Molecular force field parameter determination method, device and equipment and readable storage medium

By multi-level encoding of the structural information of molecules, encoding representations are generated to describe atoms and their chemical environment, the problems of finite atom types and inflexible parameter binding are solved, and accurate matching and efficient query of molecular force field parameters are achieved.

CN118942558APending Publication Date: 2024-11-12DOUYIN VISION CO LTD +1

Patent Information

Application Number
CN202411060355.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

The existing molecular force field parameter determination methods have problems such as limited atom types, inflexible parameter binding and incorrect identification, making it difficult to effectively describe the chemical environment of complex molecules and predict their properties.

Method used

By multi-level encoding of the structural information of the target molecule, an encoded representation is generated, where the characteristic information of each level is attached on the basis of the previous adjacent level to flexibly describe the atoms and their chemical environment, thereby determining the molecular force field parameters. .

Benefits of technology

It improves the accuracy and flexibility of molecular force field parameter matching, reduces complexity, optimizes the matching process, and is suitable for efficient query of large-scale molecular data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118942558A_ABST
    Figure CN118942558A_ABST
Patent Text Reader

Abstract

The embodiment of the invention provides a molecular force field parameter determination method, device and equipment and a readable storage medium. The method includes acquiring structural information of a target molecule. Encoding the structural information to obtain an encoded representation of the target molecule, the encoded representation comprising encoded representations corresponding to a plurality of hierarchies, and a specified hierarchy in the plurality of hierarchies indicating feature information of at least one atom included in the target molecule, and the feature information indicated by the specified hierarchy includes feature information added on the basis of the feature information indicated by the previous adjacent hierarchy. And determining a molecular force field parameter corresponding to the target molecule based on the coded representation of the target molecule. Therefore, through coding representation corresponding to multiple levels, the complexity of molecular force field parameter matching is reduced, and the efficiency is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] Example embodiments of the present disclosure generally relate to the field of computers, and more particularly to methods, devices, apparatuses, and readable storage media for determining molecular force field parameters. Background Art

[0002] In the fields of chemistry and bioinformatics, the representation and analysis of molecular structures are important steps in studying the properties and behaviors of molecules. Molecular force fields are tools for describing the potential energy surface of molecules, that is, given the three-dimensional coordinates of each atom in a molecule, the energy of the molecule can be predicted. Common molecular force field generation methods artificially formulate rules, specify atom types according to the chemical environment, and obtain force field parameters such as bonds, bond angles, and dihedral angles based on the combination of atom types. However, these methods have problems such as limited atom types, inflexible parameter binding, and misidentification. Summary of the invention

[0003] In a first aspect of the present disclosure, a method for determining molecular force field parameters is provided. The method may include: obtaining structural information of a target molecule. Encoding the structural information to obtain a coded representation of the target molecule, the coded representation including coded representations corresponding to multiple levels, a specified level among the multiple levels indicating characteristic information of at least one atom included in the target molecule, and the characteristic information indicated by the specified level includes additional characteristic information based on the characteristic information indicated by the previous adjacent level. Based on the coded representation of the target molecule, determine the molecular force field parameters corresponding to the target molecule.

[0004] In a second aspect of the present disclosure, a device for determining molecular force field parameters is provided. The device may include: a structural information determination module, configured to obtain structural information of a target molecule. An encoding module, configured to perform encoding on the structural information to obtain an encoded representation of the target molecule, the encoded representation including encoded representations corresponding to multiple levels, a specified level among the multiple levels indicating characteristic information of at least one atom included in the target molecule, and the characteristic information indicated by the specified level includes additional characteristic information based on the characteristic information indicated by the previous adjacent level. The molecular force field parameter determination module is configured to determine the molecular force field parameters corresponding to the target molecule based on the encoded representation of the target molecule.

[0005] In a third aspect of the present disclosure, an electronic device is provided. The device includes at least one processing unit; and at least one memory, the at least one memory is coupled to the at least one processing unit and stores instructions for execution by the at least one processing unit. When the instructions are executed by the at least one processing unit, the electronic device executes the method of the first aspect.

[0006] In a fourth aspect of the present disclosure, a computer-readable storage medium is provided, wherein a computer program is stored on the medium, and when the computer program is executed by a processor, the method of the first aspect is implemented.

[0007] In a fifth aspect of the present disclosure, a computer program product is provided, wherein the computer program product comprises computer executable instructions, wherein the computer executable instructions implement the method of the first aspect when executed by a processor.

[0008] It should be understood that the content described in this section is not intended to limit the key features or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0010] Figure 1 A schematic diagram showing an example environment in which embodiments of the present disclosure can be implemented;

[0011] Figure 2 A flow chart showing a method for determining molecular force field parameters according to some embodiments of the present disclosure;

[0012] Figure 3 shows an example diagram of a target molecule according to some embodiments of the present disclosure;

[0013] Figure 4 An example diagram showing an example structural fragment according to some embodiments of the present disclosure;

[0014] Figure 5 Schematic diagrams showing example structural fragments according to other embodiments of the present disclosure;

[0015] Figure 6 Schematic diagrams showing example structural fragments according to further embodiments of the present disclosure;

[0016] Figure 7 A schematic structural block diagram of a molecular force field parameter determination device according to some embodiments of the present disclosure is shown; and

[0017] Figure 8 A block diagram of an electronic device in which one or more embodiments of the present disclosure may be implemented is shown. DETAILED DESCRIPTION

[0018] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0019] In the description of the embodiments of the present disclosure, the term "including" and similar terms should be understood as open inclusion, that is, "including but not limited to". The term "based on" should be understood as "based at least in part on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The term "some embodiments" should be understood as "at least some embodiments". Other explicit and implicit definitions may also be included below.

[0020] Herein, unless explicitly stated, executing a step “in response to A” does not mean executing the step immediately after “A” but may include one or more intermediate steps.

[0021] It is understandable that the data involved in this technical solution (including but not limited to the data itself, the acquisition, use, storage or deletion of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.

[0022] It is understandable that before using the technical solutions disclosed in the various embodiments of the present disclosure, the types, scopes of use, usage scenarios, etc. of the information involved in the present disclosure should be informed to relevant users and their authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations. The relevant users may include any type of right holders, such as individuals, enterprises, and groups.

[0023] For example, in response to receiving an active request from a user, a prompt message is sent to the relevant user to clearly prompt the relevant user that the operation requested to be performed will require obtaining and using the information of the relevant user, so that the relevant user can independently choose whether to provide information to software or hardware such as an electronic device, application, server or storage medium that executes the operation of the technical solution of the present disclosure based on the prompt message.

[0024] As an optional but non-limiting implementation, in response to receiving an active request from a relevant user, a prompt message is sent to the relevant user, for example, in the form of a pop-up window, in which the prompt message may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to select "agree" or "disagree" to provide information to the electronic device.

[0025] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that meet relevant laws and regulations may also be applied to the implementation of the present disclosure.

[0026] As used herein, the term "model" can learn the association between the corresponding input and output from the training data, so that after the training is completed, the corresponding output can be generated for a given input. The generation of the model can be based on machine learning technology. Deep learning is a machine learning algorithm that processes inputs and provides corresponding outputs by using multi-layer processing units. A neural network model is an example of a model based on deep learning. In this article, "model" may also be referred to as "machine learning model", "learning model", "machine learning network" or "learning network", and these terms are used interchangeably in this article.

[0027] Figure 1 A schematic diagram of an example environment 100 in which embodiments of the present disclosure can be implemented is shown. The environment 100 includes an electronic device 110. It is desirable to use such an electronic device 110 to determine a molecular force field parameter 120 corresponding to a target molecule 102. To this end, in some implementations, a machine learning model 150 may be deployed in the electronic device 110 for determining the molecular force field parameter 120 of the target molecule.

[0028] Exemplarily, the electronic device 110 may include any computing system with computing capabilities, such as various computing devices / systems, terminal devices, server devices, etc. The terminal device may be any type of mobile terminal, fixed terminal or portable terminal, including a mobile phone, a desktop computer, a laptop computer, a notebook computer, a netbook computer, a tablet computer, a media computer, a multimedia tablet, a handheld computer, a portable game terminal, a game device, or any combination of the foregoing, including accessories and peripherals of these devices or any combination thereof. The server device may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content distribution networks, and big data and artificial intelligence platforms. The server device may include, for example, a computing system / server, such as a mainframe, an edge computing node, a computing device in a cloud environment, and the like. It should be understood that the structure and function of each element in the environment 100 are described for exemplary purposes only, without implying any limitation on the scope of the present disclosure.

[0029] At present, in the field of molecular force field parameter determination, the commonly used method is based on an artificial rule system to assign an atomic type to each atom in the target molecule according to its chemical environment, and obtain the corresponding molecular force field parameters such as bonds, bond angles and dihedral angles according to the combination of atomic types. These molecular force field parameters are used to calculate the energy of molecules under specific three-dimensional structures. For example, the CH bond can indicate the bond length between carbon atoms and hydrogen atoms. The molecular force field parameters are used to describe the length of the CH bond and its effect on energy. Similarly, the bond angle or dihedral angle can describe the atoms that form the bond angle or dihedral angle, the size of the bond angle or dihedral angle, and the effect of the bond angle or dihedral angle on energy. Molecular force field parameters can describe the energy changes of molecular fragments, thereby predicting the stability and reactivity of molecules. In traditional force field models, by assigning different parameters to these specific structures (such as bonds, bond angles and dihedral angles), the physical and chemical properties of molecules can be simulated more accurately.

[0030] There are three main problems with the commonly used methods. The first is that the types of atoms that can be used by the manually formulated rule system are very limited. As the data set increases, the number of force field parameters is difficult to grow at a synchronous rate. For example, for common atom types such as carbon, hydrogen, oxygen, and nitrogen, corresponding molecular force field parameters can be formulated. However, if it is a more complex molecular structure or a newly added molecular result, it may not be able to cope with it.

[0031] The second is that molecular force field parameters such as bonds and bond angles are strongly bound to the atom type and lack flexibility. For example, for ethanol (CH3CH2OH), there are corresponding molecular force field parameters to describe the carbon-oxygen bonds in ethanol. However, if you encounter molecules similar to ethanol, such as methyl ether (CH3OCH3), the same molecular force field is no longer applicable.

[0032] The third is that the bond type obtained by combining atom types may lead to incorrect identification of certain specific chemical structures, thus affecting the accuracy. For example, the carbon-carbon bond in the benzene ring has conjugated properties, rather than a simple single bond or double bond. In the commonly used methods, this special structure may not be correctly identified, resulting in incorrect force field parameter matching.

[0033] For example, any molecule can be represented by a graph, that is, G = (V, E), where V can represent the set of atoms in the molecule and E can represent the set of chemical bonds between two atoms. i It can represent the i-th atom in a molecule, which has characteristics such as element type, number of connections, aromaticity, smallest ring, number of ring connections, etc. i ∈V. E can represent the connection of the i-th atom v i and the jth atom v jThe chemical bond between them has characteristics such as bond order (single, double, triple or aromatic) and whether it is in the ring. The local part of the molecule can be represented as a derived subgraph, that is, G'=G[V']=(V',E'), G' can represent an induced subgraph, i.e., a local region or substructure of a molecule. V' can represent the set of nodes in the induced subgraph G', i.e., the atoms in the local region. E' can represent the set of edges in the induced subgraph G', i.e., the chemical bonds in the local region. i ' can be the i-th atom in the derived subgraph G' and has the corresponding partial features. ij ' can be the connected atom v in the derived subgraph G' i ' and the atom v j 'Chemical bonds between and have corresponding partial characteristics.

[0034] Only fully connected subgraphs are considered. In a chemical environment, this means that all atoms and the bonds between them are interconnected, with no isolated atoms or broken bonds. Based on this, the chemical environment can be defined as: f(G') = (f v (V'), f e (E')), f v and f e Act on atoms and bonds separately, retaining some of their characteristics. v (V') can form a feature set corresponding to the partial features of all atoms in the subgraph, f e (E') can form a feature set corresponding to the partial features of all chemical bonds in the subgraph.

[0035] f(G') can be understood as a collection of different subgraphs. Considering that G' must satisfy the topological relationship and atomic and bond characteristics required by f(G'), G'∈f(G'). At the same time, since f v and f e Only some features of atoms and bonds are retained, so if any subgraph G" differs from G' only in the unretained features, then G"∈f(G') is also true.

[0036] The essence of the force field is to establish a mapping between f(G') and the force field parameters θ. However, if f(G') itself is recorded directly, for any given molecule, graph matching is required for all recorded chemical environments. Graph matching itself is a complex problem, and it must be multiplied by the number of recorded force field parameters. It can be seen that this process is very inefficient.

[0037] With the development of computing power, the number of molecules obtained by quantitative calculation can reach the order of millions, and the number of conformations can reach the order of tens of millions. Correspondingly, the force field also needs more parameters to provide accurate modeling for these different chemical environments. Therefore, there are two main problems: how to automatically generate and record the description of f(G') and the corresponding force field parameters based on massive data. When the number of force field parameters is particularly large, how to efficiently query the corresponding parameters given a chemical structure.

[0038] The embodiments of the present disclosure propose a scheme for determining molecular force field parameters. According to the scheme, the structural information of the target molecule is obtained. The structural information is encoded to obtain a coded representation of the target molecule, the coded representation includes coded representations corresponding to multiple levels, a specified level among the multiple levels indicates characteristic information of at least one atom included in the target molecule, and the characteristic information indicated by the specified level includes additional characteristic information based on the characteristic information indicated by the previous adjacent level. Based on the coded representation of the target molecule, the molecular force field parameters corresponding to the target molecule are determined.

[0039] Therefore, through multiple levels of coding representation, atoms and their chemical environment can be flexibly described, thus solving the problem of limited atom types. The characteristic information indicated by each level is additional characteristic information based on the previous adjacent level, making parameter generation and matching more accurate and flexible. In addition, through the coding representation corresponding to multiple levels, the complexity of molecular force field parameter matching is reduced, the efficiency is improved, and the matching process of molecular force field parameters is optimized.

[0040] The embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. Figure 2 200 according to some embodiments of the present disclosure. The process 200 may be implemented at the electronic device 110. For ease of discussion, the following will refer to Figure 1 The process 200 is described below.

[0041] like Figure 2 As shown, in block 201, the electronic device 110 may obtain structural information of the target molecule by traversing the atoms in the target molecule. For example, a root node may be specified in the target molecule, so that the atoms in the target molecule are traversed based on the root node.

[0042] Figure 3A schematic diagram of a target molecule 300 according to some embodiments of the present disclosure is shown. Take the traversal of a molecular structure fragment of the target molecule 300 as an example. The node labeled 2 can be used as the root node. The root node includes 3 branches, and branch 1 is traversed to the node labeled 7. Branch 2 is traversed to the node labeled 3. Branch 3 is traversed to the node labeled 0. That is, starting from the root node, each atom contained in the molecular fragment structure can be traversed, and structural information can be generated. The structural information may include attribute information of the atoms passed by the traversal path. Exemplarily, the attribute information may indicate the characteristic information of the atom, for example, the characteristic information may indicate the element type, aromaticity, number of connections, number of ring connections, and the smallest ring in which the atom is located, etc. For example, for an atom labeled 3, the attribute information of the atom may indicate: the atom is labeled 3, the element type of the atom is an oxygen atom, and the atom is bonded to the atom labeled 2 by a double bond. It should be noted that hydrogen atoms are omitted in the presentation of the molecular fragment structure involved in the present disclosure and in the encoding process, and the purpose of the omission is to save the encoding amount.

[0043] In box 202, the electronic device 110 performs encoding on the structural information to obtain an encoded representation of the target molecule, the encoded representation includes encoded representations corresponding to multiple levels, a specified level among the multiple levels indicates characteristic information of at least one atom included in the target molecule, and the characteristic information indicated by the specified level includes additional characteristic information based on the characteristic information indicated by the previous adjacent level.

[0044] Based on the structural information of the target molecule, the structural information can be encoded to obtain a coded representation. The coded representation may include multiple levels of coded representations. Each level may include encoding at least one feature information of the target molecule, and the feature information indicated by the current level includes feature information additional to the feature information indicated by the previous adjacent level.

[0045] For example, if the feature information indicated by the previous adjacent level of the current level includes the feature information of atom A, the feature information of the current level may include the feature information of atom A and atom B, and atom B is bonded to atom A. For another example, if the feature information indicated by the previous adjacent level of the current level includes the feature information of the element type of atom A, the feature information of the current level may include the feature information of the element type and the number of connections of atom A. The number of levels may be set according to actual conditions. In some embodiments, the number of levels may be set to 3.

[0046] The coding can be performed on the structural information of the target molecule using the coding rules corresponding to the concise molecular linear input search (SMiles Arbitrary Target Specification, SMARTS) to obtain a character string describing the local chemical language, which can be used by the force field for parameter matching. In the present disclosure, some functions of SMARTS can be retained. Specifically, the description of atoms only retains the element type (corresponding to the character string "#"), the number of connections (corresponding to the character string "X"), the aromaticity (corresponding to the character string "A / a"), the smallest ring (corresponding to the character string "r"), the number of ring connections (corresponding to the character string "x"), etc., which can be described by a number or character. For the description of chemical bonds, the bond order (for example, the symbol "-", the symbol "=", the symbol "#", the symbol ":" respectively represent single bonds, double bonds, triple bonds and aromatic bonds) and whether it is intra-ring (the symbol "@" and the symbol "!@" can represent intra-ring bonds and non-intra-ring bonds respectively) and other features can also be represented by the symbol "~" to represent any bond. In addition, fuzzy features are deleted. For example, if a feature is described by a numerical range, it can be considered a fuzzy feature and can be discarded. For example, for a ring, if the encoding rule includes less than or greater than an N-membered ring (N is a positive integer), it belongs to the feature described by the numerical range, which can be discarded. On the contrary, if it is a definite numerical value such as a three-membered ring or a five-membered ring, it can be retained.

[0047] Based on the above process, different levels of coding representation (SMARTS) can be constructed. If based on the original coding representation, more descriptions of features such as atoms or chemical bonds are added to atoms, or atoms and bonds are added to obtain new coding representations. For example, the coding representation corresponding to the atoms indicated at the top level is "[#6]-[#6]". In the second level, by adding the features of the number of atomic connections to the atoms at the top level, the coding representation of the second level can be obtained. For example, the coding representation of the second level can be "[#6X4]-[#6X4]". The structure that can be matched by the coding representation corresponding to the new level must be a subset of the coding representation that can be matched by the coding representation corresponding to the previous adjacent level. Secondly, it is possible to determine whether the coding representations of two molecules are the same. The coding representations of the two molecules are reduced to graph structures. If the two graphs are isomorphic and the corresponding atomic and bond descriptors are the same, the coding representations of the two molecules are considered to be the same.

[0048] In block 203, the electronic device 110 determines the molecular force field parameters corresponding to the target molecule based on the coded representation of the target molecule. The coded representation of the target molecule can be used as a language for describing the molecular structure. The coded representation can represent the characteristic information of the target molecule. In some embodiments, the coded representation describes a complex target molecule through a specific string pattern.

[0049] Force field parameters can be used to describe the interactions between atoms within a molecule, including bond lengths, bond angles, dihedral angles, and non-bonded interactions. These force field parameters are determined through experimental data and computational chemistry methods and are used to simulate molecular structure and behavior.

[0050] A matching relationship between force field parameters and coded representations can be pre-constructed. The coded representations and their corresponding force field parameters are recorded in the matching relationship. The coded representations are used to reflect the structure in the molecule, and the electronic device 110 can find the force field parameters corresponding to the coded representations according to the matching relationship table and apply them to the molecular simulation.

[0051] The coding representation of the target molecule includes coding representations corresponding to multiple levels. Therefore, in the process of determining the molecular force field parameters, it can be determined level by level. For example, starting from the top level, the coding representation of the top level is matched with the molecular force field parameters. Next is the matching of the second level, the matching of the third level, etc., until the corresponding sub-force field parameters cannot be matched. Finally, the force field parameters successfully matched for the last time can be used as the molecular force field parameters corresponding to the target molecule.

[0052] Through the above process, the structural information of the target molecule is hierarchically encoded, and the chemical environment of the target molecule can be described in detail, so as to more accurately capture the characteristics of the target molecule. This hierarchical encoding can ensure that all relevant chemical features of the target molecule are included, thereby solving the problem of limited atomic types and inflexible binding of molecular force field parameters. By performing the matching of force field parameters through hierarchical encoding, the corresponding molecular force field parameters can be found efficiently. For example, in the traditional matching method, the complexity corresponding to the matching process is linearly related to the input scale. That is, if the input scale is doubled, the complexity corresponding to the matching process will also be doubled. That is, in the traditional matching method, the complexity corresponding to the matching process can be expressed as O(n), and n can correspond to the input scale. If the method of determining the molecular force field parameters level by level corresponding to the present disclosure is adopted, the time consumed by the matching is proportional to the input scale. That is, if the input scale is doubled, the time consumed by the matching is only increased by a constant unit. That is, in the corresponding method of the present disclosure, the complexity corresponding to the matching process can be expressed as O(log(n)), and n can correspond to the input scale. Thus, the process of determining the molecular force field parameters is simplified.

[0053] In addition, through the above process, the advantages can be better utilized in scenarios where the number of molecules is in the millions or even tens of millions. For example, for molecules in the millions, if a one-by-one matching method is used, the amount of calculation will be very large and time-consuming. The multi-level matching method can significantly reduce the number of matches and the complexity of calculations through level-by-level matching. Specifically, the multi-level structure can decompose the complex matching process into multiple simpler level matches, and each level of matching only needs to process a small amount of data, which can greatly improve the matching efficiency. At the top level, the matching is relatively rough structural information, while at the deeper levels, the matching is more detailed and specific structural information. This layer-by-layer refinement of matching can ensure that the final matching results are more accurate and precise, which is especially important for large-scale molecular data.

[0054] In one implementation, the electronic device 110 performs encoding on the structural information in two cases. One is to perform encoding on the top level of multiple levels, and the other is to perform encoding on other levels other than the top level. For the top level of multiple levels, a first structural fragment in the target molecule is obtained, and the first structural fragment includes at least one atom. The characteristic information of the at least one atom included in the first structural fragment is encoded to obtain the encoded representation corresponding to the top level. For the non-top levels of multiple levels, a second structural fragment in the target molecule is obtained, and the second structural fragment is determined on the basis of the structural fragment corresponding to the previous adjacent level of the non-top level. The characteristic information of the atoms included in the second structural fragment is encoded to obtain the encoded representation corresponding to the non-top level.

[0055] The feature information at different levels indicates the atomic situation of the target molecule. For example, for the top level of multiple levels, the first structural fragment in the target molecule can be selected. The first structural fragment is usually a single atom or a simple combination of atoms. The selection of the first structural fragment in the target molecule can be based on the scene or can be based on a specified selection order.

[0056] Combination Figure 3 As shown, the first structural fragment 301 may be a structural fragment of the target molecule 300 . Figure 4 A schematic diagram of a topological structure 400 of a first structural fragment 301 according to some embodiments of the present disclosure is shown. The first structural fragment 301 may be a basic structural fragment of a target molecule 300 . Figure 4In the example shown, the first structure fragment 301 includes a carbon atom, a nitrogen atom connected to the carbon atom, and another carbon atom. The characteristic information of each atom in the first structure fragment is encoded to obtain the encoded representation corresponding to the top level. Exemplarily, the encoded representation corresponding to the top level may include: "[#6X3x0:2](-[#6X4x0:1])-[#7X3x2:3]". Correspondingly, the encoding may indicate the carbon atom (C:2) at the 2nd position, trivalent, without forming a ring. The carbon atom (C:1) at the 1st position, tetravalent, without forming a ring, connected to the carbon atom (C:2) at the 2nd position. The nitrogen atom at the 3rd position, trivalent, with two lone pairs of electrons, is connected to the carbon atom (C:2) at the 2nd position.

[0057] The characteristic information of each atom in the first structural fragment may include the chemical element type, bonding situation, aromaticity, number of intra-ring connections, etc. of the atom. Such information will be used as the top-level coding representation to describe the basic characteristics of the target molecule.

[0058] For a non-top level of multiple levels, encoding can be performed on the basis of the previous adjacent level of the non-top level. Specifically, a second structure fragment is obtained from the structure fragment corresponding to the previous level, and a new atom or atom combination is added thereto. The feature information of the expanded atom is encoded to obtain the encoding representation corresponding to the non-top level. Take the non-top level as the second level after the top level as an example. Figure 5 A schematic diagram of a topological structure 500 of a second structural fragment according to some embodiments of the present disclosure is shown. The second structural fragment is a new structural fragment obtained by adding atoms on the basis of the first structural fragment. Figure 5 In the example shown, new atoms are added to the first structure fragment 301 to obtain a second structure fragment. For example, the second structure fragment adds a nitrogen atom bonded to the first carbon atom (C:1) and an oxygen atom bonded to the second carbon atom (C:2), and forms a double bond. The second structure fragment is encoded to obtain the encoding representation corresponding to the second level. Exemplarily, the encoding representation corresponding to the second level can be: "[#6X3x0:2](-[#6X4x0:1](-[#6X4x0])-[#7X4x0])(-[#7X3x2:3](-[#6X4x2])-[#6X4x2])=[#8X1x0]".

[0059] Correspondingly, the code can indicate that the carbon atom at the second position (C:2) is trivalent and does not form a ring. The carbon atom at the first position (C:1) is tetravalent and does not form a ring, and is connected to the carbon atom at the second position (C:2). The tetravalent carbon atom (C:4) connected to the first carbon atom (C:1) does not form a ring. The tetravalent carbon atom (C:4) is connected to a tetravalent nitrogen atom (N:5) and does not form a ring. The trivalent nitrogen atom at the third position has two lone pairs of electrons. The third nitrogen atom is connected to a tetravalent carbon atom (C:6), and the tetravalent carbon atom (C:6) is connected to a tetravalent carbon atom (C:7). The carbon atom at the second position (C:2) also forms a double bond with a monovalent oxygen atom.

[0060] On the basis of the second level, a third level may be included, and the third level may be a level after the second level. Figure 6 A schematic diagram of a topological structure 600 of a third structural fragment according to some embodiments of the present disclosure is shown. Figure 6 In the example shown, a new atom is added to the second structure fragment, thereby obtaining a third structure fragment. The third structure fragment is a new structure fragment obtained by adding atoms to the second structure fragment. By encoding the third structure fragment, the encoding representation corresponding to the third level can be obtained. Exemplarily, the encoding representation corresponding to the third level can be:

[0061] "[#6X3x0:2](-[#6X4x0:1](-[#6X4x0]-[#6X3x2])-[#7X4x0])(-[#7X3x2:3] (-[#6X4x2](-[#6X3x0])-[#6X4x2]-1)-[#6X4x2]-[#6X4x2]-1)=[#8X1x0]”.

[0062] The feature indicated by the code is a feature further added on the basis of the feature indicated at the second level. The feature is only for illustrative purposes and does not impose any limitation on the actual situation, so it will not be elaborated here.

[0063] Through the above-mentioned encoding method of molecular structure information, the corresponding encoding representation of each level can be obtained. These encoding representations can not only accurately describe the basic characteristics of the target molecule, but also expand layer by layer to fully cover the complex chemical environment of the molecule. The encoding representation of each level not only includes the characteristic information of the previous level, but also adds new characteristic information, making the encoding result more detailed and accurate.

[0064] In one implementation, the electronic device 110 may further include encoding the structural information by: for a top level of multiple levels, obtaining a chemical element type representation of a given atom in at least one atom indicated by the top level, the chemical element type representation being part of the characteristic information. Encoding the chemical element type representation of the given atom to obtain an encoded representation corresponding to the top level. For a non-top level of multiple levels, obtaining a structural type representation of a given atom in at least one atom indicated by the top level, the structural type representation being part of the characteristic information. Encoding the structural type representation and the chemical element type representation of the given atom to obtain an encoded representation corresponding to the non-top level.

[0065] In addition to expanding the structural fragments of the target molecule between different levels, the structural types of the target molecule can also be expanded between different levels. For example, for a target molecule with a relatively simple atomic composition, the atoms indicated at different levels are the same, but the difference is that the structural types of the atoms are different. For example, the atoms indicated at different levels are the same, corresponding to a carbon atom, a nitrogen atom connected to the carbon atom, and another carbon atom.

[0066] For the top level of the multiple levels, for the atoms in the target molecule, obtain the chemical element type of the atom (e.g., carbon, hydrogen, oxygen, etc.). Encode the chemical element type representation to obtain the coded representation corresponding to the top level. The coded representation can describe the basic components of the molecule. Exemplarily, the coded representation corresponding to the top level may include:

[0067] “[#6:2](-[#6:1])-[#7:3]”.

[0068] Correspondingly, the code can indicate that the carbon atom at the second position (C:2). The carbon atom at the first position (C:1) is connected to the carbon atom at the second position (C:2). The nitrogen atom at the third position (N:3) is connected to the carbon atom at the second position (C:2).

[0069] The reason why the coding representation corresponding to the top level in the current implementation is different from the coding representation corresponding to the top level in the aforementioned example is that the level of detail is different. For example, the coding representation corresponding to the top level in the aforementioned example includes features such as valence state, ring information, atom number, lone pair electron information, etc., while the coding representation corresponding to the top level in the current implementation only includes features such as atom number and atom type. This can cope with different scenarios. For example, including more feature representations can be applicable to scenarios that require precise description and analysis. Including fewer features can be applicable to scenarios that require rapid screening and preliminary matching.

[0070] For a non-top level among multiple levels, encoding can be performed on the basis of the previous adjacent level of the non-top level. Taking the second level as an example, based on the chemical element type representation of each atom of the previous level (top level), select at least one atom and obtain its structural type representation. The structural type representation may include bond level characteristic information (such as single bond, double bond, etc.), valence state characteristic information, etc. (such as divalent, trivalent, etc.). By encoding the chemical element type representation and bond level characteristic information of the atom, the coded representation corresponding to the second level can be obtained. Taking the addition of valence state characteristic information as an example, the coded representation corresponding to the second level may include: "[#6X3:2](-[#6X4:1])-[#7X3:3]".

[0071] Taking the third level as an example, based on the chemical element type representation of the atom and the bond level characteristic information, characteristic information such as the number of ring connections and lone pair electron information can be added. By encoding the chemical element type representation, bond level characteristic information and ring connection number characteristic information, the corresponding coded representation of the third level can be obtained. Taking the addition of features such as the number of ring connections and lone pair electron information as an example, the corresponding coded representation of the third level can include:

[0072] "[#6X3x0:2](-[#6X4x0:1])-[#7X3x2:3]".

[0073] The above encoding of different levels extends from the dimension of structural fragments, and from the chemical element type to the dimension of the structural type representation of atoms. In actual scenarios, the encoding of different levels can include the expansion of the above two dimensions at the same time. For example, the third level can include both the structural fragment expansion and the structural type representation expansion relative to the second level.

[0074] The above encoding method can systematically and accurately describe the structural information of the target molecule and provide detailed and reliable basic data for subsequent force field parameter matching and molecular simulation. This encoding method can refine the molecular encoding representation layer by layer without increasing the computational complexity, making the force field parameter matching process more efficient and accurate, and significantly improving the accuracy and efficiency of molecular simulation.

[0075] In one implementation, the electronic device 110 performs encoding on the structural information, including: for a given level in the multiple levels, performing encoding on the feature information of each atom indicated by the given level, and determining the coded representation corresponding to each atom. In response to the existence of at least two atoms having the same coded representation, performing supplementation on the feature information of the at least two atoms so that the coded representations corresponding to each atom indicated by the given level are different from each other.

[0076] Taking the determination of the coding representation of a given level among multiple levels as an example, the determination process of the coding representation of any level is described. First, the characteristic information of the atom to be coded indicated by the given level is obtained. The atom to be coded corresponds to the atom indicated by the given level, for example Figure 4 The first structure fragment 301 corresponding to the top level includes a carbon atom, a nitrogen atom and another carbon atom connected to the carbon atom. The characteristic information may include the chemical element type of the atom, bonding conditions, aromaticity, number of intra-ring connections, and the like.

[0077] For each atom indicated at a given level, obtain the characteristic information of each atom. Encode the characteristic information of each atom to obtain the corresponding coded representation of each atom. The coded representation should reflect the chemical characteristics and structural properties of the atom. Check the coded representation of each atom in a given level to determine whether there are at least two atoms with the same coded representation. If there are identical coded representations, further processing is required to ensure the uniqueness of the coded representation.

[0078] For atoms with the same coded representation, feature supplementation can be performed to make the coded representations different from each other. The supplementation can include adding more feature information, such as the stereochemical configuration of the atom, the type and connection mode of the surrounding atoms, etc. The supplemented feature information is encoded and added to the original coded representation to generate an updated and unique coded representation.

[0079] In this way, it can be ensured that at each level, the encoding representations of all atoms are mutually exclusive, thus avoiding mismatches caused by repeated encoding representations. This method improves the accuracy and uniqueness of the encoding, making the matching process of molecular force field parameters more efficient and accurate.

[0080] After determining the coded representations of multiple levels, the electronic device 110 can also traverse the multiple levels to obtain the coded representations corresponding to each level in the multiple levels. In response to the existence of at least two coded representations corresponding to each level in the coded representations corresponding to each level having the same coded representations, and the difference part indicates the structural type representation of the atom, the coded representations corresponding to the at least two levels are associated, the structural type representation is part of the feature information, and the association includes determining the coded representation corresponding to the first level as a root node, and determining the coded representations corresponding to other levels as child nodes corresponding to the root node, and the first level is the top level of the other levels.

[0081] After determining the coded representations of multiple levels, you can traverse down from the top level layer by layer. Encode the atoms indicated by each level to obtain the coded representation of that level. During the traversal process, check the coded representation of each level to determine whether there are at least two levels of coded representations that are partially the same. If there are parts that are the same, they can also be marked and the difference parts can be determined.

[0082] For example, the electronic device 110 finds that the coded representations of the second level and the third level are partially the same during the traversal process, but the difference part indicates the structural type of the atom. In other words, the difference part indicates that the third level has added features indicating bonds, bond angles or dihedral angles. In this case, the second level and the third level can be associated to merge these levels into the same node of the tree structure. Specifically, the coded representation of the second level is determined as the root node, and the coded representation of the third level is determined as the child node corresponding to the root node.

[0083] Through the above process, if some of the same coding representations appear between different levels, and the different coding representations are caused by the structural type representations, then the coding representations of different levels can be associated, thereby constructing a tree structure of coding representations between different levels. Through the tree structure, an association relationship is constructed for the original independent coding representations. Sharing the same part of the coding representation between different levels avoids repeated matching in the subsequent molecular force field parameter matching process, thereby improving the matching efficiency.

[0084] In one implementation, the process of determining the molecular force field parameters may include: obtaining a molecular force field parameter matching table, the molecular force field parameter matching table indicating a matching relationship between the molecular force field parameters and the coded representation, and determining the molecular force field parameters that match the coded representation of the target molecule from the molecular force field parameter matching table as the molecular force field parameters corresponding to the target molecule.

[0085] The molecular force field parameter matching table indicates the matching relationship between the molecular force field parameters and the coding representation. Specifically, the molecular force field parameter matching table records the molecular force field parameters corresponding to each coding representation in the molecular force field parameter matching table, which is convenient for rapid search and matching. Based on the determined coding representation of the target molecule, the molecular force field parameters that match the coding representation can be quickly determined according to the molecular force field parameter matching table.

[0086] In one implementation, the process of determining the molecular force field parameters based on the molecular force field parameter matching table may specifically include: starting from the top level of the multiple levels, sequentially based on the coding representation corresponding to each level, performing force field parameter matching in the force field parameter matching table until a matching stop condition is satisfied, the matching stop condition indicating that there is no molecular force field parameter in the force field parameter matching table that matches the coding representation corresponding to the level where the matching is performed. In the force field parameter matching table, based on the matching result of the previous adjacent level of the level where the matching is performed, the molecular force field parameters corresponding to the target molecule are determined.

[0087] For the coding representations corresponding to the determined multiple levels, you can first start from the coding representation of the top level and perform matching in the force field parameter matching table. If the corresponding molecular force field parameters are found in the force field parameter matching table for the coding representation of the top level, continue to match the coding representation of the next level. Match the coding representation of each level downward in turn until the matching stop condition is met. The matching stop condition indicates that the matching process stops when there are no molecular force field parameters matching the coding representation corresponding to the level where the matching is performed in the force field parameter matching table. For example, if no matching molecular force field parameters are found at a certain level, the matching process is stopped. When the matching stop condition is met, the molecular force field parameters corresponding to the target molecule are determined based on the matching results of the previous adjacent level. In other words, if no matching molecular force field parameters are found at a certain level, the matching results of the previous level of the level are used as the molecular force field parameters of the target molecule.

[0088] By matching layer by layer, the structural features of the target molecule can be captured more precisely, and the matching accuracy of the molecular force field parameters can be improved. In addition, the layer by layer matching avoids the computational complexity brought by the one-time matching of the entire molecular structure, reducing the computational burden. The setting of the matching stop condition can ensure that when a more precise matching result cannot be found, a higher-level matching result is used, ensuring the reliability of the matching result.

[0089] The level-by-level matching process can be performed based on a machine learning model. The following is a brief introduction to the training process of the machine learning model. First, the molecular samples in the model training data set are preprocessed to extract their structural information. Thereafter, encoding is performed on the structural information of each molecular sample to obtain a coded representation of the molecular sample. For each given molecular sample of each molecular sample, the coded representation of the given molecular sample includes coded representations corresponding to multiple levels, a specified level among the multiple levels indicates feature information of at least one atom included in the given molecular sample, and the feature information indicated by the specified level includes feature information additional to the feature information indicated by the previous adjacent level. That is, for each molecular sample, the coded representation of the molecular sample includes coded representations corresponding to multiple levels.

[0090] Take 3 levels as an example. For the top level, the following training process is performed: First, an initial molecular force field parameter is assigned to each encoding representation of the top level. Based on the difference between the initial molecular force field parameter and the annotation result, the parameters in the machine learning model are adjusted until the difference between the molecular force field parameter determined by the machine learning model based on the encoding representation and the annotation result converges.

[0091] For the second level, the following training process is performed: Based on the training results of the top level, the molecular force field parameters corresponding to each coding representation of the second level are determined. Based on the difference between the determined molecular force field parameters and the annotation results, the parameters in the machine learning model are adjusted again until the machine learning model converges with the difference between the molecular force field parameters determined by the coding representation and the annotation results. Similarly, based on the training results of the second level, the molecular force field parameters corresponding to each coding representation of the third level are determined. And the above parameter adjustment process is repeated until the difference between the molecular force field parameters determined by the machine learning model and the annotation results converges.

[0092] Through the above-mentioned step-by-step training method, it is possible to ensure that the parameters of the corresponding machine learning model are optimized during the training process, and when training the early levels, a good initial value is provided for the training of the subsequent levels. This tree-structured training method has higher efficiency and better accuracy than the traditional linear training method, because the parameters in the machine learning model are fully optimized and can better adapt to the diversity of molecular structures.

[0093] In the process of molecular force field parameter matching of the target molecule, starting from the top level in the coding representation of the target molecule, the force field parameters corresponding to the top level can be determined based on the trained machine learning model (the force field parameters corresponding to the top level are determined based on the coding table corresponding to the top level). Based on the determined force field parameters, force field parameter matching is performed in the force field parameter matching table. And so on, until the matching stop condition is met.

[0094] Figure 7 A schematic structural block diagram of a molecular encoding apparatus 700 according to some embodiments of the present disclosure is shown. The apparatus 700 may be implemented in or included in the electronic device 110, for example. Each module / component in the apparatus 700 may be implemented by hardware, software, firmware, or any combination thereof.

[0095] As shown in the figure, the device 700 includes a structural information determination module 701, which is configured to obtain the structural information of the target molecule. The encoding module 702 is configured to perform encoding on the structural information to obtain the encoded representation of the target molecule, the encoded representation includes encoded representations corresponding to multiple levels, a specified level in the multiple levels indicates characteristic information of at least one atom included in the target molecule, and the characteristic information indicated by the specified level includes additional characteristic information based on the characteristic information indicated by the previous adjacent level. The molecular force field parameter determination module 703 is configured to determine the molecular force field parameters corresponding to the target molecule based on the encoded representation of the target molecule.

[0096] In some embodiments of the present disclosure, the encoding module 702 is specifically configured as follows: for a top level of multiple levels, a first structural fragment in the target molecule is obtained, the first structural fragment includes at least one atom. Characteristic information of at least one atom included in the first structural fragment is encoded to obtain a coded representation corresponding to the top level. In addition, for a non-top level of multiple levels, a second structural fragment in the target molecule is obtained, the second structural fragment is determined based on a structural fragment corresponding to a previous adjacent level of the non-top level. Characteristic information of the atoms included in the second structural fragment is encoded to obtain a coded representation corresponding to the non-top level.

[0097] In some embodiments of the present disclosure, the encoding module 702 is specifically configured as follows: for a top level of multiple levels, obtain a chemical element type representation of a given atom in at least one atom indicated by the top level, and the chemical element type representation is part of the feature information. Encode the chemical element type representation of the given atom to obtain a coded representation corresponding to the top level. In addition, for a non-top level of multiple levels, obtain a structural type representation of a given atom in at least one atom indicated by the top level, and the structural type representation is part of the feature information. Encode the structural type representation and the chemical element type representation of the given atom to obtain a coded representation corresponding to the non-top level.

[0098] In some embodiments of the present disclosure, the encoding module 702 is specifically configured to: for a given level in the multiple levels, perform encoding on the feature information of each atom indicated by the given level, and determine the coded representation corresponding to each atom. In response to the existence of at least two atoms having the same coded representation, perform supplementation on the feature information of the at least two atoms so that the coded representations corresponding to each atom indicated by the given level are different from each other.

[0099] In some embodiments of the present disclosure, a coding representation association module is configured to: traverse multiple levels and obtain the coding representation corresponding to each level in the multiple levels. In response to the existence of at least two coding representations corresponding to each level that are partially identical and the difference part indicates the structural type representation of the atom, the coding representations corresponding to the at least two levels are associated, the structural type representation is part of the feature information, and the association includes determining the coding representation corresponding to the first level as a root node, and determining the coding representations corresponding to other levels as child nodes corresponding to the root node, and the first level is the top level of the other levels.

[0100] In some embodiments of the present disclosure, the molecular force field parameter determination module 703 may include a matching table acquisition submodule, which is configured to acquire a molecular force field parameter matching table, the molecular force field parameter matching table indicating the matching relationship between the molecular force field parameters and the coded representation. A matching execution submodule, which is configured to determine the molecular force field parameters that match the coded representation of the target molecule from the molecular force field parameter matching table as the molecular force field parameters corresponding to the target molecule.

[0101] In some embodiments of the present disclosure, the matching execution submodule is specifically configured to perform force field parameter matching in the force field parameter matching table based on the coding representation corresponding to each level in turn, starting from the top level of the multiple levels, until a matching stop condition is met, and the matching stop condition indicates that there is no molecular force field parameter in the force field parameter matching table that matches the coding representation corresponding to the level where the matching is performed. In the force field parameter matching table, the molecular force field parameters corresponding to the target molecule are determined based on the matching result of the previous adjacent level of the level where the matching is performed.

[0102] Figure 8 8 is a block diagram of an electronic device 800 in which one or more embodiments of the present disclosure may be implemented. It should be understood that Figure 8 The electronic device 800 shown is merely exemplary and should not constitute any limitation on the functionality and scope of the embodiments described herein. Figure 8 The electronic device 800 shown may include or be implemented as Figure 1 Electronic device 110 or Figure 7 Device 700.

[0103] like Figure 8As shown, the electronic device 800 is in the form of a general electronic device. The components of the electronic device 800 may include, but are not limited to, one or more processors or processing units 810, a memory 820, a storage device 830, one or more communication units 840, one or more input devices 850, and one or more output devices 860. The processing unit 810 may be an actual or virtual processor and is capable of performing various processes according to a program stored in the memory 820. In a multi-processor system, multiple processing units execute computer executable instructions in parallel to improve the parallel processing capability of the electronic device 800.

[0104] The electronic device 800 typically includes a plurality of computer storage media. Such media can be any accessible media that can be obtained by the electronic device 800, including but not limited to volatile and non-volatile media, removable and non-removable media. The memory 820 can be a volatile memory (e.g., a register, a cache, a random access memory (RAM)), a non-volatile memory (e.g., a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), flash memory), or some combination thereof. The storage device 830 can be a removable or non-removable medium, and can include a machine-readable medium, such as a flash drive, a disk, or any other medium, which can be used to store information and / or data and can be accessed within the electronic device 800.

[0105] The electronic device 800 may further include additional removable / non-removable, volatile / non-volatile storage media. Figure 8 As shown in , a disk drive for reading or writing from a removable, non-volatile disk (e.g., a "floppy disk") and an optical drive for reading or writing from a removable, non-volatile optical disk may be provided. In these cases, each drive may be connected to the bus (not shown) by one or more data media interfaces. The memory 820 may include a computer program product 825 having one or more program modules that are configured to perform various methods or actions of various embodiments of the present disclosure.

[0106] The communication unit 840 implements communication with other electronic devices through a communication medium. Additionally, the functions of the components of the electronic device 800 can be implemented with a single computing cluster or multiple computing machines that can communicate through a communication connection. Therefore, the electronic device 800 can operate in a networked environment using a logical connection with one or more other servers, a network personal computer (PC), or another network node.

[0107] The input device 850 may be one or more input devices, such as a mouse, a keyboard, a tracking ball, etc. The output device 860 may be one or more output devices, such as a display, a speaker, a printer, etc. The electronic device 800 may also communicate with one or more external devices (not shown) through the communication unit 840 as needed, such as a storage device, a display device, etc., communicate with one or more devices that allow a user to interact with the electronic device 800, or communicate with any device that allows the electronic device 800 to communicate with one or more other electronic devices (e.g., a network card, a modem, etc.). Such communication may be performed via an input / output (I / O) interface (not shown).

[0108] According to an exemplary implementation of the present disclosure, a computer-readable storage medium is provided, on which computer-executable instructions are stored, wherein the computer-executable instructions are executed by a processor to implement the method described above. According to an exemplary implementation of the present disclosure, a computer program product is also provided, which is tangibly stored on a non-transitory computer-readable medium and includes computer-executable instructions, and the computer-executable instructions are executed by a processor to implement the method described above.

[0109] According to an exemplary implementation of the present disclosure, a computer program product or a computer program is provided, the computer program product or the computer program including computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes Figure 2 The methods provided in the various optional methods are therefore not described in detail here.

[0110] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices, equipment, and computer program products implemented according to the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0111] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processing unit of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0112] Computer-readable program instructions can be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, so that the instructions executed on the computer, other programmable data processing apparatus, or other device implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0113] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple implementations of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of a module, program segment or instruction includes one or more executable instructions for realizing the logical function of the specification. In some implementations as replacements, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be realized by a special hardware-based system that performs the function or action of the specification, or can be realized by a combination of special hardware and computer instructions.

[0114] The above descriptions of various implementations of the present disclosure are exemplary, non-exhaustive, and not limited to the disclosed implementations. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described implementations. The selection of terms used herein is intended to best explain the principles of the implementations, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the various implementations disclosed herein.

Claims

1. A method for determining molecular force field parameters, comprising: Obtain structural information of target molecules; Performing encoding on the structural information to obtain an encoded representation of the target molecule, the encoded representation comprising encoded representations corresponding to a plurality of levels, a specified level among the plurality of levels indicating feature information of at least one atom included in the target molecule, and the feature information indicated by the specified level including feature information additional to the feature information indicated by a previous adjacent level; as well as Based on the encoded representation of the target molecule, molecular force field parameters corresponding to the target molecule are determined.

2. The method of claim 1, wherein encoding the structure information comprises: For a top level of the plurality of levels, Acquire a first structural fragment in the target molecule, wherein the first structural fragment includes at least one atom; Performing encoding on the feature information of the at least one atom included in the first structure fragment to obtain an encoding representation corresponding to the top level; For non-top levels of the plurality of levels, Obtaining a second structural fragment in the target molecule, wherein the second structural fragment is determined based on a structural fragment corresponding to a previous adjacent level of the non-top level; and Encoding is performed on the characteristic information of atoms included in the second structure fragment to obtain an encoded representation corresponding to the non-top level.

3. The method of claim 1 , wherein encoding the structure information comprises: For a top level of the plurality of levels, Obtaining a chemical element type representation of a given atom in at least one atom indicated by the top level, wherein the chemical element type representation is a part of the feature information; Performing encoding on the chemical element type representation of the given atom to obtain an encoded representation corresponding to the top level; For non-top levels of the plurality of levels, Obtaining a structure type representation of a given atom in at least one atom indicated by the top level, wherein the structure type representation is a part of the feature information; as well as Encoding is performed on the structure type representation of the given atom and the chemical element type representation to obtain an encoded representation corresponding to the non-top level.

4. The method of claim 1, wherein encoding the structure information comprises: For a given level in the plurality of levels, Perform encoding on the feature information of each atom indicated by the given level to determine the encoding representation corresponding to each atom; as well as In response to the existence of at least two atoms whose corresponding coded representations are the same, feature information of the at least two atoms is supplemented so that the coded representations corresponding to each atom indicated by the given level are different from each other.

5. The method according to claim 1, further comprising: Traversing the multiple levels, obtaining a coding representation corresponding to each level in the multiple levels; as well as In response to the existence of at least two coding representations corresponding to each level that are partially identical and the difference portion indicates a structural type representation of an atom, the coding representations corresponding to the at least two levels are associated, the structural type representation is part of the feature information, and the association includes determining the coding representation corresponding to the first level as a root node, and determining the coding representations corresponding to other levels as child nodes corresponding to the root node, the first level being the top level of the other levels.

6. The method according to claim 1, wherein determining the molecular force field parameters corresponding to the target molecule comprises: Obtaining a molecular force field parameter matching table, wherein the molecular force field parameter matching table indicates a matching relationship between molecular force field parameters and coded representations; as well as The molecular force field parameters that match the coded representation of the target molecule are determined from the molecular force field parameter matching table to serve as the molecular force field parameters corresponding to the target molecule.

7. The method according to claim 6, wherein determining the molecular force field parameters that match the encoded representation of the target molecule from the molecular force field parameter matching table comprises: Starting from the top level of the multiple levels, sequentially based on the encoding representation corresponding to each level, performing force field parameter matching in the force field parameter matching table until a matching stop condition is satisfied, wherein the matching stop condition indicates that there is no molecular force field parameter in the force field parameter matching table that matches the encoding representation corresponding to the level where matching is performed; as well as In the force field parameter matching table, the molecular force field parameters corresponding to the target molecule are determined based on the matching result of the previous adjacent level of the level where the matching is performed.

8. A device for determining sub-force field parameters, comprising: a structural information determination module, configured to obtain structural information of a target molecule; an encoding module, configured to perform encoding on the structural information to obtain an encoded representation of the target molecule, wherein the encoded representation includes encoded representations corresponding to a plurality of levels, a specified level among the plurality of levels indicates characteristic information of at least one atom included in the target molecule, and the characteristic information indicated by the specified level includes additional characteristic information based on the characteristic information indicated by a previous adjacent level; as well as The molecular force field parameter determination module is configured to determine the molecular force field parameters corresponding to the target molecule based on the encoded representation of the target molecule.

9. An electronic device, comprising: at least one processing unit; as well as At least one memory, the at least one memory is coupled to the at least one processing unit and stores instructions for execution by the at least one processing unit, the instructions, when executed by the at least one processing unit, causing the electronic device to perform the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, the computer program being executable by a processor to implement the method according to any one of claims 1 to 7.

11. A computer program product comprising computer executable instructions, wherein the computer executable instructions, when executed by a processor, implement the method according to any one of claims 1 to 7.

Citation Information

Patent Citations

  • Automatic generating method for force field parameter of molecular mechanics

    CN101131707A

  • Molecular force field fitting method and device

    CN117809757A

  • Molecular dynamics force field construction method based on atomic fingerprints

    CN117976068A

  • Molecular property prediction method, electronic equipment and computer readable storage medium

    CN118039000A

Cited By

  • Model training method, diffusion model-based molecule generation method and data processing device

    CN120804707A