve, or changing the shape, spiral winding method, spacing, number, voxel size, etc.), and the parameter values ​​are acquired again. The score can be arbitrarily determined based on the difference in the compared parameter values, whether it is appropriately applicable to the verification application, etc. Furthermore, the information related to these scores can also be stored in conjunction with the aforementioned dataset.

[0047] (First model generation unit 200, second model generation unit 300) The first model generation unit 200 and the second model generation unit 300 respectively use training data 210 and 310 to generate a first learned model 220 and a second learned model 320 through machine learning. More specifically, the first model generation unit 200 enables the first learned model 220 to learn in such a way that if the "input" is the three-dimensional structure of the molecule acquired by the first acquisition unit 16, then the "output" is the location of the information acquisition point with the highest score. The second model generation unit 300 enables the second learned model 320 to learn in such a way that if the "input" is the three-dimensional structure of the molecule acquired by the first acquisition unit 16 or the locations of each information acquisition point set by the setting unit 18, then the "output" is the type of parameter with the highest score. There are no particular limitations as long as the input and output as described above can be performed, and various methods such as deep learning, random forest, and support vector machine can be used.

[0048] The first learned model 220 and / or the second learned model 320 can function either by executing a program stored in the storage unit 28 via the control unit 10, or by executing a predetermined program via a device different from the molecular analysis system 1. In the latter case, the first learned model 220 and / or the second learned model 320 generated by the different device are sent to the computer 3 and stored in the storage unit 40 of the computer 3.

[0049] [Information Processing Method 1] Next, refer to Figure 5 Information processing method 1 in the molecular analysis system 1 described above will be explained. Here, it is assumed that information acquisition points are set using a virtual three-dimensional curve (curve method), and the three-dimensional structure of the molecule is acquired within the molecular analysis system 1. In the following description, structural elements marked with the same reference numerals are the same as those described above, and repeated descriptions will be omitted as appropriate. It should be noted that the processing shown below is performed by executing a program stored in the storage unit 28, but the information processing method disclosed herein is not limited to this.

[0050] First, the control unit 10 (receiving unit 12) receives information about the chemical formula of the molecule (step S1).

[0051] Next, the control unit 10 (generation unit 14) generates the three-dimensional structure of the molecule by quantum chemical calculation based on the chemical formula of the received molecule (step S2).

[0052] Next, the control unit 10 (setting unit 18) sets the axis connecting the start and end points of the three-dimensional structure of the generated and acquired molecule (step S3). Here, as mentioned above, the setting unit 18 may also set the length direction in the three-dimensional structure of the generated molecule, and set the start and end points near the ends along the length direction.

[0053] Next, when the control unit 10 (setting unit 18) arranges information acquisition points around the axis, it sets a virtual three-dimensional curve from the set starting point to the ending point, and sets multiple information acquisition points at intervals on the set virtual three-dimensional curve (step S4). Alternatively, multiple information acquisition points can be set inside or around the three-dimensional structure of the molecule, or they can be set in a way that maximizes the straight-line distance between the starting point and the ending point. In one embodiment, in this step, the virtual three-dimensional curve ( Figure 3 ).

[0054] Next, the control unit 10 (second acquisition unit 20) acquires data about one or more preset parameters at each set information acquisition point (step S5).

[0055] [Information Processing Method 2] Next, refer to Figure 6 The information processing method 2 in the molecular analysis system 1 will be explained. It should be noted that this is based on the premise of using voxels to set information acquisition points (voxel method) and obtaining information about the three-dimensional structure of molecules from the outside.

[0056] First, the control unit 10 (first acquisition unit 16) obtains the three-dimensional structure of the molecule through quantum chemical calculation based on the chemical formula of the molecule received by the receiving unit 12 (step S11) (corresponding to steps S1-S2).

[0057] Next, the control unit 10 (setting unit 18) divides the three-dimensional structure of the molecule acquired by the first acquisition unit 16 into a defined cube (step S12).

[0058] Next, the control unit 10 (setting unit 18) sets information acquisition points at the intersection of each cube (step S13).

[0059] Next, the control unit 10 (second acquisition unit 20) acquires data about one or more preset parameters at each set information acquisition point (step S14).

[0060] [Uses and Application Examples] As described above, the three-dimensional structure, information acquisition points, and parameters obtained through information processing methods 1 and 2 can be used to analyze the application of new molecules. For example, they can be used as training data 210 to generate a first learned model 220 through machine learning. With such a first learned model 220, the positions of appropriate information acquisition points can be determined based on the three-dimensional structure of the new molecule. Furthermore, the three-dimensional structure, information acquisition points, and parameters obtained through information processing methods 1 and 2 can be used as training data 310 to generate a second learned model 320 through machine learning. With such a second learned model 320, appropriate types of parameters can be set based on the three-dimensional structure of the new molecule or the positions of arbitrarily determined information acquisition points in the first learned model 220.

[0061] Furthermore, in the prior art described in Patent Document 1, simulations and preprocessing are performed to evaluate the physical properties of the compound itself. However, because all useless data is extracted, the computational cost becomes enormous (sometimes only useful for academic compound exploration and not practical). Additionally, determining a stable structure from a compound and then only obtaining data for that stable structure cannot address structural flexibility. Moreover, intermolecular interactions are not a prerequisite. Furthermore, the one-dimensional, two-dimensional, and three-dimensional structural information in existing methods is only meaningful from a chemical perspective and cannot be correlated with behavioral predictions within biological systems. Importantly, the spatial information surrounding the compound is crucial for reactions (binding) with proteins, etc.

[0062] To address the aforementioned issues, the invention disclosed herein describes the space surrounding a compound using information acquisition points. This method has not been used in any other field. Conventional methods place the molecule in its most structurally stable state within space and extract information, while the method of this disclosure defines the molecule by the space surrounding it (where information acquisition points are set), representing a completely opposite approach. Therefore, it is not constrained by structurally stable states. Furthermore, it can accurately capture molecular characteristics at low computational cost. Moreover, utilizing molecular structural characteristics is expected to have industrial applications in a wide range of fields beyond drug development.

[0063] As an example of the application of the invention disclosed herein, the design of new drugs can be cited. The similarity of the extracted chemical information can be calculated to create drugs with properties similar to the desired drug (ligand-based drug development). Furthermore, if a machine learning model is constructed using the chemical information extracted from the compound as input data, and the interaction between the compound and biological substances such as proteins, pharmacokinetic indicators such as metabolism and absorption of the compound in vivo, and the pharmacodynamic effects and side effects produced by the compound as output data, then the present invention can be used to predict various behaviors of unknown compounds in vivo. In addition, as an example of application, the prediction of compound solubility can be cited. Specifically, solubility can be predicted based on the ratio of hydrophobic to hydrophilic sites around the molecule. Furthermore, it can also be applied to the field of materials science. That is, according to this disclosure, the development of new materials such as plastics, coatings, rubber, and fibers is expected. Quantitative structure-activity relationships can also be studied based on the chemical information extracted as parameters, and new materials with desired properties can be designed.

[0064] Regarding the above application examples, machine learning can be used to determine and predict the uses of molecules. That is, it is possible to obtain experimental and predicted values ​​for a wide variety of molecules and evaluate the suitability of their uses through scoring. If data such as information acquisition points, parameters, and scores can be used as training data, the machine learning model inputting the data outputs a score, thereby determining whether the molecule is suitable for drug development, materials, or other applications, and whether the information acquisition point settings are appropriate. When calculating scores on new datasets, it is preferable to standardize and unify the information acquisition point settings (e.g., helices, voxel spacing, etc.) regardless of the molecule. If the score is poor, an attempt is made to reconfigure the information acquisition points. On the other hand, if the score (accuracy) is good, the information acquisition point-based judgment disclosed herein is good; for example, the molecule can be suggested as a lead compound candidate for a specific disease.

[0065] Furthermore, according to this disclosure, since the information acquisition points and parameters are related to the compound structure of the molecule, it is also expected that new drugs or new materials with three-dimensional structures that may also receive high scores for the same purpose can be predicted based on the information acquisition points and parameters that receive high scores for a certain purpose.

[0066] In the molecular analysis system 1 (computer), program, and information processing method disclosed herein, which are constituted by the above, the following are provided: a first acquisition unit 16, which acquires information on the three-dimensional structure of a molecule obtained by quantum chemical calculation; a setting unit 18, which sets multiple information acquisition points inside or around the three-dimensional structure of the molecule; and a second acquisition unit 20, which acquires data about one or more pre-set parameters at each information acquisition point set by the setting unit 18.

[0067] The invention disclosed herein extracts spatial information of molecules using a novel method not previously known. Since the binding capacity of a molecule depends on the presence of electrons surrounding it, the binding capacity of the molecule can be predicted based on the spatial information (particularly the information acquisition points and parameters) obtained geometrically through this disclosure, thereby supporting research and development related to the applications of molecules in drug development, new materials, and other fields. Furthermore, because molecules are identified by acquiring data along their periphery, a wide variety of information can be represented.

[0068] Furthermore, in the molecular analysis system 1, program, and information processing method disclosed herein, the setting unit 18 can also divide the three-dimensional structure of the molecule acquired by the first acquisition unit 16 into a defined cube, and set information acquisition points for each cube. As described above, according to this method, information acquisition points can be easily set.

[0069] Furthermore, in the molecular analysis system 1, program, and information processing method disclosed herein, the setting unit 18 may also set an axis connecting the start and end points of the three-dimensional structure of the molecule acquired by the first acquisition unit 16. When arranging information acquisition points around the axis, a virtual three-dimensional curve from the set start point to the end point is set, and multiple information acquisition points are set at intervals on the set virtual three-dimensional curve. Thus, the minimum required number of information acquisition points can be set according to the three-dimensional structure of the molecule.

[0070] Furthermore, in the molecular analysis system 1, program, and information processing method disclosed herein, the setting unit 18 may also set a length direction in the three-dimensional structure of the molecule acquired by the first acquisition unit 16, and set a start point and an end point near the end along this length direction. Additionally, the setting unit 18 may also set the start point and end point in a manner that maximizes the straight-line distance between them. Thus, an appropriate method for setting the start point and end point can be selected based on the three-dimensional structure of the molecule.

[0071] Furthermore, in the molecular analysis system 1, program, and information processing method disclosed herein, the setting unit 18 can also set a virtual three-dimensional curve in a manner that covers part or all of the three-dimensional structure of the molecule acquired by the first acquisition unit 16. Thus, the amount of information in the parameter values ​​can be increased or decreased based on the structure and size of the analyzed molecule.

[0072] Furthermore, in the molecular analysis system 1, program, and information processing method disclosed herein, the virtual three-dimensional curve set by the setting unit 18 can also be helical. This allows for the setting of required information acquisition points based on the three-dimensional structure of the molecule. Additionally, a helical shape can reflect the differences between enantiomers.

[0073] Furthermore, in the molecular analysis system 1, program, and information processing method disclosed herein, the setting unit 18 can also set multiple information acquisition points at equal intervals on the virtual three-dimensional curve. If the interval width is reduced, comprehensive information can be collected. Conversely, if the interval width is increased, only the minimum required information can be collected.

[0074] Furthermore, the molecular analysis system 1, program, and information processing method disclosed herein may also include: a receiving unit 12 for receiving information about the chemical formula of a molecule; and a generating unit 14 for generating the three-dimensional structure of the molecule through quantum chemical calculations based on the chemical formula of the molecule received by the receiving unit 12. As described above, the three-dimensional structure of the analyzed molecule can be obtained not only from external sources but also generated by the molecular analysis system 1.

[0075] Furthermore, in the molecular analysis system 1, program, and information processing method disclosed herein, a determining unit 22 may also be provided. This determining unit 22 uses a first model (first learned model 220) to determine the location of information acquisition points based on the three-dimensional structure of the molecule acquired by the first acquisition unit 16. The first model is generated by machine learning using training data 210 that includes the three-dimensional structure of the molecule, the location of the information acquisition points, and scores of the analysis results of the molecule. Therefore, the setting unit 18 can also set the location of the information acquisition points determined by the determining unit 22.

[0076] Furthermore, in the molecular analysis system 1, program, and information processing method disclosed herein, a parameter type setting unit 24 may also be provided. This parameter type setting unit 24 uses a second model (second learned model 320) to set the parameter types based on the three-dimensional structure of the molecule acquired by the first acquisition unit 16 or the positions of each information acquisition point set by the setting unit 18 (or the determining unit 22). The second model is generated by machine learning using training data 310 that includes the three-dimensional structure of the molecule, the positions of the information acquisition points, the parameter types, and scores of the analysis results of the molecule. Thus, the second acquisition unit 20 can also acquire data on one or more parameters set by the parameter type setting unit 24 at each information acquisition point.

[0077] Furthermore, the molecular analysis system 1, program, and information processing method disclosed herein are not limited to the above-described methods or combinations, and various modifications can be applied.

[0078] For example, although it is described that the second acquisition unit 20 acquires data about one or more preset parameters at each information acquisition point set by the setting unit 18, in curve mode, voxel mode, etc., it is also possible not to set parameters at a certain information acquisition point. That is, if more than the required information acquisition points are set, it may not be necessary to calculate parameter values ​​at all information acquisition points.

Claims

1. A molecular analysis system, characterized in that, have: The first acquisition unit acquires the three-dimensional structural information of molecules obtained through quantum chemical calculations; The unit is set up by setting multiple information acquisition points inside or around the three-dimensional structure of the molecule; as well as The second acquisition unit acquires data about one or more preset parameters at each of the information acquisition points set by the setting unit.

2. The molecular analysis system according to claim 1, characterized in that, The setting unit divides the three-dimensional structure of the molecule obtained by the first acquisition unit into a defined cube, and sets information acquisition points for each cube.

3. The molecular analysis system according to claim 1, characterized in that, The setting unit sets an axis connecting the start and end points of the three-dimensional structure of the molecule acquired by the first acquisition unit. When arranging information acquisition points around the axis, it sets a virtual three-dimensional curve from the set start point to the set end point, and sets multiple information acquisition points at intervals on the set virtual three-dimensional curve.

4. The molecular analysis system according to claim 3, characterized in that, The setting unit sets a length direction in the three-dimensional structure of the molecule obtained by the first acquisition unit, and sets the start point and the end point near the end along the length direction.

5. The molecular analysis system according to claim 3, characterized in that, The setting unit sets the parameters in a manner that maximizes the straight-line distance between the starting point and the ending point.

6. The molecular analysis system according to claim 3, characterized in that, The setting unit sets the virtual three-dimensional curve in a manner that covers part or all of the three-dimensional structure of the molecule acquired by the first acquisition unit.

7. The molecular analysis system according to claim 3, characterized in that, The setting unit sets multiple information acquisition points at equal intervals on the virtual three-dimensional curve.

8. The molecular analysis system according to claim 1, characterized in that, It also has: The receiving unit receives information about the chemical formula of the molecule; and The generating unit generates the three-dimensional structure of the molecule through quantum chemical calculations based on the chemical formula of the molecule received by the receiving unit.

9. The molecular analysis system according to claim 1, characterized in that, It also has: The unit is determined by using a first model to determine the location of the information acquisition point based on the three-dimensional structure of the molecule obtained by the first acquisition unit. The first model is generated by machine learning using training data that includes the three-dimensional structure of the molecule, the location of the information acquisition point, and the score of the analysis results about the molecule. The setting unit sets the location of the information acquisition point determined by the determining unit.

10. The molecular analysis system according to claim 1, characterized in that, It also has: The parameter type setting unit uses a second model to set the parameter types based on the three-dimensional structure of the molecule obtained by the first acquisition unit or the positions of each information acquisition point set by the setting unit. The second model is generated by machine learning using training data that includes the three-dimensional structure of the molecule, the positions of the information acquisition points, the types of parameters, and the scores of the analysis results of the molecule. The second acquisition unit acquires data about one or more parameters set by the parameter type setting unit at each information acquisition point.

11. A program that functions as a first acquisition unit, a setting unit, and a second acquisition unit by being executed by a computer, characterized in that, The first acquisition unit acquires information about the three-dimensional structure of the molecule obtained through quantum chemical calculations. The setting unit sets multiple information acquisition points inside or around the three-dimensional structure of the molecule. The second acquisition unit acquires data about one or more preset parameters at each of the information acquisition points set by the setting unit.

12. A molecular analysis method, executed by a computer, characterized in that, It has the following processes: To obtain information about the three-dimensional structure of molecules obtained through quantum chemical calculations; Multiple information acquisition points are set inside or around the three-dimensional structure of the molecule; as well as Acquire data about one or more preset parameters at each of the set information acquisition points.

Images