Protein data processing method and device, electronic equipment, storage medium and program product

Abstract

The invention provides a protein data processing method and device, electronic equipment, a storage medium and a program product. The method comprises the following steps: filling first grid data of a first protein and a second protein from a plurality of dimensions according to different postures of the first protein and the second protein to obtain second grid data of each posture corresponding to the plurality of dimensions; based on second grid data of multiple dimensions corresponding to each attitude and each translation conformation corresponding to the attitude, determining a first docking index corresponding to each translation conformation, the translation conformation being used for representing a translation operation corresponding to the first protein or the second protein; based on the first docking index corresponding to each translation conformation, determining a target docking translation conformation; and performing docking processing on the first protein and the second protein based on the target docking translation conformation to obtain a protein docking result. According to the invention, the accuracy of protein docking can be improved.

Description

Technical Field

[0001] The present application relates to computer technology, and in particular to a protein data processing method, device, electronic device, storage medium and program product. Background Art

[0002] Protein docking is a key step in computer-aided drug design (CADD), which simulates the binding process between proteins and other molecules (such as small molecule drugs, proteins, or other biomacromolecules). This process is important for understanding how biomolecules interact and for designing new drugs.

[0003] Common methods for protein docking include molecular dynamics simulation, free energy calculation, and structural alignment. Although these methods can yield corresponding protein docking results, the prediction methods for protein docking in related technologies do not cover the various complex interactions between antibody proteins and antigen proteins, affecting the accuracy of the prediction results. Summary of the Invention

[0004] The embodiments of the present application provide a protein data processing method, apparatus, electronic device, storage medium, and program product, which can improve the accuracy of protein docking.

[0005] The technical solution of the embodiment of the present application is implemented as follows:

[0006] This embodiment of the present application provides a method for processing protein data, the method comprising:

[0007] Filling the first grid data of the first protein and the second protein from multiple dimensions according to different postures of the first protein and the second protein to obtain second grid data of the multiple dimensions corresponding to each posture;

[0008] Determining a first docking index corresponding to each translational conformation based on the second grid data of the multiple dimensions corresponding to each posture and each translational conformation corresponding to the posture, wherein the translational conformation is used to characterize the translational operation corresponding to the first protein or the second protein;

[0009] determining a target docking translational conformation based on a first docking metric corresponding to each of the translational conformations;

[0010] A docking process is performed on the first protein and the second protein based on the target docking translation conformation to obtain a protein docking result.

[0011] The present invention provides a protein data processing device, comprising:

[0012] a data filling module, configured to fill the first grid data of the first protein and the second protein from multiple dimensions according to different postures of the first protein and the second protein, to obtain second grid data of the multiple dimensions corresponding to each posture;

[0013] The protein docking module is further configured to determine a first docking index corresponding to each translational conformation based on the second grid data of the multiple dimensions corresponding to each posture and each translational conformation corresponding to the posture, wherein the translational conformation is used to characterize the translation operation corresponding to the first protein or the second protein;

[0014] The protein docking module is further configured to determine a target docking translational conformation based on a first docking index corresponding to each of the translational conformations;

[0015] The protein docking module is further configured to perform docking processing on the first protein and the second protein based on the target docking translation conformation to obtain a protein docking result.

[0016] An embodiment of the present application provides an electronic device, comprising:

[0017] a memory for storing computer-executable instructions or computer programs;

[0018] The processor is configured to implement the protein data processing method provided in the embodiment of the present application when executing the computer executable instructions or computer program stored in the memory.

[0019] An embodiment of the present application provides a computer-readable storage medium storing computer-executable instructions or a computer program, which is used to implement the protein data processing method provided in the embodiment of the present application when executed by a processor.

[0020] An embodiment of the present application provides a computer program product, including a computer program or computer-executable instructions. When the computer program or computer-executable instructions are executed by a processor, the protein data processing method provided in the embodiment of the present application is implemented.

[0021] The embodiments of the present application have the following beneficial effects:

[0022] Representing the data of different postures of the protein in the form of grid data and calculating the space where the protein is located in the grid form can improve the accuracy of the calculation. The first grid data is assigned values ​​from multiple dimensions for each posture, and the docking indicators corresponding to different translational conformations are determined based on the second grid data of different dimensions, thereby improving the accuracy of determining the docking indicators. Based on the docking indicators of each translational conformation, the translational conformation used to perform the docking process is determined, thereby improving the accuracy of the protein docking process. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the application mode of the protein data processing method provided in the embodiment of the present application;

[0024] Figure 2 is a structural diagram of an electronic device provided in an embodiment of the present application;

[0025] Figure 3A This is a first flow chart of the protein data processing method provided in an embodiment of the present application;

[0026] Figure 3B 2 is a schematic diagram of a second flow chart of the method for processing protein data provided in an embodiment of the present application;

[0027] Figure 3C 3 is a schematic diagram of a third flow chart of the protein data processing method provided in an embodiment of the present application;

[0028] Figure 3D 4 is a schematic diagram of a fourth flow chart of a method for processing protein data provided in an embodiment of the present application;

[0029] Figure 4 5 is a schematic diagram of a fifth flow chart of the protein data processing method provided in an embodiment of the present application;

[0030] Figure 5 is a schematic diagram of grid data provided in an embodiment of the present application;

[0031] Figure 6 It is a schematic diagram of the principle of three-dimensional convolution processing provided in an embodiment of the present application.

[0032] It should be pointed out that the above-mentioned "first" and "second" are only used to distinguish different solutions, and do not represent the degree of distinction between the advantages and disadvantages of the solutions or the priority in the implementation process. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting this application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0034] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0035] If similar descriptions of "first / second" appear in the application documents, the following explanation is added. In the following description, the terms "first\second\third" involved are merely used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0036] It should be pointed out that the collection and processing of relevant data in this application should be strictly in accordance with the requirements of relevant national laws and regulations when applied in practice, and the informed consent or separate consent of the personal information subject should be obtained. Subsequent data use and processing should be carried out within the scope of authorization of laws and regulations and the personal information subject.

[0037] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program that has a predetermined function and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0038] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0039] Before further explaining the embodiments of the present application in detail, the nouns and terms involved in the embodiments of the present application are explained. The nouns and terms involved in the embodiments of the present application are subject to the following interpretations.

[0040] 1) Protein docking: A computational process for obtaining a model of a complex formed by two or more protein molecules using energy minimization and other rules. This is illustrated in the examples of this application using the docking between two proteins as an example.

[0041] 2) Antigen (Ag): refers to a substance that can stimulate the body to produce a (specific) immune response and can combine with the immune response products antibodies and sensitized lymphocytes in vitro to produce an immune effect (specific reaction).

[0042] 3) Antibodies: These are molecules produced by B lymphocytes that play a vital role in the immune system. Antibodies are typically composed of two heavy chains and two light chains. Light chains (L chains) are the smaller peptide chains in immunoglobulins, containing 214 amino acids. They are divided into two types: kappa light chains and lambda light chains, based on their structure and constant region antigenicity. The heavy chain of an immunoglobulin, also known as the H chain, is a polypeptide chain within an immunoglobulin monomer.

[0043] 4) Rigid body: This is an idealized physical concept. It refers to an object whose shape and size remain unchanged under the action of a force, and the relative positions of its internal points remain unchanged. In real life, an absolutely rigid body does not exist, because all objects will undergo a certain degree of deformation after being acted upon by a force. Even very small deformations may affect the function and structure of the object. However, when studying the movement and force conditions of an object, certain solids can be approximated as rigid bodies to simplify the problem. In the examples of this application, proteins are used as rigid bodies.

[0044] 5) Translational Conformation: This refers to the spatial shift of atoms or groups within a molecule without altering the molecule's basic skeletal structure. Translational conformation is typically manifested by changes in non-bonded bond angles and electron cloud distribution within the molecule. These changes can be measured through quantum mechanical calculations and experimental methods such as X-ray crystallography.

[0045] 6) Ligand: In chemistry and biology, a ligand generally refers to a molecule or ion that can form a specific non-covalent interaction with another molecule (usually a protein). Ligands are typically small molecules with specific structures and chemical properties that bind to receptors and produce specific effects through reactions. Ligands can participate in a range of biological processes, such as signal transduction, drug action, cell recognition, and assembly.

[0046] 7) Complementarity Determining Regions (CDRs): These are the parts of an antibody molecule that bind to the antigen. Each antibody molecule has six CDRs, namely CDR1, CDR2, and CDR3 of the light and heavy chains. CDR3 has the greatest diversity and is the main source of antibody diversity.

[0047] 8) Immunogenetics Database (IMGT): This database is dedicated to the storage and analysis of immune genes, including antibody and T-cell receptor genes. The IMGT provides a standardized nomenclature and numbering system for antibodies and T-cell receptors, and is widely used in immunology research and antibody engineering.

[0048] 9) Thrust Template Library: This is a C++ template library for the Compute Unified Device Architecture (CUDA), similar to the Standard Template Library (STL). The Thrust Template Library provides an interface fully compatible with CUDA devices for efficiently writing high-performance parallel programs. The Thrust Template Library offers a rich set of data-parallel algorithms, such as scan, sort, and reduce, allowing for the simple and quick construction of complex algorithms while making the code more readable. Using these highly abstract methods to express computations hides details, and the Thrust Template Library automatically selects the most efficient algorithm implementation.

[0049] 10) Van der Waals radius: This property describes the approximate size of an atom or molecule when it is not bound to other atoms. The van der Waals radius represents the "size" of interactions between atoms or molecules and is often used to describe the spatial arrangement and range of interactions between molecules.

[0050] The embodiments of the present application provide a protein data processing method, a protein data processing device, an electronic device, a computer-readable storage medium, and a computer program product, which can improve the accuracy of protein docking.

[0051] The following describes exemplary applications of electronic devices provided in embodiments of the present application. The electronic devices provided in embodiments of the present application can be implemented as terminal devices, such as laptop computers, tablet computers, desktop computers, set-top boxes, smart TVs, vehicle-mounted terminals, virtual reality (VR) devices, augmented reality (AR) devices, and other types of terminals, and can also be implemented as servers. The following describes exemplary applications of electronic devices implemented as terminal devices or servers.

[0052] refer to Figure 1 , Figure 1 This is a schematic diagram of the application mode of the protein data processing method provided in the embodiment of the present application; for example, Figure 1 The server 200, the network 300, the terminal device 400 and the database 500 are involved. The terminal device 400 is connected to the server 200 via the network 300. The network 300 can be a wide area network or a local area network, or a combination of the two.

[0053] For example, database 500 stores a large amount of protein data, and terminal device 400 can be a technician's mobile phone or computer. In response to a docking request issued by terminal device 400, server 200 determines the first and second proteins for protein docking from database 500 based on the protein numbers contained in the docking request. Server 200 then invokes the protein data processing method provided in the embodiments of this application based on the first and second proteins to obtain a protein docking result. Server 200 transmits the protein docking result, the corresponding translational conformation, and other information to terminal device 400.

[0054] In some embodiments, the protein data processing methods of the embodiments of the present application can also be applied in the following application scenarios: 1. Antibody drug research and development. In the field of antibody drug research and development, the protein data processing methods of the embodiments of the present application are used to predict the docking results between the antibody protein drug and the target antigen protein, determine the antibody protein drug corresponding to the best docking result, and improve research and development efficiency. 2. Protein molecule identification. The protein data processing methods of the embodiments of the present application are used to predict the docking results between unknown types of proteins and known types of proteins, and determine the interactions between proteins based on the docking results to improve the efficiency of identifying protein molecules.

[0055] The embodiments of the present application can be implemented through blockchain technology. The protein docking results of the embodiments of the present application can be uploaded to the blockchain for storage, and the reliability of the protein docking results can be guaranteed by the consensus algorithm. Blockchain is a new application model of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains information about a batch of protein docking results, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. The blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer.

[0056] The embodiments of the present application can be implemented using database technology. A database, in short, can be considered an electronic filing cabinet that stores electronic files, allowing users to add, query, update, and delete data in these files. A "database" is a collection of data that is stored together in a specific manner, can be shared by multiple users, has minimal redundancy, and is independent of applications.

[0057] A database management system (DBMS) is a computer software system designed for managing databases, typically providing basic functions such as storage, retrieval, security, and backup. DBMSs can be categorized by the database model they support, such as relational or XML (Extensible Markup Language); by the type of computer they support, such as server clusters or mobile phones; by the query language they use, such as SQL or XQuery; by performance priorities, such as maximum scale or maximum speed; or by other classification methods. Regardless of the classification method used, some DBMSs are cross-category, for example, supporting multiple query languages ​​simultaneously.

[0058] The embodiments of the present application can also be implemented through cloud technology. Cloud technology is a general term for network technology, information technology, integration technology, management platform technology, application technology, etc. based on the cloud computing business model. It can form a resource pool that can be used on demand and is flexible and convenient. Cloud computing technology will become an important support. The background services of the technical network system require a large amount of computing and storage resources, such as video websites, picture websites and more portal websites. With the rapid development and application of the Internet industry, as well as the promotion of search services, social networks, mobile commerce and open collaboration, each item may have its own hash code identification mark in the future, and all of them need to be transmitted to the background system for logical processing. Data of different levels will be processed separately. All kinds of industry data require strong system backing support, which can only be achieved through cloud computing.

[0059] The embodiments of the present application can be implemented through artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that machines have the functions of perception, reasoning and decision-making.

[0060] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, pre-trained models, operating / interaction systems, and mechatronics. Pre-trained models, also known as large models or basic models, can be fine-tuned and widely applied to downstream tasks across various AI disciplines. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0061] Machine learning (ML) is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications span all areas of AI. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning. Pretrained models are the latest development in deep learning, integrating these techniques.

[0062] In some embodiments, the server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be 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, CDN, and big data and artificial intelligence platforms. The electronic device can be a smart phone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited to this. The terminal device and the server can be directly or indirectly connected through wired or wireless communication, which is not limited in the embodiments of the present application.

[0063] See also Figure 2 , Figure 2 is a structural diagram of an electronic device provided in an embodiment of the present application, the electronic device may be a server 200, Figure 2 The server 200 shown includes: at least one processor 410, a memory 450, and at least one network interface 420. The various components in the server 200 are coupled together via a bus system 440. It is understood that the bus system 440 is used to achieve connection and communication between these components. In addition to the data bus, the bus system 440 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, the bus system 440 is not described in detail. Figure 2 Various buses are labeled as bus system 440 .

[0064] The processor 410 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0065] The memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state memory, hard drives, optical drives, etc. The memory 450 may optionally include one or more storage devices that are physically remote from the processor 410.

[0066] The memory 450 includes volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be a read-only memory (ROM), and the volatile memory may be a random access memory (RAM). The memory 450 described in the embodiments of the present application is intended to include any suitable type of memory.

[0067] In some embodiments, the memory 450 can store data to support various operations, examples of which include programs, modules, and data structures, or a subset or superset thereof, as exemplified below.

[0068] Operating system 451, including system programs for processing various basic system services and performing hardware-related tasks, such as the framework layer, core library layer, and driver layer, which are used to implement various basic services and process hardware-based tasks;

[0069] A network communication module 452 is used to reach other electronic devices via one or more (wired or wireless) network interfaces 420. Exemplary network interfaces 420 include Bluetooth, Wi-Fi, and Universal Serial Bus (USB).

[0070] In some embodiments, the apparatus provided in the embodiments of the present application may be implemented in software. Figure 2 A device 455 for processing protein data stored in memory 450 is shown. This device can be software in the form of a program or plug-in, and includes the following software modules: a data filling module 4551 and a protein docking module 4552. These modules are logical and can be arbitrarily combined or further separated according to the functions they implement. The functions of each module will be described below.

[0071] The protein data processing method provided in the embodiments of the present application will be described in conjunction with the exemplary application and implementation of the terminal device provided in the embodiments of the present application.

[0072] The following describes the protein data processing method provided by the embodiments of the present application. As previously described, the electronic device implementing the protein data processing method of the embodiments of the present application can be a terminal device or a server, or a combination of the two. Therefore, the execution entity of each step will not be repeated below.

[0073] See also Figure 3A , Figure 3A This is a flow chart of the protein data processing method provided in the embodiment of the present application, which will be combined with Figure 3A The steps shown are explained. Figure 3A The body of the step is Figure 1 Server in.

[0074] In step 301 , first grid data of the first protein and the second protein are filled from multiple dimensions according to different postures of the first protein and the second protein to obtain second grid data of multiple dimensions corresponding to each posture.

[0075] For example, in an embodiment of the present application, the first grid data and the second grid data are protein grid data. Protein grid data generally refers to three-dimensional protein molecular structure data, which is usually presented in the form of a grid and is used to display the spatial three-dimensional structure of protein molecules. These data are usually obtained through structural biology methods such as X-ray crystal diffraction, nuclear magnetic resonance, etc., and are obtained through computer processing and analysis. Before the first grid data is filled, the grid size corresponding to the first grid data can enclose the first protein and the second protein in the three-dimensional space. The content of the first grid data is empty, and the second grid data is the first grid data after filling.

[0076] In some embodiments, before step 301, the first protein or the second protein is fixed in three-dimensional space, and the unfixed protein is rotated around each coordinate axis in the three-dimensional space to obtain different postures between the first protein and the second protein, wherein the first protein and the second protein are rigid bodies.

[0077] For example, any one of the antigen protein and the antibody protein is fixed, and the coordinate axis of the other protein is rotated in different directions to obtain multiple different postures. Each time the posture is rotated, the coordinates of all atoms of the rotated protein will change, so the grid data of the protein needs to be reassigned. The translational conformation refers to the translational change in the spatial position of atoms or groups in the molecule without changing the basic skeletal structure of the molecule. As the posture changes, the corresponding translational conformation in each posture is also different. Different postures can be represented by Euler angles or quaternions. Euler angles can be used to describe rotations at three angles. The basic forms of rotation include: rotation order around fixed axes (ZXZ, ZYZ, etc.): Assuming that the fixed axes are the first fixed axis, the second fixed axis, and the third fixed axis, first rotate around the first fixed axis, then around the second fixed axis, and finally around the third fixed axis. The rotation order can be different.

[0078] In some embodiments, the process of obtaining the first grid data is explained below, referring to Figure 3D , Figure 3D This is a fourth flow chart of the protein data processing method provided in the embodiment of the present application. Before step 301, you can execute Figure 3D Steps 3051 to 3054 are described in detail below.

[0079] In step 3051, the protein structure information of the first protein and the second protein are converted to obtain a first amino acid sequence of the first protein and a second amino acid sequence of the second protein.

[0080] Protein structure information refers to protein structure data measured by techniques such as nuclear magnetic resonance (NMR) or X-ray crystallography and recorded in the Protein Data Bank (PDB). PDB structural data is usually stored in the PDB format. Protein structure information includes the following types of information: biomacromolecule descriptions (ATOM, HETATM), which contain information such as the coordinates, types, and chemical properties of atoms in biomacromolecules; binding information (CONECT), which contains the bonding relationships between atoms; and structural information (HELIX, SHEET), which contains structural information such as helices and β-sheets in biomacromolecules.

[0081] For example, based on the order of the amino acids in the peptide chains of the proteins, the protein structure information corresponding to each amino acid in the first and second proteins is converted into a first amino acid sequence and a second amino acid sequence. The first protein is an antigen protein, and the second protein is an antibody protein. If the antibody protein is composed of two peptide chains, a heavy chain and a light chain, then the heavy chain and the light chain are logically considered to be one protein. The sequences of the amino acid residues corresponding to the heavy and light chains are combined in order to form the sequence of the amino acid residues of one chain, thereby obtaining the second amino acid sequence of the second protein.

[0082] In step 3052, the first amino acid sequence and the second amino acid sequence are respectively coded to obtain a first amino acid feature of the first protein and a second amino acid feature of the second protein.

[0083] For example, the encoding process can be implemented by calling the Immunogenetics database (IMGT), which provides a naming and numbering system for antibodies and T cell receptors. Based on the first amino acid sequence and the second amino acid sequence, the Immunogenetics database is called to obtain a first amino acid feature of the first protein and a second amino acid feature of the second protein.

[0084] In step 3053, the complementarity determining region residue indices in the second amino acid feature are marked to obtain a third amino acid feature.

[0085] For example, the second amino acid signature belongs to an antibody protein, and the complementarity determining region is the portion of the antibody molecule that binds to the antigen. The third amino acid signature is obtained by marking multiple amino acid residues in the second amino acid signature that correspond to the complementarity determining region.

[0086] In step 3054, first grid data is constructed based on the first amino acid feature and the third amino acid feature.

[0087] For example, the protein structure information of the first protein and the second protein is preprocessed to obtain the first amino acid feature and the third amino acid feature, the first amino acid feature and the third amino acid feature are traversed to determine the range of the first protein and the second protein in the three-dimensional space, and a grid that can contain the first protein and the second protein is constructed according to the corresponding range to obtain the first grid data.

[0088] In some embodiments, step 3054 may be implemented by determining a coordinate boundary of the first protein and the second protein in three-dimensional space based on the first amino acid signature and the third amino acid signature, determining a grid space size and a cell size of the first grid data based on the coordinate boundary, and constructing the first grid data based on the grid space size and the cell size.

[0089] For example, multiple coordinate points corresponding to the atoms of the first protein and the second protein are determined, and the coordinate point farthest from the origin in the positive and negative directions of the three coordinate axes in the three-dimensional space (Z axis, X axis, Y axis) is determined among the multiple coordinate points to obtain the coordinate boundary on the coordinate axis. For example: for the X axis, the coordinate point farthest from the origin in the negative direction of the X axis is determined, and the point farthest from the origin in the positive direction of the X axis is determined to obtain the coordinate boundary on the X axis. The coordinate boundaries on the other axes are obtained respectively to obtain the coordinate boundaries of the first protein and the second protein in the three-dimensional space, and the length corresponding to each coordinate boundary is determined. For example: for the X axis, the length corresponding to the coordinate boundary is the difference between the coordinate point farthest from the origin in the negative direction of the X axis and the point farthest from the origin in the positive direction of the X axis on the X axis. Then the grid space size is larger than the length corresponding to each coordinate boundary to reserve for the calculation of related energy. The portion of the grid space size that is slightly larger than the length corresponding to each coordinate boundary can be determined according to the actual application scenario. In the embodiment of the present application, 10% is used as an example for illustration.

[0090] For example, if the lengths of the protein on the X, Y, and Z axes are 5, 8, and 10, respectively, the preconfigured side length for the cell is 1. The lengths 5, 8, and 10 on the X, Y, and Z axes are expanded by 10% to 5.5, 8.8, and 11, respectively. Next, the number of cells needed to fill each axis is calculated, and the size of the space is divided by the preconfigured side length to get the number of cells corresponding to each axis. On the X axis, ceil(5.5 / 1) = 6, on the Y axis, ceil(8.8 / 1) = 9, and on the Z axis, ceil(11 / 1) = 11. Therefore, the entire grid space contains a total of 6911 = 594 cells.

[0091] For example, multiple dimensions refer to different types of intermolecular interaction forces, and the dimensions can be chemical bonds, van der Waals forces, electrostatic interactions, polar interactions, and dipole interactions. Among them, the types of chemical bonds include ionic bonds and hydrogen bonds. The embodiments of the present application are applied to protein molecules, and the van der Waals forces between proteins and ligands have shape complementarity. Shape complementarity means that in a protein-ligand complex, the ligand usually complements the active site of the protein through its specific shape.

[0092] Filling the first grid data for the first and second proteins in multiple dimensions can be accomplished by filling each cell in the first grid data with a value or index representing the corresponding type of intermolecular interaction force, thereby obtaining second grid data corresponding to each dimension. The value or index of the intermolecular interaction force is determined by the distance between the cell and the atoms in the molecule and the force acting on the cell.

[0093] The dimensions exemplified in the embodiments of this application can be divided into the following types according to the factors that form intermolecular forces: shape complementary dimension, ionic bond dimension and hydrogen bond dimension, among which the shape complementary dimension corresponds to van der Waals force. The filling method of each dimension will be explained below.

[0094] In some embodiments, reference Figure 3B , Figure 3B This is a second flow chart of the protein data processing method provided in the embodiments of the present application. Figure 3A Step 301 can be achieved by Figure 3B Steps 3011 to 3013 are implemented as described below.

[0095] In step 3011 , the following processing is performed on the first protein and the second protein in different postures respectively: each atom of the first protein and the second protein is traversed, and the traversed atom is used as the current atom.

[0096] For example, the order of traversing each atom of the first protein and the second protein can be traversed according to the order of atoms in the amino acid sequence of the protein, or traversed row by row according to the position of the atoms in the three-dimensional space of the first grid data.

[0097] In step 3012, each first cell associated with the current atom in the first grid data is traversed based on each dimension.

[0098] For example, each first cell associated with the current atom refers to a cell within a preconfigured range from the current atom, and a method for traversing the first cells may be a binary fast search.

[0099] In some embodiments, step 3012 may be implemented by determining a second cell closest to the currently traversed atom, determining a cube centered at the second cell and having a radius of a first preconfigured length, wherein the cube includes a plurality of first cells, and traversing each first cell in the cube.

[0100] For example, candidate cells closest to the current atom in the positive and negative directions of each coordinate axis in three-dimensional space are obtained, and the cell closest to the current atom is selected from the candidate cells as the second cell. A cube with a radius of a first preconfigured length is constructed with the second cell as the center. If the cube includes multiple first cells, the second cell at the center of the cube can also serve as the first cell.

[0101] In step 3013, data is filled in the traversed first cell based on the data filling method of different dimensions to obtain the filled first cell of different dimensions.

[0102] In the examples of this application, the different dimensions include shape complementarity, ionic bonding, and hydrogen bonding. The fill data of the shape complementarity dimension is used to characterize the constraints of the complementarity-determining regions of antibodies, the van der Waals forces between molecules, or the van der Waals forces between atoms within a molecule. The ionic bonding dimension is used to characterize the forces exerted by ionic bonds on proteins during protein docking. The hydrogen bonding dimension is used to characterize the forces exerted by hydrogen bonds on proteins during protein docking. Each dimension is explained below.

[0103] In some embodiments, when the dimension is a shape-complementary dimension, step 3013 may be implemented by filling the first traversed cell with data based on the data filling method of the shape-complementary dimension to obtain the filled first cell of the shape-complementary dimension.

[0104] For example, the value of each cell is divided into real and imaginary parts. A positive influence on the cell increases the real part, while a penalty influence increases the imaginary part. A positive influence refers to positive feedback on protein docking, while a penalty refers to negative feedback on protein docking. The shape complementarity dimension determines whether to fill the corresponding data for the first cell based on the van der Waals radius. The van der Waals radius represents the "size" of the interaction between atoms or molecules and is often used to describe the spatial arrangement and interaction range of molecules.

[0105] Example methods for filling data with complementary dimensions include:

[0106] Method 1: In response to the current atom being within the complementarity determination region and the distance between the first unit cell and the current atom being less than a second preconfigured length, a first real parameter is configured for the first unit cell. For example, if the second preconfigured length is 6 angstroms, a first real parameter (x, 0i) is applied to the real part of the first unit cell within 6 angstroms of the atom.

[0107] Method 2: In response to the distance between the first cell and the current atom being less than the van der Waals radius of the current atom, a first imaginary parameter is configured for the first cell. For example, if the van der Waals radius of the current atom is 2.8 angstroms and the distance between the first cell and the current atom is less than 2.8 angstroms, the first imaginary parameter is configured for the first cell (0, 9i).

[0108] Mode 3: In response to the distance between the first unit cell and the current atom being greater than or equal to the van der Waals radius, a second real parameter is configured for the first unit cell. For another example, if the distance between the first unit cell and the current atom is greater than 2.8 angstroms of the van der Waals radius, the second real parameter (1, 0i) is configured for the first unit cell.

[0109] In some embodiments, when the dimension is an ionic bond dimension, step 3013 can be implemented by performing data filling on the first cell traversed based on the data filling method for the ionic bond dimension, thereby obtaining the filled first cell of the ionic bond dimension. Ionic bond types include anions and cations. For example, asparagine (Asn) and glutamine (Gln) have anions, while lysine (Lys) and arginine (Arg) have cations.

[0110] For example, the data filling methods for the ionic bond dimension include:

[0111] Method 1: In response to the current atom being a positive ion and the distance between the first unit cell and the current atom being less than a third preconfigured length, configuring a third real parameter for the first unit cell, wherein the third real parameter is a negative real parameter. For example, the third preconfigured length may be 4.5 angstroms. When the distance between the first unit cell and the current atom being a positive ion is less than 4.5 angstroms, configuring the third real parameter to be (-4, 0i).

[0112] Mode 2: In response to the current atom being an anion and the distance between the first unit cell and the current atom being less than a third preconfigured length, configuring a fourth real parameter for the first unit cell, wherein the fourth real parameter is a positive real parameter. For example, the third preconfigured length may be 4.5 angstroms. When the distance between the first unit cell and the current atom being an anion is less than 4.5 angstroms, configuring the third real parameter (4, 0i).

[0113] For example, based on the above examples, the superposition product of like ions is 16, and the superposition product of opposite ions is -16. By inverting the results of all ionic bonds, we can obtain that like ions repel each other, which has negative feedback on the protein docking results, and opposite ions attract each other, which has positive feedback on the protein docking results.

[0114] In some embodiments, when the dimension is a hydrogen bond dimension, step 3013 can be implemented as follows: in response to the current atom being a hydrogen bond acceptor, data is filled in the first cell traversed based on the data filling method of the hydrogen bond dimension to obtain the filled first cell of the hydrogen bond dimension. With respect to hydrogen bonds, the oxygen, nitrogen, and sulfur atoms in amino acids such as glycine (Gly), serine (Ser), threonine (Thr), tyrosine (Tyr), asparagine (Asn), glutamine (Gln), cysteine ​​(Cys), aspartic acid (Asp), glutamic acid (Glu), lysine (Lys), histidine (His), and arginine (Arg) can all serve as hydrogen bond acceptors.

[0115] For example, the data filling methods for hydrogen bond dimension include:

[0116] Method 1: In response to the current atom being a hydrogen atom and the distance between the first cell and the current atom being less than a fourth preconfigured length, a fifth real parameter and a second imaginary parameter are configured for the first cell, wherein the fifth real parameter is a positive real parameter and the second imaginary parameter is a positive imaginary parameter. For example, if the fourth preconfigured length is 3.5 angstroms, the current atom is a hydrogen atom, and the distance between the first cell and the current atom is less than 3.5 angstroms, the fifth real parameter and the second imaginary parameter are configured for the first cell to obtain parameter (1, 1i). Wherein, (1, 0) is the fifth real parameter and (0, 1i) is the second imaginary parameter.

[0117] Method 2: In response to the current atom not being a hydrogen atom and the distance between the first unit cell and the current atom being less than the fourth preconfigured length, a sixth real parameter and a third imaginary parameter are configured for the first unit cell, wherein the sixth real parameter is a positive real parameter and the third imaginary parameter is a negative imaginary parameter. For example, if the fourth preconfigured length is 3.5 angstroms, the current atom is a nitrogen atom, and the distance between the first unit cell and the current atom is less than 3.5 angstroms, the sixth real parameter and the third imaginary parameter are configured for the first unit cell to obtain parameter (1, -1i). Wherein, (1, 0) is the sixth real parameter and (0, -1i) is the third imaginary parameter.

[0118] For example, for hydrogen bonds, one end must be a donor and the other an acceptor. Therefore, the grid around the acceptor and donor is assigned values ​​of (1, 1i) and (1, -1i), respectively. Therefore, for complex number operations, the multiplication of the donor and acceptor is equal to 2i and -2i. Both results have only imaginary parts and no real parts. When calculating docking metrics, only the real part is counted, so it has no effect on the results. If one end is an acceptor and the other end is a donor, the product is 2, which has positive feedback on the protein docking results.

[0119] In the embodiment of the present application, the grid data of the protein is replicated in multiple different dimensions, taking into account the role of chemical bonds (hydrogen bonds and ionic bonds) and complementarity determining regions in the protein docking process, thereby improving the accuracy of the assignment function, thereby improving the accuracy of determining the optimal translational conformation, and improving the protein docking efficiency and the accuracy of the docking results.

[0120] Continue to refer Figure 3B In step 3014, each first grid data after data filling is used as second grid data of different dimensions to obtain second grid data of multiple dimensions corresponding to the posture.

[0121] For example, when each first cell in the first grid data is traversed and data is filled, the first grid data after data filling is obtained. The same posture corresponds to multiple dimensions, and each dimension corresponds to a second grid data. Therefore, after data filling, the second grid data of multiple dimensions corresponding to the posture is obtained.

[0122] In the embodiment of the present application, the first grid data is filled with data in different ways to obtain second grid data of different dimensions, which takes into account the various forces in the protein docking process and can improve the accuracy of predicting protein docking results.

[0123] Continue to refer Figure 3A In step 302, based on the second grid data of multiple dimensions corresponding to each posture and each translation conformation corresponding to the posture, a first docking index corresponding to each translation conformation is determined.

[0124] For example, the translational conformation is used to characterize the translational operation of the first protein or the second protein during protein docking.

[0125] In some embodiments, reference Figure 3C , Figure 3C This is a third flow chart of the protein data processing method provided in the embodiment of the present application. Step 302 can be performed by Figure 3C Steps 3021 to 3022 are implemented as described below.

[0126] In step 3021 , the following processing is performed for each translational conformation corresponding to the posture: based on the translational conformation and the second grid data of multiple dimensions, second docking indicators corresponding to the multiple dimensions are determined.

[0127] For example, if each translational conformation is different, then the translation results of the current posture of the protein based on the translational conformation are different, and the overlap between the proteins obtained after translation is different. Based on the portion of the grid data corresponding to the overlap between the proteins corresponding to each translational conformation, a second docking metric for the translational conformation with respect to the grid data is determined.

[0128] In some embodiments, step 3021 can be implemented by determining an overlapping portion between the first protein and the second protein in the second grid data based on the translational conformation, wherein the overlapping portion is the portion where the first protein and the second protein overlap during the protein docking process. For each dimension of the second grid data, the following processing is performed: extracting third grid data corresponding to the overlapping portion from the second grid data corresponding to the dimension, performing a three-dimensional convolution process on the third grid data, and obtaining a second docking index corresponding to the dimension.

[0129] For example, the input data of a three-dimensional convolution operation is usually a three-dimensional tensor. Similar to two-dimensional convolution, a three-dimensional convolution operation also uses a filter (convolution kernel) to extract features. The filter is a three-dimensional tensor whose size corresponds to the depth, height, width, and number of channels of the input data. In three-dimensional convolution, the filter moves along the depth, height, and width of the input data, multiplies the data at the corresponding position element by element, and adds all the elements to obtain a value in the output feature map. The second docking indicator is the combination of the results of the three-dimensional convolution processing corresponding to each cell in the overlapping part.

[0130] In some embodiments, the three-dimensional convolution processing is implemented by a convolutional neural network, which includes a three-dimensional convolutional neural network layer.

[0131] In step 3022, the sum of each second docking index is used as the first docking index corresponding to the translational conformation.

[0132] For example, the second docking indicators of each dimension are added together to obtain the first docking indicator. In some embodiments, the importance of different dimensions can be determined according to the specific application scenario, and the weight value of the second docking indicator of each dimension can be determined according to the importance of each dimension. The weighted sum of the second docking indicators of each dimension is used as the first docking indicator.

[0133] Continue to refer Figure 3A In step 303, a target docking translational conformation is determined based on the first docking index corresponding to each translational conformation.

[0134] For example, the score of the first docking metric is positively correlated with the docking result, that is, the higher the score of the first docking metric, the better the docking result. The highest first docking metric is selected from the first docking metrics corresponding to each translational conformation, and the translational conformation with the highest first docking metric is used as the target docking translational conformation.

[0135] In some embodiments, step 303 can be implemented by calling an interface of a parallel computing architecture based on the first docking index corresponding to each translational conformation, sorting each first docking index to obtain a sorted list, and selecting the translational conformation corresponding to the largest first docking index in the sorted list as the target docking translational conformation.

[0136] For example, the sorted list can be a descending sorted list or an ascending sorted list. For a descending sorted list, the first docking indicator ranked first in the list is the largest first docking indicator. For an ascending sorted list, the first docking indicator ranked last in the list is the largest first docking indicator.

[0137] In some embodiments, the C++ template library Thrust library in the general parallel computing architecture CUDA environment can be used to accelerate the parallel sorting of image processors.

[0138] In step 304, docking processing is performed on the first protein and the second protein based on the target docking translation conformation to obtain a protein docking result.

[0139] In some embodiments, step 304 can be implemented in the following manner: based on the target docking translation conformation, performing translation processing on the first protein or the second protein in the target posture to obtain a protein docking result, wherein the target posture is the posture between the first protein and the second protein corresponding to the target docking translation conformation.

[0140] For example, based on the index of the translational conformation, a database is called to query the translation operation corresponding to the translational conformation, and based on the translation operation corresponding to the translational conformation, the protein that is not set to a fixed state in the first protein or the second protein is translated to obtain the docking result of the protein.

[0141] In an embodiment of the present application, data of different postures of the protein are represented in the form of grid data, and calculating the space where the protein is located in the form of a grid can improve the accuracy of the calculation. The first grid data is assigned values ​​from multiple dimensions for each posture, and the docking indicators corresponding to different translational conformations are determined based on the second grid data of different dimensions, thereby improving the accuracy of determining the docking indicators, and based on the docking indicators of each translational conformation, the translational conformation used to perform the docking process is determined, thereby improving the accuracy of the protein docking process.

[0142] Below, an exemplary application of the protein data processing method of the embodiment of the present application in a practical application scenario will be described.

[0143] Protein docking is used in many fields such as antibody drug development, antibody design, and molecular recognition. The protein docking solutions of related technologies include: (1) Molecular dynamics simulation: This is a physics-based method that predicts the interaction between antibodies and antigens by simulating the Newtonian equations of molecular motion. This method can provide detailed dynamic information, but the computational cost is high and usually requires large-scale parallel computing resources. (2) Monte Carlo simulation: This is a statistical method that explores the possible binding modes of antibodies and antigens by randomly changing the conformations of antibodies and antigens and then using the Metropolis criterion to decide whether to accept this change. This method can handle large-scale search spaces, but may require a large number of simulation steps to reach convergence. (3) Free energy calculation: This is a thermodynamics-based method that predicts the binding force of antibodies and antigens by calculating the free energy difference between binding and unbinding. This method can provide accurate binding force predictions, but the computational cost is high and requires large-scale parallel computing resources. (4) Structural alignment: This is a template-based method that predicts the binding mode of antibodies and antigens by comparing the structural similarity of antibodies and known antigen-antigen complexes. Structural comparison methods rely on known structural data and may not be able to handle new antigen-antibody pairings, thereby limiting the scope of applicability of structural comparison methods and resulting in low versatility.

[0144] Related technologies, such as molecular dynamics simulations and free energy calculations, require extensive computing resources. These methods typically require large-scale parallel computing environments, which can be impractical for the hardware equipment available in most research laboratories. Related technologies also fail to fully exploit the advantages of parallel computing in image processor optimization, which can lead to computational inefficiencies and prolonged research cycles. Furthermore, the scoring functions of these technologies fail to fully account for the complex interactions between antibodies and antigens, resulting in inaccurate predictions.

[0145] The present application embodiment is directed to the above-mentioned problems existing in the related art, proposes a method for processing protein data, by considering the characteristics of antigen protein and antibody protein, from the ionic bond, hydrogen bond and complementary determination region of antigen protein and antibody protein, more accurately predict the interaction of antibody and antigen, to enhance the accuracy of protein docking. More effectively utilize computing resources, by using parallel computing and image processor optimization, computing efficiency can be significantly improved, thereby saving computing resources. This makes large-scale antigen-antibody docking research possible, and can be applicable to resource-limited environments. Not only can known antigen-antibody pairings be processed, new protein pairings can also be processed. Adapting to various protein structures and chemical properties, the scope of application of the processing method of the protein data of the present application embodiment has been expanded, and various complex biomolecule docking problems can be addressed.

[0146] The following is explained with reference to the accompanying drawings. Figure 4 , Figure 4 This is the fifth flow chart of the protein data processing method provided in the embodiments of the present application, which is described in detail below.

[0147] In step 401, the encoding of protein structure information of the antigen protein and the antibody protein is obtained.

[0148] For example, protein structure information refers to protein structure data measured by techniques such as nuclear magnetic resonance (NMR) or X-ray crystallography and recorded in the Protein Data Bank (PDB). PDB structural data is usually stored in the PDB format, which contains information about the atomic coordinates, binding state, etc. of biological macromolecules. Taking the processing process of antibody protein as an example, the protein structure information of the antibody protein is converted into a sequence of amino acid residues, and the sequence of amino acid residues is re-encoded using the immune genome database encoding. After re-encoding, the complementary determining region residue index on the antibody is marked. If the antibody is composed of two peptide chains, heavy chain and light chain, they need to be logically regarded as one protein of one chain. For example: the sequences of amino acid residues corresponding to the two peptide chains of the heavy chain and the light chain are combined into the sequence of amino acid residues of one chain according to the order of precedence.

[0149] In step 402, a grid to be assigned is constructed based on the processed protein structure information.

[0150] For example, based on the structural information of the recoded and labeled proteins, all atoms of the antigen protein and the antibody protein are scanned to obtain the coordinate boundaries of the atoms in the antigen protein and the antibody protein. The size of the rectangular grid that can encompass the antigen protein and the antibody protein is calculated based on the coordinate boundaries. The total size of the rectangular grid is 10% larger than the size of the protein to reserve for the calculation of related energies. After obtaining the size of the rectangular grid, the size of the cells in the rectangular grid is set, and the cells are filled with the space corresponding to the entire rectangular grid to obtain the constructed grid to be assigned.

[0151] This can be achieved as follows: If the lengths of the protein on the X, Y, and Z axes are 5, 8, and 10, respectively, the preconfigured side length for the cell is 1. The lengths 5, 8, and 10 on the X, Y, and Z axes are expanded by 10% to 5.5, 8.8, and 11, respectively. Then, the number of cells required to fill each axis is calculated, and the size of the space is divided by the preconfigured side length to obtain the number of cells corresponding to each axis. On the X axis, ceil(5.5 / 1) = 6, on the Y axis, ceil(8.8 / 1) = 9, and on the Z axis, ceil(11 / 1) = 11. Therefore, the entire grid space contains a total of 6911 = 594 cells.

[0152] refer to Figure 5 , Figure 5 Grid 501 is a schematic diagram of grid data provided by an embodiment of the present application. Grid 501 is the grid to be filled. The size of grid 501 is larger than the size of protein 502, so that protein 502 is included in grid 501. Grid 501 is filled with the values ​​corresponding to protein 502 to obtain grid data.

[0153] In step 403, a plurality of postures between the antigen protein and the antibody protein are enumerated, and an assignment process is performed based on any posture of the antigen protein and the antibody protein.

[0154] For example, in the process of executing the method for processing protein data provided by the embodiment of the present application, the antigen protein and the antibody protein are regarded as rigid bodies in three-dimensional space, any one of the antigen protein and the antibody protein is fixed, and the coordinate axis of the other protein is rotated in different directions to obtain a plurality of different postures. With each rotation posture, the coordinates of all atoms will change, so it is necessary to re-assign the grid of the protein and enumerate different translational conformations. The various postures between the antigen protein and the antibody protein can be represented by Euler angles or quaternions.

[0155] In some embodiments, the assignment processing of different postures may be serial. If the computing resources are sufficient to support the parallel assignment processing of multiple postures, the assignment processing of different postures may also be parallel.

[0156] Here, step 404, step 405 and step 406 are performed in parallel. Step 404, step 405 and step 406 are different implementations of the assignment process in step 403.

[0157] In step 404, each cell in the grid to be assigned a value is traversed by a binary quick search method, and the traversed cells are assigned a value based on the shape complementary dimension to obtain a grid with the assigned shape complementary dimension.

[0158] As an example, the following explains the binary fast search method. For each atom of the protein, a computing thread is assigned, and a CUDA-based image processor parallel algorithm can be used. According to the X, Y, and Z axis coordinates of the atom, a binary fast search is performed in the entire grid space to find the index of the cell closest to the atom in each coordinate axis direction, and the cell closest to the atom is determined from the nearest cells corresponding to the three coordinate axes. With the cell closest to the atom as the center and the radius of the atomic staining effect (the first preconfigured length above) as the side length, all cells in the cube within the range are obtained, and each cell in the cube is traversed. The assignment of the cell is determined based on the assignment method of different dimensions and the distance between the cell and the atom. Among them, staining is a process of quickly assigning values ​​to the grid around the atom based on the coordinates and category of the atom.

[0159] The following explains how the shape complementarity dimension is assigned. The value of each cell is divided into a real part and an imaginary part. Adding a positive influence to the cell increases the value of the real part, and generating a penalty influence increases the value of the imaginary part. Among them, a positive influence refers to positive feedback on protein docking, and a penalty refers to negative feedback on protein docking. A positive influence means increasing the value of the real part. For example, if the initial value is (0, 0i), if a positive influence is required on this grid, x is added to the real part (x is a positive number), and finally (x, 0i) is obtained. At this time, the influence of this grid on the result is positive. The specific reason is that if this grid is for the receptor, and a grid of another antibody is (y, 0i), y is a positive number, then the product of the two grids is x*y, and the final value is positive, indicating that the influence on the final total score is increasing, so it is a positive influence. The penalty is to increase the value of the imaginary part only. If the feedback of the previous section is (0, xi) and (0, yi), the final product is 0-(x*y). At this time, the impact on the total score (the total score is the sum of the products of all grids) is reduced, so this situation is regarded as a penalty.

[0160] For example: First, for the constraints of the complementarity determining region of an antibody, the residue positions obtained by preprocessing (complementarity determining region) are extracted, and appropriate values ​​(determined according to the specific application scenario) are applied to the real part of the cells within 6 angstroms around the residue positions to enhance positive feedback. The most basic shape complementarity score is applied to the cells. The process of determining the shape complementarity score includes: based on the coordinates and van der Waals radius of each atom, a shape penalty (i.e., an imaginary part parameter) is applied to all cells within a range less than or equal to the van der Waals radius centered on the atomic coordinates. Then, a bonus value (i.e., a real part parameter) is assigned to cells outside the van der Waals radius of each atom and within a certain area to obtain the most basic protein shape energy score of the cell.

[0161] In some implementations, taking shape complementarity as an example, the grid within the range of +2.8 angstroms of the atomic coordinates is centered, and the real part of the score is applied. The specific value of shape complementarity is (1, 0i) and the penalty is (0, 9i).

[0162] In step 405, each cell in the grid to be assigned a value is traversed by a binary quick search method, and the traversed cells are assigned a value based on the ionic bond dimension to obtain a grid of the assigned ionic bond dimension.

[0163] For example, for the ionic bond dimension, asparagine (Asn) and glutamine (Gln) carry anions, and lysine (Lys) and arginine (Arg) carry cations. In the embodiment of the present application, a positive real part is applied to the grid within 4.5 angstroms around the anion, and a negative real part is applied to the grid within 4.5 angstroms around the cation. For example: the grids around the anion and the cation are only assigned values ​​of (4, 0i) and (-4, 0i). The superposition product of the same ions is 16, and the superposition product of the opposite ions is -16. By negating the results of all ionic bonds, we can obtain like-sex repulsion, which forms a negative feedback effect on the protein docking result, and opposite-sex attraction forms a positive feedback effect on the protein docking result.

[0164] In step 406, each cell in the grid to be assigned a value is traversed by a binary quick search method, and the traversed cells are assigned a value based on the hydrogen bond dimension to obtain a grid with the assigned hydrogen bond dimension.

[0165] For example, with respect to the hydrogen bond dimension, the oxygen, nitrogen, and sulfur atoms in amino acids such as glycine (Gly), serine (Ser), threonine (Thr), tyrosine (Tyr), asparagine (Asn), glutamine (Gln), cysteine ​​(Cys), aspartic acid (Asp), glutamic acid (Glu), lysine (Lys), histidine (His), and arginine (Arg) can all serve as hydrogen bond acceptors. In the embodiment of the present application, a positive real part and a negative imaginary part are assigned to the grid within 3.5 angstroms of the oxygen, nitrogen, and sulfur atoms, and a positive real part and a positive imaginary part are assigned to the hydrogen atoms within 3.5 angstroms of the side chains of all residues.

[0166] For example, for hydrogen bonds, one end must be a donor and the other an acceptor. Therefore, the grids around the acceptor and donor are assigned values ​​of (1, 1i) and (1, -1i), respectively. Therefore, for complex number operations, the multiplication of the donor and acceptor equals 2i and -2i. Both results only have imaginary parts, not real parts. However, when performing score statistics in this embodiment, only the real part is counted, so this has no effect on the result. If one end is an acceptor and the other is a donor, the product is 2, forming a positive feedback effect on the final score.

[0167] In step 407 , a plurality of translational conformations corresponding to a plurality of postures between the antigen protein and the antibody protein are enumerated, and docking indices are calculated for each translational conformation and the assigned grids of different dimensions.

[0168] For example, docking metrics are scores for different translational conformations. The score for each translational conformation in the ligand grid is determined by summing the products of the overlap between the receptor grid and the ligand grid. This product of the overlap generally indicates the strength of the interaction between the two, with larger values ​​indicating stronger interactions. The docking metric is the sum of the scores for each translational conformation across different dimensions.

[0169] The process of determining the score of each translational conformation can be viewed as a 3D convolution process. Since convolution in the time domain is equal to multiplication in the frequency domain, this embodiment of the application uses Fourier transforms to accelerate the entire process. This embodiment of the application uses the CU Fourier transform library to call the image processor to accelerate Fourier transform-related calculations.

[0170] refer to Figure 6 , Figure 6Schematic diagram of the principle of three-dimensional convolution processing provided in an embodiment of the present application. In grid data 601, unprocessed portion 602 is the overlapping portion of the antibody protein and the antigen protein. Unprocessed portion 602 is extracted from grid data 601, and three-dimensional convolution processing is performed on unprocessed portion 602 to obtain convolution result 603. Three-dimensional convolution processing is performed on each unprocessed portion of the overlapping portion of the antibody protein and the antigen protein in grid data 601 to obtain the final convolution result.

[0171] In step 408, the template library interface of the parallel computing architecture is called to sort the docking index of each translational conformation to obtain a sorting result.

[0172] For example, the scores of each translation conformation are sorted in descending order. Since the amount of data may be large, the embodiment of the present application uses the C++Thrust template library in the universal parallel computing architecture (CUDA) environment to accelerate the parallel sorting of the image processor.

[0173] In step 409 , the target translational conformation corresponding to the highest docking index is determined based on the ranking results, and the docking result corresponding to the target translational conformation is determined.

[0174] For example, the index of the highest-scoring translational conformation can be used to infer the specific translation operation required. Since the docking metrics for translational conformations include scores from different dimensions such as ionic bonds, hydrogen bonds, and shape complementarity, the scoring accuracy is improved.

[0175] The protein data processing methods provided in the embodiments of this application have the following beneficial effects: They fully utilize the concurrency advantages of image processors, significantly improving computational efficiency when optimizing tasks involving massive computations such as grid coloring, Fourier transforms, and large number sorting. They also consider not only shape complementarity but also the influence of chemical bonds, which can improve docking accuracy. Simultaneously, they optimize the most likely contact areas for antibodies, imposing constraints to increase the probability of successful docking.

[0176] The following continues to describe the exemplary structure of the protein data processing device 455 provided in the embodiments of the present application implemented as a software module. In some embodiments, Figure 2As shown, the software modules in the protein data processing device 455 stored in the memory 450 may include: a data filling module 4551, which is used to fill the first grid data of the first protein and the second protein from multiple dimensions according to the different postures of the first protein and the second protein, and obtain second grid data of multiple dimensions corresponding to each posture; a protein docking module 4552, which is also used to determine the first docking index corresponding to each translation conformation based on the second grid data of multiple dimensions corresponding to each posture and each translation conformation corresponding to the posture, wherein the translation conformation is used to characterize the translation operation corresponding to the first protein or the second protein; the protein docking module 4552, which is also used to determine the target docking translation conformation based on the first docking index corresponding to each translation conformation; the protein docking module 4552 is also used to perform docking processing on the first protein and the second protein based on the target docking translation conformation to obtain a protein docking result.

[0177] In some embodiments, the data filling module 4551 is configured to perform the following processing on the first protein and the second protein in different postures:

[0178] Traverse each atom of the first protein and the second protein, and use the traversed atom as the current atom; traverse each first cell associated with the current atom in the first grid data based on each dimension; fill the traversed first cell with data based on data filling methods of different dimensions to obtain filled first cells of different dimensions; use each first grid data after data filling as second grid data of different dimensions to obtain second grid data of multiple dimensions corresponding to the posture.

[0179] In some embodiments, the data filling module 4551 is used to determine the second cell that is closest to the current atom traversed; determine a cube with the second cell as the center and the first preconfigured length as the radius, wherein the cube includes multiple first cells; and traverse each first cell in the cube.

[0180] In some embodiments, the data filling module 4551 is configured to fill the traversed first cell with data based on a data filling method of a shape complementary dimension to obtain the first cell filled with the shape complementary dimension, wherein the data filling method of the shape complementary dimension includes:

[0181] In response to the current atom being in the complementary determination region and the distance between the first cell and the current atom being less than the second preconfigured length, the first real parameter is configured for the first cell; in response to the distance between the first cell and the current atom being less than the van der Waals radius of the current atom, the first imaginary parameter is configured for the first cell; in response to the distance between the first cell and the current atom being greater than or equal to the van der Waals radius, the second real parameter is configured for the first cell.

[0182] In some embodiments, when the dimension is an ionic bond dimension, the data filling module 4551 is configured to fill the traversed first cell with data based on a data filling method for the ionic bond dimension to obtain a filled first cell of the ionic bond dimension, wherein the data filling method for the ionic bond dimension includes:

[0183] In response to the current atom being a cation and the distance between the first cell and the current atom being less than a third preconfigured length, a third real parameter is configured for the first cell, wherein the third real parameter is a negative real parameter; in response to the current atom being an anion and the distance between the first cell and the current atom being less than the third preconfigured length, a fourth real parameter is configured for the first cell, wherein the fourth real parameter is a positive real parameter.

[0184] In some embodiments, when the dimension is a hydrogen bond dimension, the data filling module 4551 is configured to, in response to the current atom being a hydrogen bond acceptor, fill the first cell traversed with data based on a data filling method of the hydrogen bond dimension to obtain a filled first cell of the hydrogen bond dimension, wherein the data filling method of the hydrogen bond dimension includes:

[0185] In response to the current atom being a hydrogen atom and the distance between the first cell and the current atom being less than the fourth preconfigured length, a fifth real parameter and a second imaginary parameter are configured for the first cell, wherein the fifth real parameter is a positive real parameter and the second imaginary parameter is a positive imaginary parameter; in response to the current atom being not a hydrogen atom and the distance between the first cell and the current atom being less than the fourth preconfigured length, a sixth real parameter and a third imaginary parameter are configured for the first cell, wherein the sixth real parameter is a positive real parameter and the third imaginary parameter is a negative imaginary parameter.

[0186] In some embodiments, the protein docking module 4552 is configured to perform the following processing for each translation conformation corresponding to a pose:

[0187] Based on the translational conformation and the second grid data of multiple dimensions, second docking indices corresponding to the multiple dimensions are determined; and the sum of each second docking indices is used as the first docking indices corresponding to the translational conformation.

[0188] In some embodiments, the protein docking module 4552 is used to determine the overlapping part between the first protein and the second protein in the second grid data based on the translation conformation; and perform the following processing on the second grid data corresponding to each dimension: extract the third grid data corresponding to the overlapping part from the second grid data corresponding to the dimension, and perform three-dimensional convolution processing on the third grid data to obtain the second docking index corresponding to the dimension.

[0189] In some embodiments, the three-dimensional convolution processing is implemented by a convolutional neural network, which includes a three-dimensional convolutional neural network layer.

[0190] In some embodiments, the protein docking module 4552 is used to call the interface of the parallel computing architecture based on the first docking index corresponding to each translational conformation, sort each first docking index, and obtain a sorted list; and the translational conformation corresponding to the largest first docking index in the sorted list is used as the target docking translational conformation.

[0191] In some embodiments, the protein docking module 4552 is used to perform translation processing on the first protein or the second protein in a target posture based on the target docking translation conformation to obtain a protein docking result, wherein the target posture is the posture between the first protein and the second protein corresponding to the target docking translation conformation.

[0192] In some embodiments, the data filling module 4551 is used to convert the protein structure information of the first protein and the second protein respectively to obtain the first amino acid sequence of the first protein and the second amino acid sequence of the second protein before filling the first grid data of the first protein and the second protein from multiple dimensions according to the different postures of the first protein and the second protein to obtain the second grid data of multiple dimensions corresponding to each posture; encode the first amino acid sequence and the second amino acid sequence respectively to obtain the first amino acid feature of the first protein and the second amino acid feature of the second protein; mark the complementary determining region residue index in the second amino acid feature to obtain the third amino acid feature; and construct the first grid data based on the first amino acid feature and the third amino acid feature.

[0193] In some embodiments, the data filling module 4551 is used to determine the coordinate boundaries of the first protein and the second protein in three-dimensional space based on the first amino acid feature and the third amino acid feature; determine the grid space size and cell size of the first grid data based on the coordinate boundaries; and construct the first grid data based on the grid space size and cell size.

[0194] In some embodiments, the data filling module 4551 is used to fix the first protein or the second protein in three-dimensional space and rotate the unfixed protein around each coordinate axis in the three-dimensional space to obtain different postures between the first protein and the second protein before filling the first grid data of the first protein and the second protein from multiple dimensions according to the different postures of the first protein and the second protein to obtain second grid data of multiple dimensions corresponding to each posture, wherein the first protein and the second protein are rigid bodies.

[0195] The present invention provides a computer program product comprising a computer program or computer-executable instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer program or computer-executable instructions from the computer-readable storage medium and executes the computer program or computer-executable instructions, causing the electronic device to perform the protein data processing method described in the present invention.

[0196] The present application provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions or computer programs are stored. When the computer-executable instructions or computer programs are executed by a processor, the processor will execute the protein data processing method provided in the present application, for example, Figure 3A The protein data processing method is shown.

[0197] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface storage, optical disk, or CD-ROM; or various devices including one or any combination of the above memories.

[0198] In some embodiments, computer-executable instructions may be in the form of a program, software, software module, script, or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.

[0199] As an example, computer-executable instructions may, but need not, correspond to a file in a file system, may be stored as part of a file that stores other programs or data, such as, for example, in one or more scripts in a HyperText Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinating files (e.g., files storing one or more modules, subroutines, or code portions).

[0200] By way of example, computer-executable instructions may be deployed to be executed on one electronic device, or on multiple electronic devices located at one site, or on multiple electronic devices distributed across multiple sites and interconnected by a communication network.

[0201] In summary, through the embodiments of the present application, the structural data of different postures of the protein are represented in a grid form, and the calculation accuracy can be improved by calculating the space in which the protein is located in a grid form. The first grid data is assigned values ​​from multiple dimensions for each posture, and the docking indicators corresponding to different translational conformations are determined based on the second grid data of different dimensions, thereby improving the accuracy of determining the docking indicators, and determining the translational conformation used to perform the docking process based on the docking indicators of each translational conformation, thereby improving the accuracy of the protein docking process.

[0202] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent replacements, and improvements made within the spirit and scope of the present application are included in the scope of protection of the present application.

Claims

1. A method for processing protein data, characterized in that: The method comprises: Filling the first grid data of the first protein and the second protein from multiple dimensions according to different postures of the first protein and the second protein to obtain second grid data of the multiple dimensions corresponding to each posture; Determining a first docking index corresponding to each translational conformation based on the second grid data of the multiple dimensions corresponding to each posture and each translational conformation corresponding to the posture, wherein the translational conformation is used to characterize the translational operation corresponding to the first protein or the second protein; determining a target docking translational conformation based on a first docking metric corresponding to each of the translational conformations; A docking process is performed on the first protein and the second protein based on the target docking translation conformation to obtain a protein docking result.

2. The method according to claim 1, characterized in that The step of filling the first grid data of the first protein and the second protein from multiple dimensions according to the different postures of the first protein and the second protein to obtain second grid data of the multiple dimensions corresponding to each posture includes: The following processing is performed on the first protein and the second protein in different postures respectively: traversing each atom of the first protein and the second protein, and taking the traversed atom as the current atom; traverse each first cell associated with the current atom in the first grid data based on each dimension; Filling the traversed first cells with data based on data filling methods of different dimensions to obtain filled first cells of different dimensions; Each of the first grid data after the data filling is used as second grid data of different dimensions to obtain the second grid data of the multiple dimensions corresponding to the posture.

3. The method according to claim 2, characterized in that The traversing each first cell associated with the current atom in the first grid data based on each dimension includes: Determine a second cell that is closest to the traversed current atom; determining a cube with the second cell as the center and a first preconfigured length as the radius, wherein the cube includes a plurality of the first cells; Each of the first cells in the cube is traversed.

4. The method according to claim 2, characterized in that When the dimension is a shape-complementary dimension, the data filling method based on different dimensions is used to fill the traversed first cell with data to obtain the filled first cell of different dimensions, including: The first cell traversed is filled with data based on the data filling method of the shape complementary dimension to obtain the first cell filled with the shape complementary dimension, wherein the data filling method of the shape complementary dimension includes: In response to the current atom being in a complementarity determination region and the distance between the first unit cell and the current atom being less than a second preconfigured length, configuring a first real parameter for the first unit cell; In response to a distance between the first unit cell and the current atom being smaller than a van der Waals radius of the current atom, configuring a first imaginary parameter for the first unit cell; In response to the distance between the first unit cell and the current atom being greater than or equal to the van der Waals radius, a second real parameter is configured for the first unit cell.

5. The method according to claim 2, characterized in that When the dimension is an ionic bond dimension, the data filling method based on different dimensions is used to fill the traversed first cell with data to obtain the filled first cell of different dimensions, including: The first cell traversed is filled with data based on the data filling method of the ionic bond dimension to obtain the first cell filled with the ionic bond dimension, wherein the data filling method of the ionic bond dimension includes: In response to the current atom being a cation and the distance between the first unit cell and the current atom being less than a third preconfigured length, configuring a third real parameter for the first unit cell, wherein the third real parameter is a negative real parameter; In response to the current atom being an anion and the distance between the first unit cell and the current atom being less than a third preconfigured length, a fourth real parameter is configured for the first unit cell, wherein the fourth real parameter is a positive real parameter.

6. The method according to claim 2, characterized in that When the dimension is a hydrogen bond dimension, the data filling method based on different dimensions is used to fill the traversed first cell with data to obtain the filled first cell of different dimensions, including: In response to the current atom being a hydrogen bond acceptor, data is filled in the traversed first cell based on the data filling method of the hydrogen bond dimension to obtain the filled first cell of the hydrogen bond dimension, wherein the data filling method of the hydrogen bond dimension includes: In response to the current atom being a hydrogen atom and the distance between the first unit cell and the current atom being less than a fourth preconfigured length, configuring a fifth real parameter and a second imaginary parameter for the first unit cell, wherein the fifth real parameter is a positive real parameter and the second imaginary parameter is a positive imaginary parameter; In response to the current atom not being a hydrogen atom and the distance between the first cell and the current atom being less than a fourth preconfigured length, a sixth real parameter and a third imaginary parameter are configured for the first cell, wherein the sixth real parameter is a positive real parameter and the third imaginary parameter is a negative imaginary parameter.

7. The method according to claim 1, characterized in that The determining, based on the second grid data of the multiple dimensions corresponding to each posture and each translation conformation corresponding to the posture, a first docking index corresponding to each translation conformation, includes: The following processing is performed for each translation constellation corresponding to the posture: Determining second docking indices corresponding to the multiple dimensions respectively based on the translation conformation and the second grid data of the multiple dimensions; The sum of each of the second docking indices is used as the first docking indices corresponding to the translational conformations.

8. The method according to claim 7, characterized in that Determining second docking indices corresponding to the multiple dimensions respectively based on the translation conformation and the second grid data of the multiple dimensions includes: determining an overlapping portion between the first protein and the second protein in the second grid data based on the translational conformation; The following processing is performed on the second grid data corresponding to each dimension: the third grid data corresponding to the overlapping part is extracted from the second grid data corresponding to the dimension, and the third grid data is subjected to three-dimensional convolution processing to obtain the second docking indicator corresponding to the dimension.

9. The method according to claim 8, characterized in that The three-dimensional convolution processing is implemented by a convolutional neural network, which includes a three-dimensional convolutional neural network layer.

10. The method according to any one of claims 1 to 9, characterized in that The determining of a target docking translational conformation based on the first docking index corresponding to each of the translational conformations includes: Based on the first docking index corresponding to each of the translational conformations, calling the interface of the parallel computing architecture to sort each of the first docking indexes to obtain a sorted list; The translation conformation corresponding to the largest first docking index in the sorted list is used as the target docking translation conformation.

11. The method according to any one of claims 1 to 9, characterized in that The performing docking processing on the first protein and the second protein based on the target docking translation conformation to obtain a protein docking result includes: Based on the target docking translation conformation, a translation process is performed on the first protein or the second protein in a target posture to obtain a protein docking result, wherein the target posture is the posture between the first protein and the second protein corresponding to the target docking translation conformation.

12. The method according to any one of claims 1 to 9, characterized in that Before filling the first grid data of the first protein and the second protein from multiple dimensions according to the different postures of the first protein and the second protein to obtain second grid data of the multiple dimensions corresponding to each posture, the method further includes: performing conversion processing on the protein structure information of the first protein and the second protein respectively to obtain a first amino acid sequence of the first protein and a second amino acid sequence of the second protein; performing encoding processing on the first amino acid sequence and the second amino acid sequence respectively to obtain a first amino acid feature of the first protein and a second amino acid feature of the second protein; marking the complementarity determining region residue index in the second amino acid feature to obtain a third amino acid feature; The first grid data is constructed based on the first amino acid feature and the third amino acid feature.

13. The method according to claim 12, characterized in that The constructing the first grid data based on the first amino acid feature and the third amino acid feature includes: determining a coordinate boundary between the first protein and the second protein in three-dimensional space based on the first amino acid feature and the third amino acid feature; determining a grid space size and a cell size of the first grid data based on the coordinate boundary; The first grid data is constructed based on the grid space size and the cell size.

14. The method according to any one of claims 1 to 9, characterized in that Before filling the first grid data of the first protein and the second protein from multiple dimensions according to the different postures of the first protein and the second protein to obtain second grid data of the multiple dimensions corresponding to each posture, the method further includes: The first protein or the second protein is fixed in a three-dimensional space, and the unfixed protein is rotated around each coordinate axis in the three-dimensional space to obtain different postures between the first protein and the second protein, wherein the first protein and the second protein are rigid bodies.

15. A protein data processing device, characterized in that: The device comprises: a data filling module, configured to fill the first grid data of the first protein and the second protein from multiple dimensions according to different postures of the first protein and the second protein, to obtain second grid data of the multiple dimensions corresponding to each posture; The protein docking module is further configured to determine a first docking index corresponding to each translational conformation based on the second grid data of the multiple dimensions corresponding to each posture and each translational conformation corresponding to the posture, wherein the translational conformation is used to characterize the translation operation corresponding to the first protein or the second protein; The protein docking module is further configured to determine a target docking translational conformation based on a first docking index corresponding to each of the translational conformations; The protein docking module is further configured to perform docking processing on the first protein and the second protein based on the target docking translation conformation to obtain a protein docking result.

16. An electronic device, characterized in that: The electronic device comprises: a memory for storing computer-executable instructions or computer programs; A processor, configured to implement the protein data processing method according to any one of claims 1 to 14 when executing the computer executable instructions or computer program stored in the memory.

17. A computer-readable storage medium storing computer-executable instructions or a computer program, characterized in that: When the computer executable instructions or computer program are executed by a processor, the protein data processing method according to any one of claims 1 to 14 is implemented.

18. A computer program product comprising computer executable instructions or a computer program, characterized in that When the computer executable instructions or computer program are executed by a processor, the protein data processing method according to any one of claims 1 to 14 is implemented.

Images