Molecular docking method, computer readable storage medium and computing device
Through the combination of the semi-flexible molecular docking method and the Isin model, the optimization of calculations using hardware accelerators is solved, and the calculation is improved. The execution efficiency and accuracy are improved. It is suitable for resource-constrained environments.
Patent Information
- Application Number
- CN202410962862.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
Smart Images

Figure CN118918983A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the fields of biomedicine and computers, and in particular to a molecular docking method, a computer-readable storage medium, and a computing device. Background Art
[0002] Molecular docking is an important technology in computer-aided drug design, which aims to predict how small molecules bind to the active sites of biological macromolecular targets and evaluate their binding modes and affinities. It can quickly and high-throughput virtual screening of compound libraries through search algorithms, and predict the best binding conformation using scoring functions, thereby improving the efficiency of new drug discovery. The core of molecular docking technology includes efficient conformational sampling algorithms, such as those based on molecular dynamics, evolutionary algorithms, Monte Carlo methods, etc.; and accurate scoring evaluation functions, which need to balance various molecular interaction energy terms. Molecular docking provides a powerful tool for structural bioinformatics and computer-aided drug design, but its prediction accuracy still needs to be further improved and verified by experimental data. However, after research, the inventors found that the existing molecular docking technology has the problem of large computational complexity.
[0003] To solve this problem, this paper proposes an improved molecular docking scheme to reduce the amount of calculation and improve the execution efficiency of molecular docking. Summary of the invention
[0004] As mentioned in the background art, the technical problem solved by the present application is how to reduce the amount of calculation and improve the execution efficiency of molecular docking.
[0005] To solve the above problems, in the first aspect, the present application provides a molecular docking method, which includes: obtaining receptor information and ligand information; inputting the receptor information and the ligand information into a molecular docking model, and obtaining the docking result of the receptor and the ligand at the output end of the molecular docking model; wherein the molecular docking model is used to make a flexible change to the ligand, and based on the receptor and the ligand after the flexible change, the receptor and the ligand are docked to obtain a semi-flexible molecule, the semi-flexible molecule is scored, and the docking result is obtained according to the score of the semi-flexible molecule.
[0006] Optionally, the molecular docking model is used to perform the following steps: obtain a conformational space, divide the conformational space into grids, and obtain grid data; convert the grid data into a QUBO mathematical model; convert the QUBO mathematical model into an Ising model; calculate the optimal solution of the Ising model, and use the optimal solution as the docking result.
[0007] Optionally, the Ising model is implemented by a field programmable gate array chip or a dedicated integrated circuit.
[0008] Optionally, obtaining the conformational space includes: obtaining the thermodynamic approximate distribution of semi-flexible molecules; sampling the conformations of semi-flexible molecules from the thermodynamic approximate distribution, and taking the conformations of the semi-flexible molecules that meet the conditions as candidate conformations; generating a conformational space based on the candidate conformations; wherein the probability density function of the thermodynamic approximate distribution is: P(x)∝exp(-E(x) / kT); wherein P(x) is the probability of node x, E(x) is the energy of the node x, k is the Boltzmann constant, and T is the system temperature.
[0009] Optionally, the sampling of the conformations of the semi-flexible molecules and taking the conformations of the semi-flexible molecules that meet the conditions as candidate conformations include: step A, generating a current conformation according to a probability distribution; step B, determining whether to take the current conformation as a candidate conformation according to the quality of the current conformation; if the decision result of step B is yes, executing step C to retain the candidate conformation; repeating steps A to C as above until the number of the retained candidate conformations is greater than or equal to a preset value.
[0010] Optionally, the sampling of the conformation of the semi-flexible molecule and taking the conformation of the semi-flexible molecule that meets the conditions as the candidate conformation includes: initializing the conformation and speed of the semi-flexible molecule; calculating the motion trajectory of the semi-flexible molecule according to Newtonian mechanics; recording the conformation of the semi-flexible sub-molecule at certain time intervals; and using the recorded conformation of the semi-flexible sub-molecule as the candidate conformation.
[0011] Optionally, the semi-flexible molecule is scored based on at least one of the following factors: squared mean deviation, ligand-receptor score, energy function, interaction energy, hydrophobicity matching, number of hydrogen bonds.
[0012] Optionally, calculating the optimal solution of the Ising model includes: calculating the optimal solution of the Ising model through simulated annealing or a genetic algorithm.
[0013] In a second aspect, the present application also provides a molecular docking device, which includes: an input information acquisition module for acquiring receptor information and ligand information; a model processing module for inputting the receptor information and the ligand information into a molecular docking model, and obtaining the docking result of the receptor and the ligand at the output end of the molecular docking model; wherein the molecular docking model is used to make a flexible change to the ligand, and based on the receptor and the ligand after the flexible change, dock the receptor and the ligand to obtain a semi-flexible molecule, score the semi-flexible molecule, and obtain the docking result according to the score of the semi-flexible molecule.
[0014] In a third aspect, the present application further provides a computing device, comprising a processor and a storage device, wherein the storage device is suitable for storing a plurality of program codes, and the program codes are suitable for being loaded and run by the processor to execute any one of the molecular docking methods.
[0015] In a fourth aspect, the present application also provides a computer-readable storage medium, in which a plurality of program codes are stored, wherein the program codes are suitable for being loaded and run by a processor to execute any one of the molecular docking methods.
[0016] Compared with the prior art, the technical solution of the embodiment of the present application has the following beneficial effects:
[0017] In the molecular docking method of the embodiment of the present application, a semi-flexible molecular docking scheme is provided. The molecular docking model only makes flexible changes to the ligand information without destroying the molecular formula of the ligand, which can effectively reduce the amount of calculation and ensure the accuracy of the molecular docking results.
[0018] Furthermore, the present application also provides a hardware system specially designed for the Ising solution system for semi-flexible molecular docking, which realizes the efficient operation of the Ising model through spatial discretization and the QUBO mathematical model. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 A schematic diagram of a molecular docking method according to an embodiment of the present application;
[0020] Figure 2 for Figure 1 A flowchart of a specific embodiment of step S102;
[0021] Figure 3 A schematic diagram of a process for sampling the conformation of a semi-flexible molecule according to an embodiment of the present application;
[0022] Figure 4 A schematic diagram of another process for sampling the conformation of a semi-flexible molecule according to an embodiment of the present application;
[0023] Figure 5 A schematic structural diagram of a molecular docking device according to an embodiment of the present application. DETAILED DESCRIPTION
[0024] As mentioned in the background art, existing molecular docking methods have the problem of high computational complexity.
[0025] Specifically, existing molecular docking methods have the following problems:
[0026] Problem 1: Existing hardware accelerators cannot effectively support the semi-flexible molecular docking model. The complexity of solving the semi-flexible molecular docking problem grows exponentially with the scale of the problem. When the scale of the problem exceeds a certain level, even with the most advanced computers, it is difficult to obtain an accurate solution within a limited time. This makes the accurate solution of this problem extremely challenging. Due to the huge amount of computation required to solve the semi-flexible molecular docking problem, powerful hardware acceleration capabilities are required, which still puts a lot of pressure on existing general-purpose CPUs and GPUs. The design of a dedicated hardware accelerator for the semi-flexible molecular docking problem is a major challenge.
[0027] Problem 2: Inefficient computing speed and energy consumption. The process of accurately solving the semi-flexible molecular docking problem is usually accompanied by extremely high energy consumption and heat generation, which brings new challenges to the energy supply and heat dissipation design of the computing system. For some application scenarios with high computing accuracy requirements and limited resources, such as cryptography and traveling salesman, this inefficiency will limit the application scope and performance of the model.
[0028] Problem 3: There is a lack of a model framework for semi-flexible molecular docking. Due to the special properties of the semi-flexible molecular docking problem, it is very difficult to design an efficient and accurate solution algorithm. At present, there is still a lack of a general and efficient solution algorithm framework to solve all semi-flexible molecular docking problems. Due to the difficulty in accurately solving semi-flexible molecular docking problems, people usually resort to various heuristic approximate algorithms, but the solution quality of these algorithms is difficult to guarantee, and improving their accuracy level is a major challenge.
[0029] To solve the above problem, see Figure 1 The present application provides a schematic flow chart of a molecular docking method, characterized in that the method includes steps S101 and S102, which are described in detail as follows.
[0030] Step S101, obtaining receptor information and ligand information.
[0031] In one embodiment, the receptor information includes the spatial coordinate distribution and charge information of each atom in the receptor, and / or the ligand information includes the spatial coordinate distribution and charge information of each atom in the ligand.
[0032] It should be noted that the receptor information and / or ligand information includes but is not limited to the above information, and may also be other information, such as the characteristics and graphs of the receptor and / or ligand, etc. As long as the information can be recognized and processed by the molecular docking model and the molecular docking results of the receptor and the ligand are obtained.
[0033] Step S102: input the receptor information and the ligand information into a molecular docking model, and obtain the docking result of the receptor and the ligand at the output end of the molecular docking model.
[0034] The molecular docking model is used to make flexible changes to the ligand, and based on the receptor and the ligand after the flexible changes, the receptor and the ligand are docked to obtain a semi-flexible molecule, the semi-flexible molecule is scored, and the docking result is obtained according to the score of the semi-flexible molecule.
[0035] Specifically, the ligand information is flexibly changed in the present application. In the semi-flexible molecular docking of the present application, the receptor is rigid, that is, its atomic coordinates will not change during the docking process; while the ligand is flexible and can change without destroying the molecular formula.
[0036] The ligand information after the flexibility change includes multiple groups of ligand information. The receptor is docked with these ligand information respectively to obtain multiple docked semi-flexible molecules. The semi-flexible molecules are scored according to certain rules, and the optimal solution is obtained from the semi-flexible molecules based on the scoring results as the docking result, and the docking result is output at the output end of the molecular docking model.
[0037] The optimal solution may be one or more, which means selecting one or more with the best docking quality from the semi-flexible molecules.
[0038] In one embodiment, multiple semi-flexible molecules may be scored based on multiple factors, and one or more semi-flexible molecules with the highest score may be output as the docking result.
[0039] Optionally, the semi-flexible molecule may be scored based on at least one of the following factors: Root Mean Square Deviation (RMSD), Ligand-receptor score, energy function, interaction energy, hydrophobicity matching, and number of hydrogen bonds.
[0040] Among them, RMSD is a metric used to measure the difference between the structures of two molecules. The smaller the RMSD value, the more similar the two molecular structures are, and the better the docking quality is. The ligand-receptor score is a metric used to measure the strength of the interaction between the ligand and the receptor. The higher the ligand-receptor score value, the stronger the interaction between the ligand and the receptor, and the better the docking quality is. The energy function is used to calculate the total energy of the docking of two molecules. The lower the energy value, the more stable the docking and the better the quality. The interaction energy is used to calculate the energy of the interaction between the two molecules. The larger the interaction energy value, the stronger the interaction and the better the docking quality. The hydrophobic matching is used to evaluate the matching degree of the hydrophobic groups between the ligand and the receptor. The higher the hydrophobic matching degree, the better the docking quality is. The number of hydrogen bonds is used to calculate the number of hydrogen bonds formed between the two molecules. The more hydrogen bonds there are, the stronger the interaction and the better the docking quality is.
[0041] In a specific embodiment, one or more of the following scoring functions may be used to score the semi-flexible molecules:
[0042] Empirical scoring function: The force field-based empirical scoring function used by Vina, AutoDock, DOCK, etc. estimates the intermolecular binding free energy by calculating van der Waals forces, electrostatic forces, hydrogen bond forces, etc.
[0043] Knowledge-based scoring functions: such as DrugScore and PMF, obtain the statistical potential function between atomic pairs by statistically analyzing the structural characteristics of known protein-ligand complexes, and then predict the binding strength of new protein-ligands.
[0044] Machine learning scoring functions: such as RF-Score and NNScore, which learn protein-ligand binding characteristics from known complex data, establish machine learning models, and predict the binding affinity of new complexes.
[0045] Consensus scoring: that is, the comprehensive use of the prediction results of multiple scoring functions, such as equal-weighted summation, weighted summation, etc., can reduce the limitations of a single scoring function and improve prediction accuracy.
[0046] In a preferred embodiment, the Vina scoring function used in the present application is used to score semi-flexible molecules. The advantage of the Vina scoring function is that it has good robustness and can handle different types of proteins and complex and diverse small molecule shapes, thereby achieving better solution results in various situations.
[0047] It should be noted that the scoring function includes but is not limited to the examples given above, as long as the scoring function can meet the accuracy, scoring efficiency and robustness. Among them, accuracy means that the scoring function should be able to accurately reflect the quality of the docking of two molecules. Scoring efficiency means that the calculation efficiency of the scoring function should be high so that a large number of semi-flexible molecules can be quickly evaluated; robustness means that the scoring function should have strong robustness and be able to resist the influence of noise and interference.
[0048] pass Figure 1 The provided molecular docking method, in which the molecular docking model only performs flexible changes on the ligand information on the premise of not destroying the molecular formula of the ligand, can effectively reduce the amount of calculation and ensure the accuracy of the molecular docking results.
[0049] In one embodiment, see Figure 2 , Figure 2 for Figure 1 FIG. 5 is a flow chart of a specific embodiment of step S102. Figure 1 The molecular docking model in step S102 performs the following steps:
[0050] Step S201, acquiring a conformation space, dividing the conformation space into grids, and obtaining grid data.
[0051] Step S202, converting the grid data into a quadratic unconstrained binary optimization (Quadratic Unconstrained Binary Optimization, referred to as QUBO) mathematical model.
[0052] Step S203: convert the QUBO mathematical model into an Ising model.
[0053] Step S204, calculating the optimal solution of the Ising model, and using the optimal solution as the docking result.
[0054] In a specific embodiment, the Ising model is implemented by a Field-Programmable Gate Array (FPGA) chip or an Application Specific Integrated Circuit (ASIC).
[0055] In a specific embodiment, the molecular docking model may include: a spatial gridding unit for S201, a QUBO mathematical model conversion unit for executing step S202, an Ising model conversion unit for executing step S203, and an Ising calculation unit for executing step S204.
[0056] After the receptor information and the ligand information are input into the molecular docking model, the conformational space is obtained according to the spatial coordinates of the semi-flexible molecule after the docking of the receptor and the ligand, and the molecular docking conditions of the receptor and the ligand.
[0057] The spatial gridding unit is used to divide the conformational space into grids. Each grid can be numbered and the state monitored, which converts the protein docking problem into a discretization problem that is easier for computers to handle. For example, if a node placed on a grid has a specific identifier, it means that there are atoms on the grid.
[0058] Optionally, after spatial gridding, the force field data of the protein on a specific atom at each grid point can be obtained (the smaller the force field, the more stable the atom is near the grid point). Proteins have different interaction forces on different types of atoms (e.g., carbon atoms, nitrogen atoms, etc.) in the surrounding space, so the grid data can at least include information such as grid point number, atom type, force field, etc.
[0059] The QUBO mathematical model conversion unit is used to convert grid data into a mathematical model for calculation. It converts discretized data into the format required in the mathematical model. In molecular docking, the QUBO model may be a quadratic unconstrained binary optimization problem, which converts the gridded spatial distribution (i.e., grid data) into a mathematical model, which contains complex biophysical information such as atomic interactions and charge distribution. This mathematical model is the basis for subsequent scoring and searching for the optimal conformation.
[0060] Among them, the Lagrangian formula can be used to convert spatial grid data into a necessary step for the QUBO mathematical model. The Lagrangian formula is a mathematical derivation step that is used to consider the constraints in the problem (such as resource limitations, biophysical constraints, etc.) as part of the objective function and add them to the solution process to ensure that the objective function is optimized under the premise of meeting all constraints. In molecular docking, this can ensure that the molecular conformation with the lowest energy (i.e., the highest affinity) is found under the premise of meeting all constraints (such as the length and angle of chemical bonds, the three-dimensional configuration of molecules, etc.), that is, "the ligand should not destroy the molecular formula." In addition, the Lagrangian formula can be used as part of the scoring function to evaluate the feasibility and quality of each conformation.
[0061] The Hamiltonian of the Ising model (also known as the Ising algorithm) can also be called the potential energy function, which is mainly used to find the optimal solution. In the transformation process of molecular docking, the potential energy function is used to construct a scoring function, which is used to score each molecular conformation obtained by sampling. The lower the value of the potential energy function, the lower the energy of the molecular conformation, that is, the higher the affinity between the molecule and the protein. Therefore, the potential energy function helps determine which molecular conformations are better, thereby selecting the best protein docking results.
[0062] Specifically, the potential energy function may include multiple energy terms, such as electrostatic interactions, van der Waals forces, hydrogen bonds, etc., which are calculated based on the relative positions and charge distributions between atoms in the docked molecule. By adding or combining these energy terms, a total potential energy value is obtained, which reflects the strength of the interaction between the molecular conformation and the receptor. The scoring function uses this potential energy value to score the conformation and guide the subsequent search and optimization process to find the molecular conformation with the lowest energy (i.e., the highest affinity) as the optimal solution.
[0063] Commonly used potential energy functions include:
[0064] Bond stretching energy: reflects the effect of bond length change on molecular energy. Bond stretching energy is usually expressed using Morse potential energy function or Harmony potential energy function.
[0065] Bond angle energy: reflects the effect of bond angle changes on molecular energy. Bond angle energy is usually expressed using the Harmony potential energy function.
[0066] Torsional energy: reflects the effect of torsion angle change on molecular energy. Torsional energy is usually expressed using a periodic potential energy function.
[0067] Non-bonded interaction energy: reflects the influence of non-bonded interactions such as van der Waals forces, electrostatic forces and hydrogen bonds on molecular energy. Non-bonded interaction energy is usually expressed by Lennard-Jones potential energy function or Coulomb potential energy function.
[0068] Choosing a suitable potential energy function is one of the key steps in semi-flexible molecular docking. When choosing a potential energy function, the following factors should be considered:
[0069] Precision: The potential energy function should have sufficient precision to accurately reflect the conformations and interactions of semiflexible molecules.
[0070] Efficiency: The computational efficiency of the potential energy function should be high so that the quality of candidate conformations can be quickly evaluated in semiflexible molecular docking.
[0071] Versatility: The potential energy function should be highly versatile and applicable to various types of semi-flexible molecular docking problems.
[0072] The parameterization of potential energy function is a key step to obtain accurate potential energy function. Parameterization is usually carried out using experimental data, quantum chemical calculations or molecular mechanics simulations.
[0073] Through this embodiment, a hardware system specially designed for the Ising solution system of semi-flexible molecular docking is provided, and the efficient operation of the Ising algorithm is realized through spatial discretization and QUBO mathematical model. The Ising model of semi-flexible molecular docking is directly run in the hardware system, and the computing efficiency and memory utilization are improved through the parallel calculation and storage of Ising algorithm model parameters. By designing an efficient spatial gridding module and a QUBO mathematical model conversion module, the calculation and storage operations of semi-flexible molecular docking are optimized to speed up the Ising algorithm solution process and reduce energy consumption. A complete overlap of calculation and data acquisition is achieved, and the parallel computing architecture is used to realize the efficient collaboration of various functional modules, further improving the overall performance and energy efficiency.
[0074] In one embodiment, Figure 2 Step S201 of obtaining the conformational space may include: obtaining the thermodynamic approximate distribution of the semi-flexible molecules; sampling the conformations of the semi-flexible molecules from the thermodynamic approximate distribution, and taking the conformations of the semi-flexible molecules that meet the conditions as candidate conformations; and generating the conformational space according to the candidate conformations.
[0075] That is, the conformational space is the space where candidate conformations are located.
[0076] Among them, the thermodynamic approximate distribution is a probability distribution based on statistical mechanics, which can be used to simulate the conformational space of semi-flexible molecules, with the aim of generating a large number of semi-flexible molecular conformations and providing candidate conformations for subsequent scoring and screening. The probability density function of the thermodynamic approximate distribution can be expressed by the following formula: P(x)∝exp(-E(x) / kT); where P(x) is the probability of state x; E(x) is the energy of state x; k is the Boltzmann constant; and T is the system temperature.
[0077] In the thermodynamic distribution, the force fields of different atoms at different grid points (which can be understood as whether an atom is stable at a certain position in space) can be filled in the bias vector of the Ising model, and the relative position information between two atoms can be filled in the weight matrix of the Ising model. Thus, the probability distribution of semi-flexible molecules can be obtained through the thermodynamic distribution.
[0078] In one embodiment, see Figure 3 , Figure 3 A schematic diagram of a process for sampling conformations of a semi-flexible molecule; the sampling of the conformations of the semi-flexible molecule and taking the conformations of the semi-flexible molecule that meet the conditions as candidate conformations may include:
[0079] Step A, generating the current conformation according to the probability distribution.
[0080] Step B, determining whether to use the current conformation as a candidate conformation according to the quality of the current conformation;
[0081] If the decision result of step B is yes, then executing step C to retain the candidate conformation;
[0082] Repeat steps A to C above until the number of the candidate conformations retained is greater than or equal to a preset value.
[0083] Specifically, in semi-flexible molecular docking, a current conformation of a semi-flexible molecule is generated according to the probability distribution, and the energy of the current conformation is calculated. A random number is generated, and if the random number is less than the specified value, the current conformation is accepted as a candidate conformation and retained; otherwise, the current conformation is rejected. The above steps are repeated until the required number of candidate conformations is generated.
[0084] Thus, a method for sampling from probability distribution is provided to sample sufficient candidate conformations of semi-flexible molecules from the thermodynamic approximation distribution.
[0085] In another embodiment, see Figure 4 , Figure 4This is another schematic diagram of a process for sampling conformations of semi-flexible molecules. The sampling of conformations of the semi-flexible molecules and taking conformations of the semi-flexible molecules that meet the conditions as candidate conformations may also include:
[0086] Step S401, initializing the conformation and velocity of the semi-flexible molecule;
[0087] Step S402, calculating the motion trajectory of the semi-flexible molecule according to Newtonian mechanics;
[0088] Step S403, recording the conformation of the semi-flexible sub-element within a certain time interval;
[0089] Step S404: using the recorded conformation of the semi-flexible sub-element as the candidate conformation.
[0090] Thus, a method for sampling from the thermodynamic approximate distribution by molecular dynamics simulation is provided to obtain candidate conformations of semi-flexible molecules. Molecular dynamics simulation is a time evolution method based on Newtonian mechanics, which can be used to simulate the conformational space of semi-flexible molecules. In molecular dynamics simulation, the state of the system changes over time and eventually reaches an equilibrium state. In the equilibrium state, the conformational distribution of the system follows the thermodynamic approximate distribution.
[0091] It should be noted that in addition to the above sampling methods, other methods of sampling from thermodynamic approximate distributions can also be used to obtain candidate conformations, such as the MH (Metropololis-Hasting) sampling method or the Gibbs sampling method. The MH sampling method is an improved version of the acceptance-rejection sampling method, which can improve the sampling efficiency. Gibbs sampling is a method for sampling from multidimensional probability distributions, which can be used to sample complex conformational spaces.
[0092] Optionally, the above sampling method and molecular docking process can be run on a GPU (graphics processing unit). With its powerful parallel computing capabilities, the GPU provides an efficient hardware acceleration solution for semi-flexible molecular docking. Compared with the traditional CPU (central processing unit), the GPU has thousands of parallel computing cores and can handle a large number of docking tasks at the same time, thereby significantly improving the computing efficiency. The sampling method and molecular docking are implemented on the GPU, and a variety of optimization strategies such as batch molecular parallel docking, full GPU computing, and single-precision floating-point operations are introduced to maximize the parallel performance of the GPU.
[0093] It should be noted that the Ising algorithm is a heuristic optimization algorithm that simulates the cooling process of solids. It is often used to solve combinatorial optimization problems that are difficult for non-deterministic polynomials (NP). It equates the optimization problem to solving the lowest energy state of solids. By controlling the cooling process and the probability acceptance strategy, it gradually evolves from a disordered state at a higher temperature to an optimal solution state at a lower temperature. The algorithm starts with a higher initial temperature and a random solution, generates new solutions through small perturbations, and accepts new solutions that are better or worse than the objective function according to a certain probability criterion, while continuously cooling down. In this way, by simulating atomic thermal motion, it eventually converges to an approximate global optimal solution. The Ising algorithm has a strong global search capability, but a low convergence efficiency. It is often used in combination with other algorithms to improve the solution efficiency and is widely used in combinatorial optimization problems in engineering practice.
[0094] In order to solve the problem of low convergence efficiency of the Ising algorithm, in order to solve the problem of low convergence efficiency of the Ising algorithm, in one embodiment, the calculating the optimal solution of the Ising model includes: calculating the optimal solution of the Ising model through simulated annealing or genetic algorithms (GA).
[0095] Among them, simulated annealing is an optimization method that simulates the annealing process of a physical system. In simulated annealing, the scores of candidate conformation scores are used as the energy of the system, and an annealing algorithm is used to reduce the energy of the system.
[0096] Genetic algorithm is an optimization method that simulates the biological evolution process. In the genetic algorithm, candidate docking poses are represented as chromosomes, and genetic operations are used to improve the quality of candidate docking poses.
[0097] Finding the optimal molecular docking posture through simulated annealing or genetic algorithm can better achieve the optimization and convergence of semi-flexible molecular docking.
[0098] Optionally, one or more of the following convergence criteria may be set: maximum number of iterations, scoring function change threshold, structure RMSD threshold, etc. If one or more convergence criteria are met, sampling is terminated and the molecular docking method is terminated.
[0099] The convergence criterion of the maximum number of iterations may be: if the maximum number of iterations is reached, the sampling is terminated.
[0100] The convergence criterion of the scoring function change threshold may be: if the change of the scoring function in several consecutive iterations is less than a certain threshold, the sampling is terminated.
[0101] The convergence criterion of the structural RMSD threshold may be: if the change in the structural RMSD of the docking pose in several consecutive iterations is less than a certain threshold, the search is terminated.
[0102] Through the above-mentioned convergence algorithm, the problem of low convergence efficiency of the Ising algorithm can be effectively solved. This optimization significantly improves the calculation speed and reduces energy consumption, so that the model can achieve efficient calculation even in a resource-constrained environment, solving the problems of slow calculation speed and low energy efficiency of existing semi-flexible molecular docking.
[0103] This application is specifically designed for the Ising algorithm of semi-flexible molecular docking, taking into account its unique computing and storage requirements. By optimizing the hardware architecture and computing process, efficient support for the Ising algorithm is achieved. This specially designed semi-flexible molecular docking model can give full play to the performance of the Ising algorithm on the hardware platform, so that the model can also be fully utilized in a resource-constrained environment, solving the problem that the existing solution model cannot quickly and efficiently solve the Ising algorithm.
[0104] In one embodiment, the application also provides a dedicated hardware accelerator for solving the Ising algorithm for semi-flexible molecular docking, including: an FPGA chip, as the core processor of the accelerator, responsible for executing the Ising model; an external memory, used to store the parameters and input data of the molecular docking model; a controller, used to manage the input and output of data, and schedule the operation process of the accelerator; a PCIe interface, the device exchanges data with other components through the PCIe bus interface.
[0105] Therefore, this paper meets the needs of the Ising algorithm to run efficiently on resource-constrained hardware platforms. By implementing the hardware accelerator as an application-specific integrated circuit (ASIC) core or a field-programmable gate array (FPGA), a flexible and high-performance hardware acceleration solution is provided to meet the needs of different application scenarios.
[0106] In one embodiment, see Figure 5 The present application also provides a molecular docking device 50, which may include: an input information acquisition module 501, used to acquire receptor information and ligand information; a model processing module 502, used to input the receptor information and the ligand information into a molecular docking model, and obtain the docking result of the receptor and the ligand at the output end of the molecular docking model;
[0107] The molecular docking model is used to make flexible changes to the ligand, and based on the receptor and the ligand after the flexible changes, the receptor and the ligand are docked to obtain a semi-flexible molecule, the semi-flexible molecule is scored, and the docking result is obtained according to the score of the semi-flexible molecule.
[0108] In one embodiment, the model processing module 502 may include:
[0109] A spatial grid unit, used for obtaining a conformational space, dividing the conformational space into grids, and obtaining grid data;
[0110] A QUBO mathematical model conversion unit, used for converting the grid data into a QUBO mathematical model;
[0111] An Ising model conversion unit, used for converting the QUBO mathematical model into an Ising model;
[0112] An Ising calculation unit is used to calculate the optimal solution of the Ising model and use the optimal solution as the docking result.
[0113] In one embodiment, the spatial grid unit may include:
[0114] Thermodynamic approximation subunit, used to obtain the thermodynamic approximation distribution of semi-flexible molecules;
[0115] A sampling subunit, used for sampling conformations of semi-flexible molecules from the thermodynamic approximate distribution, and taking conformations of the semi-flexible molecules that meet the conditions as candidate conformations;
[0116] A scoring subunit, used for generating a conformational space according to the candidate conformations;
[0117] Wherein, the probability density function of the thermodynamic approximate distribution is:
[0118] P(x)∝exp(-E(x) / kT);
[0119] Where P(x) is the probability of state x, E(x) is the energy of state x, k is the Boltzmann constant, and T is the system temperature.
[0120] In one embodiment, the sampling subunit can perform the following steps: step A, generating a current conformation according to a probability distribution; step B, determining whether to use the current conformation as a candidate conformation according to the quality of the current conformation; if the decision result of step B is yes, executing step C to retain the candidate conformation; repeating steps A to C as above until the number of retained candidate conformations is greater than or equal to a preset value.
[0121] In one embodiment, the sampling subunit can perform the following steps: initialize the conformation and speed of the semi-flexible molecule; calculate the motion trajectory of the semi-flexible molecule according to Newtonian mechanics; record the conformation of the semi-flexible sub-molecule at a certain time interval; and use the recorded conformation of the semi-flexible sub-molecule as the candidate conformation.
[0122] Optionally, the semi-flexible molecule is scored based on at least one of the following factors: squared mean deviation, ligand-receptor score, energy function, interaction energy, hydrophobicity matching, number of hydrogen bonds.
[0123] In one embodiment, the Ising calculation unit is further used to calculate the optimal solution of the Ising model through simulated annealing or genetic algorithm.
[0124] For more information about the working principle and working method of the molecular docking device 50, please refer to Figures 1 to 4 The description of molecular docking method in will not be repeated here.
[0125] The embodiment of the present application also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the molecular docking method of the present application are executed. The computer-readable storage medium may include a non-volatile memory or a non-transitory memory, and may also include an optical disk, a mechanical hard disk, a solid-state hard disk, etc.
[0126] An embodiment of the present application also provides a computing device, including a memory and a processor, wherein the memory stores a computer program that can be executed on the processor, and when the processor runs the computer program, the steps of the molecular docking method of the present application are executed.
[0127] In the embodiments of the present application, the processor may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, and may implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application may be directly embodied as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.
[0128] In the embodiment of the present application, the memory may be a non-volatile memory, such as a hard disk drive (HDD) or a solid-state drive (SSD), etc., or a volatile memory (volatile memory), such as a random-access memory (RAM). The memory is any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory in the embodiment of the present application may also be a circuit or any other device that can realize a storage function, for storing computer programs and / or data.
[0129] In the molecular docking method provided in the embodiment of the present application, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program is loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user device or other programmable device. The computer program can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instruction can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, referred to as DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media integrations. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital video disc (DVD)), or a semiconductor medium (eg, an SSD).
[0130] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
[0131] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.
[0132] In the above embodiments, the description of each embodiment has its own emphasis, and any multiple embodiments can be used in combination. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0133] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in a processor or an instruction in the form of software. The steps of the method disclosed in conjunction with the embodiment of the present application can be directly embodied as a hardware processor for execution, or a combination of hardware and software units in a processor for execution. The software unit can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor executes the instructions in the memory, and completes the steps of the above method in conjunction with its hardware. To avoid repetition, it is not described in detail here.
[0134] In the embodiment of the present application, the processor of the above-mentioned device may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.
[0135] The present application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute some or all of the steps of any method described in the above method embodiment. The computer program product may be a software installation package.
[0136] In the several embodiments provided in the present application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of the above-mentioned units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.
[0137] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present application.
[0138] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0139] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or TRP, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.
[0140] It should be understood that the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article indicates that the associated objects before and after are in an "or" relationship.
[0141] The "plurality" appearing in the embodiments of the present application refers to two or more.
[0142] The first, second, etc. descriptions appearing in the embodiments of the present application are only used for illustration and distinction of the description objects. There is no order, nor do they indicate any special limitation on the number of devices in the embodiments of the present application, and cannot constitute any limitation on the embodiments of the present application.
[0143] The "connection" that appears in the embodiments of the present application refers to various connection methods such as direct connection or indirect connection to achieve communication between devices, and the embodiments of the present application do not impose any limitations on this.
[0144] The embodiments of the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of the present application. At the same time, for general technical personnel in this field, according to the idea of the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.
Claims
1. A molecular docking method, characterized in that: The method comprises: Obtain receptor information and ligand information; Inputting the information of the receptor and the information of the ligand into a molecular docking model, and obtaining the docking result of the receptor and the ligand at the output end of the molecular docking model; The molecular docking model is used to make flexible changes to the ligand, and based on the receptor and the ligand after the flexible changes, the receptor and the ligand are docked to obtain a semi-flexible molecule, the semi-flexible molecule is scored, and the docking result is obtained according to the score of the semi-flexible molecule.
2. The method according to claim 1, characterized in that The molecular docking model was used to perform the following steps: Acquiring a conformational space, dividing the conformational space into grids, and obtaining grid data; Converting the grid data into a QUBO mathematical model; Converting the QUBO mathematical model into an Ising model; An optimal solution of the Ising model is calculated, and the optimal solution is used as the docking result.
3. The method according to claim 2, characterized in that The Ising model is implemented by a field programmable gate array chip or a dedicated integrated circuit.
4. The method according to claim 2 or 3, characterized in that: The obtaining of conformational space comprises: obtaining the thermodynamic approximate distribution of semi-flexible molecules; Sampling conformations of semi-flexible molecules from the thermodynamic approximate distribution, and taking conformations of the semi-flexible molecules that meet the conditions as candidate conformations; generating a conformational space according to the candidate conformations; Wherein, the probability density function of the thermodynamic approximate distribution is: P(x)∝exp(-E(x) / kT); Where P(x) is the probability of state x, E(x) is the energy of state x, k is the Boltzmann constant, and T is the system temperature.
5. The method according to claim 4, characterized in that The sampling of the conformations of the semi-flexible molecule and taking the conformations of the semi-flexible molecule that meet the conditions as candidate conformations comprises: Step A, generating the current conformation according to the probability distribution; Step B, determining whether to use the current conformation as a candidate conformation according to the quality of the current conformation; if the determination result of step B is yes, executing step C to retain the candidate conformation; Repeat steps A to C above until the number of the candidate conformations retained is greater than or equal to a preset value.
6. The method according to claim 4, characterized in that The sampling of the conformations of the semi-flexible molecule and taking the conformations of the semi-flexible molecule that meet the conditions as candidate conformations comprises: initializing the conformation and velocity of the semi-flexible molecule; Calculating the motion trajectory of the semi-flexible molecule according to Newtonian mechanics; recording the conformation of the semiflexible subson at certain time intervals; The recorded conformation of the semiflexible subson is used as the candidate conformation.
7. The method according to claim 1, characterized in that The semi-flexible molecules are scored based on at least one of the following factors: squared mean deviation, ligand-receptor score, energy function, interaction energy, hydrophobicity matching, number of hydrogen bonds.
8. The method according to claim 4, characterized in that The calculating the optimal solution of the Ising model comprises: The optimal solution of the Ising model is calculated by simulated annealing or genetic algorithm.
9. A computing device comprising a processor and a storage device, wherein the storage device is suitable for storing a plurality of program codes, characterized in that: The program code is adapted to be loaded and executed by the processor to perform the method of any one of claims 1 to 8.
10. A computer-readable storage medium storing a plurality of program codes, characterized in that: The program code is suitable for being loaded and executed by a processor to perform the method of any one of claims 1 to 8.
Citation Information
Patent Citations
Sensitivity analysis algorithm for computer-aided pilot drug optimization design
CN111415703A
Data processing method and device, storage medium and electronic device
CN116468211A
Molecular docking method and device based on light quantum computer
CN116597917A
Molecular docking method and apparatus based on coherent ising machine
WO2023185658A1
Cited By
Molecular docking method, system and medium based on hybrid quantum algorithm
CN122781417A
