Covalent molecule binding conformation prediction method and device

Abstract

The invention discloses a covalent molecule binding conformation prediction method and device. The method comprises the following steps: acquiring a three-dimensional structure of a protein, determining the position of a nucleophilic residue which is quasi-covalently bound, and intercepting a protein pocket; the method comprises the following steps: acquiring a ligand molecular structure, determining a covalent reaction type of the ligand molecular structure and protein, identifying a potential reaction warhead, carrying out structure transformation on the ligand molecule before and after the reaction, labeling a reaction atom, adding a pseudo atom, generating an initial three-dimensional conformation of a new ligand molecule, and carrying out coordinate superposition on the pseudo atom of the ligand and the reaction atom of a reaction residue; initial representations of the two molecules are obtained, the distance matrix of the two molecules is predicted, finally, the conformation is updated through the geometric optimization process limited by covalent bonds, and the final covalent binding conformation is obtained. The embodiment of the invention provides a novel covalent molecule binding conformation prediction scheme, and the prediction precision of covalent molecule binding conformation can be remarkably improved.

Description

Technical Field

[0001] The invention belongs to the technical field of drug design, and relates to a method and device for predicting covalent molecule binding conformation. Background Art

[0002] Covalent drug design is currently a hot area in new drug research and development, but how to combine computational methods to discover and optimize covalent lead molecules remains a major difficulty in the field.

[0003] Molecular docking is the most commonly used structure-based computer-aided drug design method and has been widely used in the discovery and optimization of lead molecules. Traditional docking methods usually use conformational search algorithms to sample the binding conformations of ligand small molecules in protein pockets, and evaluate the binding strength of the two through a pre-defined scoring function. However, traditional covalent molecular docking algorithms such as CovDock, Dockcovalent, and CovalentDock have general predictive capabilities for binding conformations and cannot accurately screen out near-native conformations of ligands. On the one hand, this is because traditional heuristic search algorithms are difficult to fully collect near-native conformations; on the other hand, traditional empirical scoring functions are difficult to effectively capture the interaction between ligands and target proteins, and thus cannot accurately rank conformations.

[0004] In recent years, artificial intelligence (AI) technology has been widely introduced into the development of docking methods, and emerging methods such as DiffDock, KarmaDock, and CarsiDock have emerged. These methods can predict the binding conformation of ligands and proteins end-to-end based on the two-dimensional structure of target proteins and ligands. Compared with traditional docking methods, they often show significantly better prediction accuracy and computational efficiency, but their applicability to covalent ligands is limited. Therefore, the development of dedicated covalent docking algorithms based on new AI technologies is expected to significantly improve the prediction accuracy of docking programs for the binding conformation of covalent molecules, thereby improving the research and development efficiency of covalent drugs. Summary of the invention

[0005] The technical problem to be solved by the embodiments of the present application is to provide a method and device for predicting the binding conformation of a covalent molecule, so as to achieve accurate prediction of the binding conformation of a covalent molecule.

[0006] In a first aspect, an embodiment of the present application provides a method for predicting covalent molecular binding conformations, the method comprising:

[0007] Obtain the three-dimensional structure corresponding to the protein molecule;

[0008] Pre-process the protein three-dimensional structure;

[0009] Obtain the structure of the ligand molecule;

[0010] Preprocessing the ligand structure;

[0011] Characterize the pre-processed protein pocket and ligand three-dimensional conformations to obtain their respective initial atomic representations and atom pair representations;

[0012] Convert the initial representation of proteins and ligands into corresponding predicted distance matrices;

[0013] The three-dimensional conformation of the ligand molecule is rotated and the dihedral angle is twisted to update the conformation of the ligand molecule and output the final covalent binding conformation.

[0014] In a second aspect, an embodiment of the present application provides a covalent molecule binding conformation prediction device, the device comprising:

[0015] Protein structure acquisition module, used to obtain the three-dimensional structure corresponding to the protein molecule;

[0016] Protein preprocessing module, used to preprocess the three-dimensional structure of proteins;

[0017] Ligand structure acquisition module, used to obtain the structure of ligand molecules;

[0018] Ligand preprocessing module, used to preprocess the ligand structure;

[0019] The molecular characterization module is used to characterize the pre-processed protein pockets and ligand three-dimensional conformations and obtain their respective initial atomic representations and atom pair representations;

[0020] The distance matrix prediction module is used to convert the initial representation of proteins and ligands into the corresponding predicted distance matrices;

[0021] Covalent bond-constrained geometry optimization module: used to rotate the three-dimensional conformation of the ligand molecule and perform dihedral angle torsion operations, update the conformation of the ligand molecule, and output the final covalent binding conformation.

[0022] In a third aspect, an embodiment of the present application provides an electronic device, including:

[0023] A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes any of the above-described covalent molecular binding conformation prediction methods.

[0024] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium. When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute any of the above-mentioned covalent molecular binding conformation prediction methods.

[0025] Compared with the prior art, the embodiments of the present application have the following advantages:

[0026] In the embodiment of the present application, by obtaining the three-dimensional structure of the protein, the position of the nucleophilic residues to be covalently bound is determined and the protein pocket is intercepted; the structure of the ligand molecule is obtained, the type of covalent reaction to be performed with the protein is determined, the potential reaction warhead is identified, the structure of the ligand molecule is transformed before and after the reaction, the reaction atoms are marked and pseudo-atoms are added, the initial three-dimensional conformation of the new ligand molecule is generated and the coordinates of the pseudo-atoms of the ligand and the reaction atoms of the reaction residues are superimposed; the initial representation of the two molecules is obtained, and their distance matrix is ​​predicted, and finally the conformation is updated through a covalent bond-constrained geometric optimization process to obtain the final covalent binding conformation. The embodiment of the present application proposes a new covalent molecule binding conformation prediction scheme, which can significantly improve the prediction accuracy of covalent molecule binding conformations.

[0027] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A flowchart of the steps of a method for predicting covalent molecular binding conformation provided in an embodiment of the present application;

[0029] Figure 2 A flowchart of the steps of a ligand molecule pretreatment method provided in an embodiment of the present application;

[0030] Figure 3 A schematic diagram of the structural transformation of a ligand before and after reaction provided in an embodiment of the present application;

[0031] Figure 4 A flow chart of steps of a covalent bond-constrained geometry optimization process provided in an embodiment of the present application;

[0032] Figure 5 A schematic diagram of the structure of a covalent molecular binding conformation prediction device provided in an embodiment of the present application;

[0033] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0034] The present invention is further described below in conjunction with the accompanying drawings.

[0035] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0036] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings.

[0037] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or terminal including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the existence of other identical elements in the process, method, article or terminal including the elements.

[0038] Reference Figure 1 , showing a flow chart of the steps of a covalent molecular binding conformation prediction method provided in an embodiment of the present application, such as Figure 1 As shown, the covalent molecular binding conformation prediction method may include: step 101, step 102, step 103 and step 104.

[0039] Step 101, obtaining a protein molecule and pre-processing it to obtain a protein pocket marked with a nucleophilic residue.

[0040] In the embodiments of the present application, the three-dimensional structure of the protein molecule can be obtained from the resolved crystal structure in the PDB database, or through protein structure prediction methods such as homology modeling.

[0041] For the obtained protein molecules, it is necessary to identify the nucleophilic residues to which covalent bonding is to occur. These residues are generally reactive amino acids such as Cys, Ser, His, and Lys. In practical applications, the nucleophilic reactivity of these residues must be identified. Subsequently, open source python libraries such as Prody and MDAnalysis are used to read and manipulate protein molecules, annotate the nucleophilic residues to be reacted, and select a certain range around these residues (such as ) amino acids form protein pockets.

[0042] After the three-dimensional structure of the protein molecule is obtained and pre-processed to obtain the protein pockets annotated with nucleophilic residues, step 102 is performed.

[0043] Step 102: Obtain the ligand molecule and perform pretreatment to obtain an initial three-dimensional conformation that is covalently linked to the protein nucleophilic residue.

[0044] The ligand molecule can be in the form of a one-dimensional SMILES string, or in the form of a two-dimensional or three-dimensional SDF or MOL2 molecular file format, and the reading and operation of the molecule can be achieved through open source Python libraries such as RDKit and OpenBabel.

[0045] After obtaining the ligand molecular structure and the protein pockets annotated with nucleophilic residues, the ligand molecule can be pre-processed to meet the requirements of covalent docking. Figure 2 This is described in detail as follows.

[0046] Reference Figure 2 , shows a flow chart of the steps of a ligand molecule pretreatment method provided in an embodiment of the present application. Figure 2 As shown, the ligand molecule pretreatment process may include: step 201, step 202, step 203 and step 204.

[0047] Step 201: Identify the ligand reaction warhead, perform structural transformation on the ligand molecule before and after the reaction, and mark the reaction atoms.

[0048] First, the potential reaction warheads of the ligand molecule need to be identified according to the preset covalent reaction type. When the ligand does not contain the preset reaction warhead, the process will be terminated directly; if the ligand contains the preset reaction warhead, the subsequent processes will be carried out one by one according to the number of potential reaction warheads. Then, according to the preset reaction rules, the structure of the ligand molecule before and after the reaction is transformed. Some commonly used covalent reaction rules can be as follows: Figure 3 shown.

[0049] In a specific implementation, the reaction types supported by the embodiments of the present application are not limited to Figure 3 The Michael addition reaction, double bond nucleophilic addition reaction, triple bond nucleophilic addition reaction, boronic acid addition reaction, electrophilic substitution reaction, epoxide ring opening reaction, β-lactam addition reaction, γ-lactone addition reaction, disulfide bond formation, and imine condensation reaction are shown, and the reaction rules supported by each reaction type are not limited to Figure 3 The part shown.

[0050] After obtaining the structure of the ligand molecule after the reaction, the reactive atoms in the ligand molecule that are intended to form covalent bonds with the nucleophilic residues of the protein are marked, and then step 202 can be executed.

[0051] Step 202: Add pseudo atoms for ligand reaction atoms.

[0052] In this embodiment, the atomic properties of the pseudo-atom to be added as the ligand reaction atom are determined according to the amino acid type of the nucleophilic residue to which the ligand molecule is to be covalently bound. For example, if the amino acid type of the nucleophilic residue is Cys, a sulfur atom is attached to the ligand reaction atom; if the amino acid type is Ser, an oxygen atom is attached; if it is His or Lys, a nitrogen atom is attached.

[0053] After the pseudo-atom is connected, step 203 may be performed.

[0054] Step 203: Generate an initial three-dimensional conformation of the ligand molecule.

[0055] In this embodiment, a small molecule conformation generation tool in an open source program such as RDKit or OpenBabel is used to generate an initial three-dimensional conformation of a ligand molecule with a pseudo atom attached. The generated initial conformation may be one or more. When multiple initial conformations are generated, the subsequent processes are performed one by one.

[0056] After the three-dimensional conformation of the ligand is obtained, step 204 may be performed.

[0057] Step 204: Overlay the coordinates of the ligand pseudo-atoms with the protein reactive atoms.

[0058] After obtaining the initial three-dimensional conformation of the ligand molecule, a coordinate translation operation will be performed on the ligand molecule so that the coordinates of its pseudo-atom are completely aligned with the coordinates of the reactive atoms of the protein nucleophilic residues, thereby forming a covalent bond between the ligand and the protein in three-dimensional space.

[0059] After obtaining the protein pocket annotated with the nucleophilic residue and the three-dimensional initial conformation of the ligand covalently linked to the protein nucleophilic residue, step 103 may be performed.

[0060] Step 103: Obtain the distance matrix between the ligand and the protein.

[0061] In this embodiment, the preprocessed protein pocket and ligand three-dimensional conformation will be further converted into corresponding atoms and atom pairs, and then input into a deep learning model (such as CarsiDock's distance matrix prediction model) based on conventional molecular training to obtain the atom pair distance matrix between the protein and the ligand and the atom distance matrix inside the ligand. In a specific implementation, the distance matrix prediction model can also be fine-tuned based on a dedicated covalent complex, or trained separately based on other deep learning frameworks.

[0062] After the distance matrix is ​​obtained, step 104 may be performed.

[0063] Step 104: Perform a covalent bond-restrained geometry optimization process to obtain the final binding conformation.

[0064] After the distance matrix between the protein and the ligand is predicted, the covalent bond-constrained geometry optimization process can be performed to obtain the final binding conformation. Figure 4 This is described in detail as follows.

[0065] like Figure 4 As shown, for the input protein pocket, the initial ligand three-dimensional conformation, and the predicted distance matrix of the two, the pseudo-atoms of the ligand molecule are used as the center to perform rotation and dihedral angle torsion operations to obtain an updated ligand three-dimensional conformation; based on the updated ligand three-dimensional conformation and protein pocket, the corresponding updated distance matrix can be obtained; the distance difference loss between the updated distance matrix and the predicted distance matrix is ​​calculated. If the loss is greater than the preset threshold, the rotation and dihedral angle torsion parameters of the current three-dimensional conformation of the ligand will continue to be updated; if the loss meets the predetermined requirements, the conformation will be output as the final predicted covalent binding conformation. In the specific implementation, the pseudo-atoms of the ligand are not taken into account in the calculation of the distance matrix and the corresponding loss, and are only used to assist in the update of rotation and dihedral parameters.

[0066] In a specific implementation of the present application, after the above step 104, a deep learning-based scoring function (such as RTMScore and GenScore) or a traditional scoring function (such as AutoDock Vina, Smina) can also be used to evaluate the binding strength of the output conformation.

[0067] The advantages of the covalent molecular docking method provided in the embodiments of the present application are described in combination with two aspects:

[0068] 1. Docking accuracy of molecular docking methods

[0069] The present application example compares the docking accuracy of this method and traditional covalent docking programs on two public covalent complex data sets. As shown in Table 1, this method achieves significantly better performance than traditional methods on both test sets.

[0070] Table 1

[0071]

[0072] The quality of a docked conformation can usually be evaluated by calculating the root mean square deviation (RMSD) between the conformation and the native conformation in the resolved crystal structure. The smaller the RMSD value, the higher the quality of the docked conformation. Therefore, the mean or median RMSD of a group of samples can be calculated to evaluate the prediction accuracy of the docking method. The docking success rate indicates the number of successful predictions under different RMSD thresholds (such as RMSD < The percentage of samples with ) in the total number of samples. The higher the success rate, the better the docking accuracy. The other methods in Table 1 are all commonly used traditional covalent docking programs.

[0073] 2. Virtual Screening Performance

[0074] Docking programs have been widely used in virtual screening, so screening ability has become an important indicator for evaluating the performance of a docking program. The screening ability depends not only on the quality of the docking conformation generated by the docking program, but also on the scoring function's evaluation of the binding affinity based on the docking conformation. The embodiment of the present application adopts the proposed covalent docking method to generate a covalent binding conformation, and uses RTMScore to re-score the conformation, and evaluates its screening ability on the CovDock test set. The test set contains a total of 10 PDB entries, covering 3 kinase targets, each of which contains 10 active covalent molecules and 100 bait molecules containing the same reaction warhead. As shown in Table 2, this method showed a screening ability that was significantly better than HCovDock and CovDock on the covalent screening test set.

[0075] Table 2

[0076]

[0077] The screening ability in Table 2 is mainly evaluated by the area under the receiver's operating curve (AUROC). The larger the AUROC, the stronger the virtual screening ability. HCovDock and CovDock are two traditional covalent docking methods. The relevant data in Table 2 are from their respective original literature reports.

[0078] The covalent molecular binding conformation prediction method provided in the embodiment of the present application obtains the three-dimensional structure of the protein, pre-processes it to obtain the protein pocket marked with nucleophilic residues; obtains the molecular structure of the ligand, identifies the potential reaction warhead and performs structural transformation before and after the reaction, and finally obtains the initial three-dimensional conformation of the ligand covalently connected to the nucleophilic residue; obtains the initial atomic and atomic pair representations of the two molecules, predicts their distance matrix, and finally updates the conformation through a covalent bond-constrained geometric optimization process to obtain the final covalent binding conformation. The embodiment of the present application proposes a new covalent molecular binding conformation prediction scheme, which can significantly improve the prediction accuracy of covalent molecular binding conformations.

[0079] An embodiment of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the above-mentioned covalent molecular binding conformation prediction method when executed by the processor.

[0080] Reference Figure 5 , shows a schematic diagram of the structure of a covalent molecular binding conformation prediction device provided in an embodiment of the present application. Figure 5 As shown, the covalent molecule binding conformation prediction device 500 may include the following modules:

[0081] The protein structure acquisition module 510 is used to acquire the three-dimensional structure corresponding to the protein molecule;

[0082] The protein preprocessing module 520 is used to preprocess the three-dimensional structure of the protein;

[0083] The ligand structure acquisition module 530 is used to acquire the structure of the ligand molecule;

[0084] A ligand preprocessing module 540 is used to preprocess the ligand structure;

[0085] A molecular characterization module 550 is used to characterize the preprocessed protein pocket and ligand three-dimensional conformation and obtain their respective initial atomic representations and atomic pair representations;

[0086] a distance matrix prediction module 560 for converting the initial representations of proteins and ligands into corresponding predicted distance matrices;

[0087] Covalent bond-constrained geometry optimization module 570: used to rotate the three-dimensional conformation of the ligand molecule and perform dihedral angle torsion operations, update the conformation of the ligand molecule, and output the final covalent binding conformation.

[0088] Optionally, the device further comprises a rescoring module for evaluating the binding affinity of the output binding conformation.

[0089] The protein pretreatment module comprises:

[0090] Nucleophilic residue determination unit, used to determine the position of the nucleophilic residue to be covalently bound and mark it;

[0091] The pocket interception unit is used to intercept residues within a certain distance range of the reactive residues as protein pockets.

[0092] The ligand pretreatment module comprises:

[0093] The ligand reaction warhead recognition unit is used to identify potential reaction warheads in ligand molecules, complete the conversion of structures before and after the reaction, and mark the reactive atoms that are to react;

[0094] The pseudo-atom adding unit is used to add a pseudo-atom with the same atomic properties as the reactive atom of the nucleophilic residue to the reactive atom of the ligand;

[0095] The ligand initial conformation generation unit is used to generate the initial three-dimensional conformation of the new ligand molecule;

[0096] The covalent linking unit is used to coordinately superimpose the pseudo-atoms of the ligand molecule with the reactive atoms of the protein nucleophilic residues to achieve covalent linking of the initial ligand conformation with the protein.

[0097] The covalent bond constrained geometry optimization module includes:

[0098] An updated conformation acquisition unit is used to acquire updated ligand conformations and calculate distance matrices;

[0099] A distance difference loss calculation unit, used for calculating the distance difference loss based on the updated distance matrix of the ligand conformation and the predicted distance matrix;

[0100] An update parameter calculation unit, used for calculating the rotation and dihedral angle torsion parameters to be updated in the next step based on the distance difference loss;

[0101] A conformation updating unit, further updating the conformation three-dimensional coordinates based on the parameters;

[0102] The final conformation output unit is used to iteratively perform the update process so that the updated distance loss is as close as possible to the distance loss predicted by the model, and output the final covalent binding conformation.

[0103] The covalent molecular binding conformation prediction device provided in the embodiment of the present application obtains the three-dimensional structure of the protein, pre-processes it to obtain the protein pocket marked with nucleophilic residues; obtains the molecular structure of the ligand, identifies the potential reaction warhead and performs structural transformation before and after the reaction, and finally obtains the initial three-dimensional conformation of the ligand covalently connected to the nucleophilic residue; obtains the initial representation of the two molecules, predicts their distance matrix, and finally updates the conformation through the geometric optimization process of covalent bond constraints to obtain the final covalent binding conformation. The embodiment of the present application proposes a new covalent molecular binding conformation prediction device, which can significantly improve the prediction accuracy of covalent molecular binding conformations.

[0104] An embodiment of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and operable on the memory, wherein the computer program implements the above-mentioned method for predicting the covalent molecular binding conformation when executed by the processor.

[0105] Figure 6 FIG. 6 is a schematic diagram showing the structure of an electronic device 600 according to an embodiment of the present invention. Figure 6As shown, the electronic device 600 includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) or computer program instructions loaded from a storage unit 608 into a random access memory (RAM) 603. In RAM 603, various programs and data required for the operation of the electronic device 600 can also be stored. CPU 601, ROM 602 and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0106] Multiple components in the electronic device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, a microphone, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a disk, an optical disk, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the electronic device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0107] The various processes and processing described above may be performed by the processing unit 601. For example, the method of any of the above embodiments may be implemented as a computer software program, which is tangibly contained in a computer-readable medium, such as the storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the CPU 601, one or more actions of the method described above may be performed.

[0108] An embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned method for predicting the covalent molecular binding conformation.

[0109] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0110] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, devices or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, terminals (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing terminal to generate a machine, which is executed by the processor of the computer or other programmable data processing terminal to generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0112] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing terminal to operate in a predetermined manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, the instruction device being implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0113] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal so that a series of operating steps are executed on the computer or other programmable terminal to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable terminal. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0114] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present application.

[0115] The above is a detailed introduction to a covalent molecule binding conformation prediction method, a covalent molecule binding conformation prediction device, an electronic device and a computer-readable storage medium provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technicians in this field, according to the ideas of the present application, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A method for predicting covalent molecular binding conformation, characterized in that: The method includes: Obtain the three-dimensional structure corresponding to the protein molecule; Pre-process the protein three-dimensional structure; Obtain the structure of the ligand molecule; Preprocessing the ligand structure; Characterize the pre-processed protein pocket and ligand three-dimensional conformations to obtain their respective initial atomic representations and atom pair representations; Convert the initial representation of proteins and ligands into corresponding predicted distance matrices; The three-dimensional conformation of the ligand molecule is rotated and the dihedral angle is twisted to update the conformation of the ligand molecule and output the final covalent binding conformation.

2. A method for predicting covalent molecular binding conformation according to claim 1, characterized in that: The pre-processing of the three-dimensional structure of the protein includes: determining the position of the nucleophilic residues to be covalently bound and marking them, and intercepting the residues within the set distance range of the reactive residues as protein pockets.

3. A method for predicting covalent molecular binding conformation according to claim 1, characterized in that: The pre-processing of the ligand structure includes: determining the type of covalent reaction between the ligand and the protein, identifying potential reaction warheads in the ligand molecule, performing structural transformation before and after the reaction, and marking reactive atoms that are to react; Add a pseudo-atom to the reactive atom of the ligand with the same atomic properties as the reactive atom of the nucleophilic residue; Generate an initial three-dimensional conformation of the new ligand molecule; The pseudo-atoms of the ligand molecule are aligned with the reactive atoms of the protein nucleophilic residues to achieve covalent connection between the initial ligand conformation and the protein.

4. A method for predicting covalent molecular binding conformation according to claim 1, characterized in that: The initial representation of the protein and ligand is converted into a corresponding predicted distance matrix, specifically: The initial representation is input into the distance matrix prediction model to obtain the predicted ligand-protein distance matrix and the ligand internal distance matrix.

5. A method for predicting covalent molecular binding conformation according to claim 4, characterized in that: The three-dimensional conformation of the ligand molecule is rotated and the dihedral angle is twisted to update the conformation of the ligand molecule and output the final covalent binding conformation. Specifically: Centering on the pseudo-atom, the initial three-dimensional conformation of the ligand molecule is rotated and the dihedral angle is twisted to update the conformation of the ligand molecule; Based on the updated distance matrix of the ligand conformation and the predicted distance matrix, the distance difference loss is calculated; Calculating updated rotation and dihedral torsion parameters based on the distance difference loss; Based on the updated parameters, the three-dimensional conformation of the ligand molecule is updated; The updating process is iteratively performed to make the updated distance loss as close as possible to the distance loss predicted by the model, and the final binding conformation is output.

6. A covalent molecular binding conformation prediction device, characterized in that: include: Protein structure acquisition module, used to obtain the three-dimensional structure corresponding to the protein molecule; Protein preprocessing module, used to preprocess the three-dimensional structure of proteins; Ligand structure acquisition module, used to obtain the structure of ligand molecules; Ligand preprocessing module, used to preprocess the ligand structure; The molecular characterization module is used to characterize the pre-processed protein pockets and ligand three-dimensional conformations and obtain their respective initial atomic representations and atom pair representations; The distance matrix prediction module is used to convert the initial representation of proteins and ligands into the corresponding predicted distance matrices; Covalent bond-constrained geometry optimization module: used to rotate the three-dimensional conformation of the ligand molecule and perform dihedral angle torsion operations, update the conformation of the ligand molecule, and output the final covalent binding conformation.

7. A covalent molecular binding conformation prediction device according to claim 6, characterized in that: The device also includes a rescoring module for evaluating the binding affinity of the output binding conformation.

8. An electronic device comprising: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the covalent molecular binding conformation prediction method according to any one of claims 1 to 5.

9. A computer-readable storage medium, when the instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to execute the covalent molecular binding conformation prediction method according to any one of claims 1 to 5.

Images