Protein-protein docking method, apparatus, intelligent device, and storage medium

Abstract

The application belongs to the technical field of bioinformatics, and provides a protein-protein docking method and device, an intelligent device and a storage medium. The method comprises the following steps: performing global search on the surface space of a receptor protein to determine a plurality of candidate docking poses of the receptor protein and a ligand protein; performing optimization on the plurality of candidate docking poses according to a preset optimization algorithm to determine a target docking pose; clustering the target docking pose; and determining an optimal docking pose of the receptor protein and the ligand protein based on the clustering result. The application can efficiently search for a binding conformation and improve the efficiency and success rate of protein-protein docking.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bioinformatics, and in particular, to a protein-protein docking method and device, an intelligent device, and a storage medium. BACKGROUND

[0002] Protein-protein interactions play an important role in biological processes such as cell signaling, metabolic regulation, gene expression regulation, and immune response. In addition to obtaining protein complex structures through traditional experimental methods (such as X-ray crystallography and nuclear magnetic resonance), protein-protein docking technology can also be used. Protein-protein docking is a key research method in computational biology, aiming to explore the interactions between proteins and their complex structures. With the continuous development of technology, it has become possible to use computers to simulate the docking of proteins and predict the structure of protein complexes.

[0003] However, protein-protein docking is a challenging task, mainly due to the extremely large conformational space of proteins, which requires traversing a vast space to search for possible binding conformations. Therefore, how to efficiently search for binding conformations and improve the efficiency and success rate of protein-protein docking is a problem that needs to be solved. SUMMARY

[0004] The embodiments of the present application provide a protein-protein docking method, device, intelligent device, and storage medium, which can efficiently search for binding conformations and thus improve the efficiency and success rate of protein-protein docking.

[0005] In a first aspect, the embodiments of the present application provide a protein-protein docking method, which includes:

[0006] globally searching the surface space of a receptor protein to determine a plurality of candidate docking poses of the receptor protein and a ligand protein;

[0007] optimizing the plurality of candidate docking poses according to a preset optimization algorithm to determine a target docking pose;

[0008] clustering the target docking pose;

[0009] determining an optimal docking pose of the receptor protein and the ligand protein based on the clustering result.

[0010] In a possible implementation manner of the first aspect, the step of globally searching the surface space of a receptor protein to determine a plurality of candidate docking poses of the receptor protein and a ligand protein includes:

[0011] The FFT algorithm is used to perform a global search of the surface space of the receptor protein to generate multiple binding conformations;

[0012] The docking configurations are screened according to preset screening criteria to determine several candidate docking postures.

[0013] In one possible implementation of the first aspect, the step of optimizing the plurality of candidate docking postures using the preset optimization algorithm to determine the target docking posture includes:

[0014] Based on the aforementioned candidate docking postures, the surface space of the receptor protein is divided into regions.

[0015] According to the preset optimization algorithm, a local search is performed on the candidate docking postures in each divided region to determine the target docking posture.

[0016] In one possible implementation of the first aspect, the candidate docking postures include ligand protein conformations, and the step of dividing the surface space of the receptor protein into regions based on the plurality of candidate docking postures includes:

[0017] Based on the position of the ligand protein conformation, the candidate docking postures are clustered to obtain several classes of candidate docking postures.

[0018] Based on the number of candidate docking postures, the surface space of the receptor protein is divided into regions, wherein each region contains one of the candidate docking postures.

[0019] In one possible implementation of the first aspect, the step of performing a local search on the candidate docking postures in each divided region according to the preset optimization algorithm to determine the target docking posture includes:

[0020] The differential evolution algorithm is used to iteratively optimize the candidate docking postures in each region.

[0021] Based on the results of the iterative optimization, the target docking posture in each region is determined.

[0022] In one possible implementation of the first aspect, the step of iteratively optimizing the candidate docking postures in each region using a differential evolution algorithm includes:

[0023] It is beneficial to use a preset evaluation function to evaluate the fitness of candidate docking attitudes in the region and determine the fitness value of candidate docking attitudes in the region;

[0024] The candidate docking postures are iteratively optimized based on the fitness value and a preset iteration strategy until a preset termination condition is met, at which point the iteration optimization ends. The preset iteration strategy includes crossover, mutation, and selection.

[0025] In one possible implementation of the first aspect, the step of clustering the target docking posture includes:

[0026] The target docking postures are merged and clustered using a hierarchical clustering algorithm to obtain several clusters of target docking postures.

[0027] Secondly, embodiments of this application provide a protein-protein docking device, the device comprising:

[0028] The first determining unit is used to perform a global search of the surface space of the receptor protein to determine several candidate docking postures of the receptor protein and the ligand protein.

[0029] The second determining unit is used to optimize the plurality of candidate docking postures according to a preset optimization algorithm to determine the target docking posture;

[0030] Clustering analysis unit, used to cluster the target docking posture;

[0031] The docking attitude determination unit is used to determine the optimal docking attitude between the receptor protein and the ligand protein based on the clustering results.

[0032] Thirdly, embodiments of this application provide an intelligent device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the protein-protein docking method as described in the first aspect above.

[0033] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the protein-protein docking method as described in the first aspect above.

[0034] Fifthly, embodiments of this application provide a computer program product that, when run on a smart device, causes the smart device to execute the protein-protein docking method as described in the first aspect above.

[0035] In this embodiment, a global search of the surface space of the receptor protein is performed to determine several candidate docking postures of the receptor protein and the ligand protein, i.e., several potential binding sites. Then, the candidate docking postures are optimized according to a preset optimization algorithm. Based on the results of the global search, potential binding sites are further optimized and screened to determine the target docking posture, which can effectively improve the search efficiency. Then, the target docking postures are clustered, and based on the clustering results, the optimal docking posture of the receptor protein and the ligand protein is determined, which can effectively improve the success rate of protein-protein docking. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart illustrating the implementation of the protein-protein docking method provided in the embodiments of this application;

[0038] Figure 2 This is a flowchart illustrating the specific implementation of step S101 of the protein-protein docking method provided in this application embodiment;

[0039] Figure 3 This is a schematic diagram of candidate docking postures obtained through global search in the protein-protein docking method provided in the embodiments of this application;

[0040] Figure 4 This is a flowchart illustrating a specific implementation of step S102 in the protein-protein docking method provided in this application embodiment;

[0041] Figure 5 This is a flowchart illustrating a specific implementation of the protein-protein docking method provided in this application, which iteratively optimizes the candidate docking postures in each region.

[0042] Figure 6 This is a schematic diagram illustrating the prediction of complex structure in the protein-protein docking method provided in the embodiments of this application;

[0043] Figure 7 This is a structural block diagram of the protein-protein docking device provided in the embodiments of this application;

[0044] Figure 8 This is a schematic diagram of the smart device provided in the embodiments of this application. Detailed Implementation

[0045] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0046] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0047] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0048] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0049] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0050] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0051] It should be understood that the protein-protein docking method provided in this application is applicable to various types of smart devices that need to simulate protein-protein docking, specifically including smart devices such as mobile phones, tablets, wearable devices, laptops, and desktop computers. This application does not impose any limitations on the specific type of smart device.

[0052] Figure 1 The implementation flow of the protein-protein docking method provided in this application embodiment is illustrated. The method flow includes steps S101 to S104. The specific implementation principle of each step is as follows:

[0053] Step S101: Perform a global search on the surface space of the receptor protein to determine several candidate docking postures of the receptor protein and the ligand protein.

[0054] Before protein-protein docking, the stationary and mobile proteins are identified. The stationary protein is labeled as the receptor protein, while the mobile protein is labeled as the ligand protein. The docking attitude refers to the relative position and orientation of the receptor and ligand proteins when they bind to each other. The ligand protein moves within the surface space of the receptor protein to search for the optimal docking attitude.

[0055] In protein-protein docking, the conformational space of proteins is typically enormous, making it difficult to achieve global coverage by directly sampling the conformational space using evolutionary algorithms. In this implementation, a global search is first performed on the surface space of the receptor protein to identify potential binding sites.

[0056] As one possible implementation of this application Figure 2 The specific implementation flow of step S101 in the protein-protein docking method provided in the embodiments of this application is shown below:

[0057] A1: The FFT algorithm is used to perform a global search of the surface space of the receptor protein to generate multiple binding conformations.

[0058] Fast Fourier Transform (FFT) is an efficient algorithm used to compute the Discrete Fourier Transform (DFT) and its inverse transform. FFT-based search methods can quickly search for potential binding conformations globally (up to 10^10), avoiding getting trapped in local optima and resulting in a more comprehensive selection of potential binding sites, thus improving docking accuracy. In the FFT-based global search algorithm, the receptor and ligand proteins are treated as rigid bodies with six degrees of freedom, specifically three-dimensional translation and three-dimensional rotation. In this embodiment, three-dimensional translation refers to the movement of all atomic coordinates of the ligand protein, and three-dimensional rotation refers to the rotation of the entire ligand protein.

[0059] In this embodiment, the receptor protein molecule is kept fixed and projected onto a three-dimensional grid of size N×N×N. The grid is discretized based on the receptor protein molecule's geometry, electrostatic properties, and hydrophobicity. The ligand protein molecule rotates around the receptor protein molecule. After each rotation, the ligand protein molecule is reprojected onto the three-dimensional grid, and its geometry, electrostatic properties, and hydrophobicity are also discretized. The resulting discretized data is a three-dimensional grid of both the receptor and ligand proteins, including information on their geometry, electrostatic properties, and hydrophobicity. Specifically, each discretized grid point represents a molecule's properties at that point, such as whether it is occupied by atoms, the magnitude of its electrostatic potential, and its hydrophobicity. This discretized data is used for subsequent Fast Fourier Transform (FFT) calculations to efficiently evaluate different docking postures in a global search. Grid discretization simplifies the calculation of complex molecular geometries and their interactions, making it easier for intelligent devices to process.

[0060] A2: Based on preset screening conditions, the combined conformations are screened to determine several candidate docking postures.

[0061] In some implementations, conformations can be initially screened based on conditions such as geometric matching and surface contact, retaining a predetermined number of docking conformations. The retained docking conformations are then transformed into Cartesian coordinate space, and site constraint information and a fast lookup table-based atom pair scoring function are applied to score them. The retained docking conformations are then ranked using a precise distance-related atom pair scoring function, and candidate docking attitudes are determined based on the ranking results.

[0062] In one possible implementation, inappropriate binding conformations are filtered out based on the interaction energy between proteins to determine candidate docking postures; each binding conformation corresponds to a candidate docking posture. The lower the interaction energy, the more stable the binding conformation. The retained binding conformations are then ranked based on their interaction energy, and a certain number of conformations are retained, for example, the top 1000, to obtain a number of candidate docking postures.

[0063] The interaction energy between proteins is represented by a comprehensive correlation function, and the calculation of the correlation function is accelerated by FFT. The correlation function is expressed as follows: (1)

[0064] S(t)=∑ v∈V R(v)L(v+o)=IFFT[FFT[R(v)]*FFT[L(v)]] (1)

[0065] Where R represents the discrete function of the receptor protein, L represents the discrete function of the ligand protein, v represents the coordinates in the three-dimensional grid, o represents the translation of the ligand protein relative to the receptor protein, V represents the entire three-dimensional grid, * represents the complex conjugate operator, and IFFT represents the inverse FFT.

[0066] The complex conjugate operator refers to taking the conjugate of a complex number, that is, having the same real part and the inverted imaginary part. In the above equation (1), R and L are respectively subjected to FFT transformation to obtain their frequency domain representations. Then, the frequency domain signal of R is subjected to conjugate operation to obtain the conjugate frequency domain signal. By taking the conjugate of a signal when multiplying in the frequency domain, we can efficiently calculate the correlation in the time domain in the frequency domain. Finally, the correlation in the time domain, that is, S(t), can be obtained by IFFT transformation.

[0067] Directly calculating the correlation function requires considering all possible relative displacements, totaling N. 3 There are several scenarios. Each relative displacement requires N... 3 The computation involves multiple multiplication and addition operations, therefore the total computational complexity is O(N). 6 However, by using the FFT transformation, the correlation function S(t) can be obtained by simply multiplying the two functions and then performing the inverse FFT, thus reducing the computational cost to O(N). 3 ln N 3 This enables efficient searching.

[0068] In this embodiment, the surface space of the receptor protein is globally searched based on the FFT algorithm to obtain a large number of binding conformations. Unreasonable binding conformations are filtered out by the interaction ability between proteins, and finally a certain number of binding conformations are retained for subsequent optimization steps.

[0069] Step S102: Optimize the candidate docking postures according to the preset optimization algorithm to determine the target docking posture.

[0070] The target docking posture is the optimized docking posture of the candidate docking posture.

[0071] In this embodiment, based on the global search in step S101, a set of candidate docking postures can be obtained. For example... Figure 3 As shown in the figure, region ① represents the receptor protein, while the particles around it represent the centroid of the ligand protein, which can be regarded as a docking posture.

[0072] like Figure 3 As shown, binding sites are mainly distributed in several regions on the receptor surface, as well as some scattered points. Therefore, to improve the efficiency of subsequent local optimization operations, the surface space of the receptor protein can be divided into regions.

[0073] As one possible implementation of this application Figure 4 A specific implementation flow of step S102 in the protein-protein docking method provided in this application embodiment is shown below:

[0074] B1: Based on the aforementioned candidate docking postures, the surface space of the receptor protein is divided into regions.

[0075] In one possible implementation, the candidate docking postures include ligand protein conformations. Based on the position of the ligand protein conformations, the candidate docking postures are clustered to obtain several classes of candidate docking postures. The surface space of the receptor protein is divided into regions according to the number of classes of the candidate docking postures, wherein each region contains one class of the candidate docking postures.

[0076] The above clustering can employ the first clustering algorithm. The position of the candidate docking posture specifically refers to the position of the ligand protein within that posture. The first clustering algorithm is a partitioning-based algorithm; for example, the K-means algorithm can be used to cluster the positions of the candidate docking postures into 10 clusters. That is, the positions of the ligand proteins in three-dimensional space are clustered, using the centroid of the ligand protein to represent its position in three-dimensional space. Based on the clustering results, the surface space of the receptor protein is divided into regions; for example, clustering into 10 clusters divides the surface space of the receptor protein into 10 regions. The specific process of using the K-means algorithm is not detailed here.

[0077] B2: According to the preset optimization algorithm, a local search is performed on the candidate docking postures in each divided region to determine the target docking posture. The target docking posture is the optimal docking posture in the region.

[0078] In this embodiment, the surface space of the receptor protein is divided into regions based on the position of the ligand protein conformation in the candidate docking postures. Then, according to a preset optimization algorithm, a local search is performed on the candidate docking postures in each region after division. The combination of global search and local optimization can ensure the comprehensiveness of the search, improve the efficiency of the search, and improve the accuracy of protein-protein docking prediction.

[0079] In one possible implementation, the preset optimization algorithm includes a Differential Evolution Algorithm (DE). The DE algorithm iteratively optimizes the candidate docking postures in each region, and determines the optimal docking posture in each region based on the results of the iterative optimization.

[0080] As one possible implementation of this application Figure 5This paper illustrates a specific implementation flow of the protein-protein docking method provided in this application, which iteratively optimizes candidate docking postures in each region using a differential evolution algorithm. The details are as follows:

[0081] C1: A pre-defined evaluation function is used to evaluate the fitness of candidate docking postures within a region, determining their fitness values. The pre-defined evaluation function can be the DFIRE function. This function calculates the fitness value for each individual (the docking conformation corresponding to the candidate docking posture), reflecting its suitability as a solution. The fitness value also indicates the docking energy; a lower fitness value indicates lower docking energy, which in turn indicates greater stability of the docking posture.

[0082] C2: Based on the fitness value and the preset iteration strategy, the candidate docking posture is iteratively optimized until the preset termination condition is met, at which point the iteration optimization ends. The preset iteration strategy includes crossover, mutation, and selection.

[0083] The DE algorithm simulates the evolutionary process in nature, continuously evolving individuals within a population through crossover, mutation, and selection strategies to obtain the optimal solution. Generally, the DE algorithm randomly generates an initial population and iteratively optimizes the individuals within it. It is worth emphasizing that the initial population in this application is not randomly generated, but rather obtained based on the global search in step S101. This ensures that the initial state of the population is closer to the optimal solution, improving the efficiency of local searches and the accuracy of search results. In this embodiment, candidate docking postures in each region constitute the initial population for that region, and the individuals in this initial population are the docking conformations corresponding to the candidate docking postures.

[0084] Individuals in the population X = (X1, X2, ... X d (where d is the dimension of the individual vector) can be viewed as an N-dimensional spatial vector. If the flexibility of the protein is not considered, the individual dimension N is 6, consisting of translation and rotation vectors (x,y,z,θ,α,β). Here, (x,y,z) represents the movement of all atoms in the ligand, and (θ,α,β) represents the rotation of the ligand as a whole.

[0085] In this embodiment, an anisotropic network model (ANM) is introduced to consider the structural motion of proteins, using a linear combination of normal modes to describe the conformational changes between unbound and bound proteins (considering only the first ten normal modes). Therefore, the dimension of individuals in the population changes from 6 to 26 dimensions, with the latter 20 dimensions representing the deformation degree of receptor and ligand proteins along each mode. The overall motion of the protein structure can be viewed as the superposition of different low-frequency vibrational modes. The ANM calculates the vibrational modes based on the protein structure, and the latter 20 dimensions in the DE (deformation model) represent the amplitude corresponding to each mode. The anisotropic network model is used to simulate protein conformational changes, further optimizing the protein backbone.

[0086] Mutation strategy is one of the core elements of the differential evolution algorithm. For each individual in the population, the algorithm randomly selects three different individuals and generates a mutated individual through the mutation strategy. The mutation strategy typically involves multiplying the difference vector between two randomly selected individuals by a scaling factor and then adding it to the third individual to generate the mutated individual.

[0087] In one possible implementation, the mutation strategy is as follows (2):

[0088]

[0089] in, Let X represent the i-th mutated individual generated in the t-th iteration, where X represents an individual in the population, F represents the scaling factor (mutation rate), t represents the t-th iteration, and r1, r2, and r3 are random numbers, randomly generated integers in [0, NP-1] (where r1, r2, and r3 are distinct, NP is the population size, and NP ≥ 4). A mutated individual is generated by scaling the difference vectors of two randomly selected individuals in the population and combining them with the vector of another random individual. The value of F is related to the performance of the algorithm. An excessively high mutation rate can increase the search space of the algorithm, but it will lead to low search efficiency. On the other hand, an excessively low mutation rate may reduce population diversity and easily lead to premature convergence. In this embodiment, an adaptive F value method is adopted, as shown in the following calculation formula (3):

[0090]

[0091] Among them, T m F0 represents the maximum number of iterations, t represents the current number of iterations, and F0 = 0.45.

[0092] The crossover strategy aims to increase population diversity. During the crossover process, the algorithm exchanges some genes between the mutated individual and the target individual (i.e., the individual currently being processed) with a certain probability.

[0093] In one possible implementation, the crossover strategy is as follows (4):

[0094]

[0095] in, This represents the j-th dimension of information for the i-th individual in the t-th iteration. This represents the j-th dimension of information for the i-th variant individual generated in the t-th iteration. This represents the j-th dimension of information for individual i in the t-th iteration.

[0096] Based on the above calculation formula (4), as long as one of the following two conditions is met, The value comes from

[0097] (1) The randomly generated number is less than or equal to CR;

[0098] (2) The current dimension j equals j rand j rand It is a pre-generated random index.

[0099] When neither of the above conditions (1) nor (2) is satisfied, The value comes from

[0100] The purpose of the above crossover strategy is to combine the parent vectors and mutated individuals Cross-recombination was performed to generate experimental individuals. Where CR represents the crossover rate, and its value ranges from [0,1]. rand The index is randomly generated from [1, d] (d = 26), where d represents the dimension of the individual, to ensure... At least one gene can come from CR controls the degree of cross-recombination between the variant individual and the parent. When the CR value is too large, Most of the information comes from mutated individuals, which is beneficial for increasing population diversity; if it is too small, the information of the experimental individuals comes more from their parents, affecting the search for the global optimal solution of the population. CR is based on the following equation (5):

[0101] CR=0.5×[1+rand(0,1)](5)

[0102] In this embodiment, the average crossover rate is kept at around 0.75, which helps maintain the diversity of the population and avoids getting trapped in local optima.

[0103] The selection operation compares the fitness values ​​of the trial individual (i.e., the individual generated after the crossover strategy) and the target individual. If the fitness value of the trial individual is better than that of the target individual (for minimization problems) or worse (for maximization problems), the trial individual replaces the target individual and becomes a member of the new generation of the population. Otherwise, the target individual remains in the population.

[0104] In one possible implementation, the selection strategy is as follows (6), based on a greedy strategy from the test individuals. With parent vector Selecting the most fit individuals and passing them on to the next generation helps preserve excellent individuals and gradually improve the overall fitness level of the population.

[0105]

[0106] Based on the above calculation formula (6), when The value comes from the test individuals otherwise, The value comes from the parent individual.

[0107] Step S103: Cluster the target docking postures.

[0108] In this embodiment, the target docking posture is the docking posture after DE optimization. The target docking posture includes the optimized ligand protein conformation. The target docking posture is clustered according to the optimized ligand protein conformation.

[0109] The population optimized by DE often contains a large number of similar structures. The presence of these structures increases data redundancy and reduces prediction accuracy. Therefore, in order to reduce the impact of redundant information, this embodiment clusters the docking postures after DE optimization and uses clustering to remove some similar structures.

[0110] In this embodiment, a second clustering algorithm is used to cluster the target docking postures. This second clustering algorithm is a different type of clustering algorithm from the first clustering algorithm. The second clustering algorithm is a hierarchical clustering algorithm.

[0111] In this embodiment, a hierarchical clustering algorithm is used to merge and cluster the target docking postures, obtaining several clusters of target docking postures. Specifically, the hierarchical clustering algorithm can be a tree-based clustering algorithm. It does not require pre-specifying the number of clusters; instead, it calculates the similarity or distance between data points, gradually merging them into increasingly larger clusters, or progressively splitting a cluster into smaller clusters, until a certain stopping condition is met. The result of this algorithm is a tree structure, which can be pruned to obtain different numbers of clusters.

[0112] In this embodiment, the root mean square deviation (RMSD) of the ligands is used as an indicator to measure structural similarity. A smaller RMSD value indicates higher structural similarity. Then, the clustering tree is cut according to a set threshold (4 Å) to divide the structures into different clusters. For example, structures with an RMSD less than 4 Å are grouped into the same cluster. By aggregating highly similar structures together, a large amount of redundant information can be reduced, thereby improving the accuracy and efficiency of protein-protein docking.

[0113] Step S104: Based on the clustering results, determine the optimal docking posture of the receptor protein and the ligand protein.

[0114] Sort the target docking attitudes from lowest to highest according to their corresponding docking energies. Select the highest-ranked attitude as the optimal docking attitude. The lower the energy, the better the docking attitude.

[0115] In this implementation, the binding conformation with the lowest docking energy in the clusters retained after clustering is taken as the representative binding conformation in that cluster, and the final predicted docking attitude is obtained based on the representative binding conformation. The docking energy can be determined according to the aforementioned evaluation function. Figure 6 The structure of the complex predicted by docking simulation using the protein-protein docking method of this application is shown. Figure 6 1B6C, 1JTG, 1WDE, 2HRK, 1R6Q, and 1WQ1 are the protein IDs in the database.

[0116] In an experimental scenario, docking simulations were performed on a benchmark dataset (containing 230 proteins) using the proposed method. The results were compared with seven other widely used protein-protein docking procedures (including Zdock, Pydock, Megadock, Cluspro, Hdocklite, Lightdock, and Patchdock). This experiment primarily focused on the success rates of the eight docking procedures for top 1, top 5, and top 10. Top 10 is considered a successful docking if at least one of the top 10 predicted structures meets the DockQ value requirement; similarly, Top 1 and Top 5 are considered. This invention is the only docking procedure among the eight with a top 1 success rate approaching 30%, while the other seven procedures have success rates below 15% when only considering the optimal energy structure. Furthermore, the proposed method maintained the highest success rates for top 5 and top 10 predictions, at 45.7% and 49.1%, respectively. Therefore, the proposed method exhibits optimal docking performance compared to other docking procedures.

[0117] In this embodiment, by performing a global search on the surface space of the receptor protein, candidate docking postures are determined, i.e., potential binding sites are identified. Then, the conformation of the ligand protein is optimized based on the candidate docking postures and a preset optimization algorithm. Based on the results of the global search, potential binding sites are further optimized and screened to determine the target docking posture, which can effectively improve the search efficiency. Then, the target docking postures are clustered, and based on the clustering results, the optimal docking posture of the ligand protein docking with the receptor protein is determined, which can effectively improve the success rate of protein-protein docking.

[0118] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0119] Corresponding to the protein-protein docking method described in the above embodiments, Figure 7 A structural block diagram of the protein-protein docking device provided in the embodiments of this application is shown. For ease of explanation, only the parts related to the embodiments of this application are shown.

[0120] Reference Figure 7 The protein-protein docking device includes: a first determining unit 71, a second determining unit 72, a clustering analysis unit 73, and a docking attitude determining unit 74, wherein:

[0121] The first determining unit 71 is used to perform a global search of the surface space of the receptor protein to determine several candidate docking postures of the receptor protein and the ligand protein.

[0122] The second determining unit 72 is used to optimize the plurality of candidate docking postures according to a preset optimization algorithm and determine the target docking posture;

[0123] Clustering analysis unit 73 is used to cluster the target docking posture;

[0124] The docking attitude determination unit 74 is used to determine the optimal docking attitude between the receptor protein and the ligand protein based on the clustering results.

[0125] As one possible implementation of this application, the first determining unit 71 includes:

[0126] Combined with the conformation generation module, it is used to perform a global search of the surface space of the receptor protein using the FFT algorithm to generate multiple binding conformations;

[0127] The first determining module is used to screen the docking conformations according to preset screening conditions and determine several candidate docking postures.

[0128] As one possible implementation of this application, the second determining unit 72 includes:

[0129] The region division module is used to divide the surface space of the receptor protein into regions based on the plurality of candidate docking postures;

[0130] The second determining module is used to perform a local search on the candidate docking postures in each divided region according to the preset optimization algorithm, and determine the target docking posture.

[0131] As one possible implementation of this application, the candidate docking posture includes a ligand protein conformation, and the aforementioned region segmentation module is specifically used for:

[0132] Based on the position of the ligand protein conformation, the candidate docking postures are clustered to obtain several classes of candidate docking postures.

[0133] Based on the number of candidate docking postures, the surface space of the receptor protein is divided into regions, wherein each region contains one of the candidate docking postures.

[0134] As one possible implementation of this application, the second determining module described above is specifically used for:

[0135] The differential evolution algorithm is used to iteratively optimize the candidate docking postures in each region.

[0136] Based on the results of the iterative optimization, the target docking posture in each region is determined.

[0137] As one possible implementation of this application, the step of iteratively optimizing the candidate docking postures in each region using a differential evolution algorithm includes:

[0138] It is beneficial to use a preset evaluation function to evaluate the fitness of candidate docking attitudes in the region and determine the fitness value of candidate docking attitudes in the region;

[0139] The candidate docking postures are iteratively optimized based on the fitness value and a preset iteration strategy until a preset termination condition is met, at which point the iteration optimization ends. The preset iteration strategy includes crossover, mutation, and selection.

[0140] As one possible implementation of this application, the clustering analysis unit 73 is specifically used for:

[0141] The target docking postures are merged and clustered using a hierarchical clustering algorithm to obtain several clusters of target docking postures.

[0142] In this embodiment, by performing a global search on the surface space of the receptor protein, several candidate docking postures of the receptor protein and the ligand protein are determined, i.e., several potential binding sites are identified. Then, the candidate docking postures are optimized according to a preset optimization algorithm. Based on the results of the global search, potential binding sites are further optimized and screened to determine the target docking posture, which can effectively improve the search efficiency. Then, the target docking postures are clustered, and based on the clustering results, the optimal docking posture of the receptor protein and the ligand protein is determined, which can effectively improve the success rate of protein-protein docking.

[0143] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0144] This application embodiment also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements... Figures 1 to 6 The steps of any protein-protein docking method are represented.

[0145] This application embodiment also provides a smart device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements... Figures 1 to 6 The steps of any protein-protein docking method are represented.

[0146] This application also provides a computer program product that, when run on a server, causes the server to execute the implementation as described above. Figures 1 to 6 The steps of any protein-protein docking method are represented.

[0147] Figure 8 This is a schematic diagram of a smart device provided in an embodiment of this application. Figure 8 As shown, the intelligent device 8 in this embodiment includes: a processor 80, a memory 81, and a computer program 82 stored in the memory 81 and executable on the processor 80. When the processor 80 executes the computer program 82, it implements the steps in the various protein-protein docking method embodiments described above, for example... Figure 1 Steps S101 to S104 are shown. Alternatively, when the processor 80 executes the computer program 82, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 7 The functions of units 71 to 74 are shown.

[0148] For example, the computer program 82 may be divided into one or more modules / units, which are stored in the memory 81 and executed by the processor 80 to complete this application. The one or more modules / units may be a series of computer-readable instruction segments capable of performing a specific function, which describe the execution process of the computer program 82 in the smart device 8.

[0149] The intelligent device 8 may include, but is not limited to, a processor 80 and a memory 81. Those skilled in the art will understand that... Figure 8 This is merely an example of the smart device 8 and does not constitute a limitation on the smart device 8. It may include more or fewer components than shown, or combine certain components, or different components. For example, the smart device 8 may also include input / output devices, network access devices, buses, etc.

[0150] The processor 80 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0151] The memory 81 can be an internal storage unit of the smart device 8, such as a hard drive or memory of the smart device 8. The memory 81 can also be an external storage device of the smart device 8, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the smart device 8. Furthermore, the memory 81 can include both internal and external storage units of the smart device 8. The memory 81 is used to store the computer program and other programs and data required by the smart device. The memory 81 can also be used to temporarily store data that has been output or will be output.

[0152] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0153] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0154] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a device / terminal equipment, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0155] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0156] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A protein-protein docking method, characterized in that, The method includes: A global search is performed on the surface space of the receptor protein to determine several candidate docking postures between the receptor protein and the ligand protein, including: using an FFT algorithm to perform a global search on the surface space of the receptor protein to generate multiple binding conformations; and filtering the binding conformations according to preset screening conditions to determine several candidate docking postures. The optimization of the candidate docking postures according to the preset optimization algorithm to determine the target docking posture includes: clustering the candidate docking postures based on the position of the ligand protein conformation to obtain several classes of candidate docking postures; dividing the surface space of the receptor protein into regions according to the number of classes of the candidate docking postures, wherein each region contains one class of the candidate docking postures. The differential evolution algorithm iteratively optimizes the candidate docking postures in each region, including: evaluating the fitness of the candidate docking postures in the region using a preset evaluation function to determine the fitness value of the candidate docking postures in the region; iteratively optimizing the candidate docking postures based on the fitness value and a preset iteration strategy until a preset termination condition is met, at which point the iteration optimization ends; the preset iteration strategy includes crossover, mutation, and selection; and determining the target docking posture in each region based on the results of the iteration optimization. Cluster the target docking postures; Based on the clustering results, the optimal docking posture between the receptor protein and the ligand protein is determined.

2. The protein-protein docking method according to claim 1, characterized in that, The step of clustering the target docking posture includes: The target docking postures are merged and clustered using a hierarchical clustering algorithm to obtain several clusters of target docking postures.

3. A protein-protein docking device, characterized in that, The device includes: The first determining unit is used to perform a global search on the surface space of the receptor protein to determine several candidate docking postures of the receptor protein and the ligand protein, including: performing a global search on the surface space of the receptor protein using an FFT algorithm to generate multiple binding conformations; and filtering the binding conformations according to preset screening conditions to determine several candidate docking postures. The second determining unit is used to optimize the plurality of candidate docking postures according to a preset optimization algorithm to determine the target docking posture, including: clustering the plurality of candidate docking postures based on the position of the ligand protein conformation to obtain a plurality of classes of candidate docking postures; dividing the surface space of the receptor protein into regions according to the number of classes of candidate docking postures, wherein each region contains one class of candidate docking postures; iteratively optimizing the candidate docking postures in each region using a differential evolution algorithm, including: evaluating the fitness of the candidate docking postures in the region using a preset evaluation function to determine the fitness value of the candidate docking postures in the region; iteratively optimizing the candidate docking postures based on the fitness value and a preset iteration strategy until a preset termination condition is met, wherein the iterative optimization ends; and determining the target docking posture in each region based on the result of the iterative optimization. Clustering analysis unit, used to cluster the target docking posture; The docking attitude determination unit is used to determine the optimal docking attitude between the receptor protein and the ligand protein based on the clustering results.

4. A smart device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the protein-protein docking method as described in any one of claims 1 to 2.

5. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the protein-protein docking method as described in any one of claims 1 to 2.

Images