Virus three-dimensional reconstruction method and system based on three-dimensional Gaussian splashing

Abstract

The embodiment of the invention discloses a virus three-dimensional reconstruction method and system based on three-dimensional Gaussian splashing. The method comprises the following steps: acquiring three-dimensional point cloud data including a target virus; performing molecular dynamics simulation on the target virus based on a coarse graining molecular dynamics model, identifying a high-energy interaction region, and obtaining binding energy data of the target virus and a host molecule in an interaction process; tracing the conformation change of the virus structure to obtain conformation change data; performing three-dimensional Gaussian splashing processing on the point cloud data to generate an in-situ visual three-dimensional model of the target virus; wherein in the Gaussian splashing processing, the shape, the direction and the weight of a Gaussian kernel are dynamically adjusted according to combined energy data and conformation change data. According to the method and the device, efficient virus three-dimensional model generation is realized aiming at virus characteristics.

Description

Technical Field

[0001] The present application relates to the fields of biomedical imaging and computer vision technology, and in particular to a three-dimensional virus reconstruction method and system based on three-dimensional Gaussian splattering. Background Art

[0002] With the deepening of virological research, the interaction mechanism between viruses and host molecules (such as receptors and antibodies) has become key to drug design and vaccine development. Traditional virus 3D reconstruction methods mainly rely on cryo-electron microscopy (Cryo-EM) or X-ray crystallography to generate static high-resolution structures, but it is difficult to capture the dynamic conformational changes and energy distribution of viruses during the interaction process. 3D Gaussian splattering is a point cloud rendering technology that converts point cloud points into ellipsoidal Gaussian kernels to achieve efficient and high-fidelity 3D reconstruction. However, existing Gaussian splattering technology is mainly used for the reconstruction of buildings or objects, and is not optimized for the characteristics of viruses, making it difficult to meet the needs of virological research.

[0003] Therefore, there is an urgent need for a virus three-dimensional reconstruction method combined with Gaussian splattering to efficiently generate a dynamic three-dimensional model of the virus based on the characteristics of the virus. Summary of the Invention

[0004] The present application provides a three-dimensional virus reconstruction method based on three-dimensional Gaussian splattering, which realizes the efficient dynamic generation of a three-dimensional virus model based on virus characteristics.

[0005] This application provides the following solutions: According to a first aspect, a method for three-dimensional reconstruction of viruses based on three-dimensional Gaussian splashing is provided, the method comprising: obtaining three-dimensional point cloud data including a target virus, the point cloud data including spatial coordinates, geometric features and biological properties; performing molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model, calculating the energy distribution of the binding sites between the target virus and the host molecule, identifying high-energy interaction regions, and obtaining binding energy data of the target virus and the host molecule during the interaction process; and tracking the conformational changes of the virus structure through time series atomic coordinate analysis to obtain conformational change data; performing three-dimensional Gaussian splashing processing on the point cloud data to generate an in-situ visualized three-dimensional model of the target virus; wherein, in the Gaussian splashing processing, the shape, direction and weight of the Gaussian kernel are dynamically adjusted according to the binding energy data and the conformational change data.

[0006] According to an achievable method in an embodiment of the present application, in the Gaussian splattering processing, dynamically adjusting the shape, direction and weight of the Gaussian kernel according to the combined energy data and conformational change data includes: converting each point in the point cloud data into a Gaussian kernel, and the parameters of the Gaussian kernel include the Gaussian distribution representation, color representation and opacity representation corresponding to the point in the point cloud data; wherein the Gaussian distribution representation includes: shape information and direction information of the Gaussian kernel; adjusting the shape information of the Gaussian kernel corresponding to each point in the point cloud according to the combined energy distribution data; adjusting the direction information in the Gaussian distribution representation according to the conformational change data so that the shape of the Gaussian kernel is along the normal vector direction of the interaction area; and dynamically adjusting the weight of each Gaussian kernel according to the biological properties of the interaction area between the target virus and the host molecule.

[0007] According to an achievable method in an embodiment of the present application, the dynamic adjustment of the weights of each Gaussian kernel according to the biological properties of the interaction region between the target virus and the host molecule includes: obtaining the biological properties of the high-energy interaction region, wherein the biological properties include binding affinity and immunogenicity; using a quantitative mapping function to calculate the weights of each Gaussian kernel according to the binding affinity and immunogenicity, wherein, on the basis of assigning higher weights to interaction regions with high binding affinity, the weights of interaction regions with high immunogenicity are further increased.

[0008] According to an achievable method in an embodiment of the present application, the method further includes: when performing molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model, pre-calculating multiple key states of the target virus and host molecules during the interaction process, the key states including binding energy and conformational change data at different infection times; in the Gaussian splashing processing, dynamically adjusting the shape, direction and weight of the Gaussian kernel according to the binding energy data and conformational change data includes: using an interpolation algorithm to generate smoothly transitioned point cloud data and kernel parameters between the key states to form a pseudo-real-time demonstration effect; based on the pseudo-real-time data, dynamically adjusting the shape, direction and weight of the Gaussian kernel to generate a continuous three-dimensional rendering model.

[0009] According to an achievable method in an embodiment of the present application, the method further includes: modifying the parameters of the point cloud area with significant conformational changes and energy distribution based on the shape, direction and weight of the Gaussian kernel of the previous infection state; and / or; using a predictive optimization algorithm based on Kalman filtering to predict the conformational changes and energy distribution of the next infection state, and pre-adjusting the parameters of the Gaussian kernel.

[0010] According to an achievable method in an embodiment of the present application, the point cloud data is subjected to three-dimensional Gaussian splattering processing to generate an in-situ visualized three-dimensional model of the target virus, including: retaining a key Gaussian kernel, deleting other Gaussian kernels, and using the key Gaussian kernel to generate a continuous three-dimensional rendering model; wherein, the method for judging the degree of importance includes: calculating the binding energy value of the point cloud points corresponding to each Gaussian kernel according to the binding energy data, and obtaining a first Gaussian kernel set whose absolute value of the binding energy is higher than a first preset threshold; evaluating the conformational change amplitude of each point cloud point according to the conformational change data, and obtaining a second Gaussian kernel set whose conformational change is higher than a second preset threshold; obtaining a third Gaussian kernel set with high immunogenicity or high hydrophilicity according to the biological properties; and calculating the union of the first Gaussian kernel set, the second Gaussian kernel set, and the third Gaussian kernel set to obtain the key Gaussian kernel.

[0011] According to an achievable method in an embodiment of the present application, when the method performs three-dimensional Gaussian splashing processing on the point cloud data to generate a continuous visualized three-dimensional real-scene model of the target virus, it also includes: adjusting the color representation and opacity representation of the Gaussian kernel corresponding to the functional area of ​​the target virus based on the biological properties of the point cloud data; wherein the biological properties include immunogenicity or hydrophilicity; the functional area includes an immune epitope or a structurally stable area; the color representation is mapped to a specific color channel according to the immunogenicity, and the opacity representation is dynamically adjusted according to the hydrophilicity value.

[0012] According to the second aspect, a virus three-dimensional reconstruction system based on three-dimensional Gaussian splashing is provided, characterized in that the system includes: a point cloud data acquisition unit, configured to acquire three-dimensional point cloud data including a target virus, the point cloud data including spatial coordinates, geometric features and biological properties; a dynamics simulation unit, configured to perform molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model, calculate the energy distribution of the binding site between the target virus and the host molecule, identify high-energy interaction areas, and obtain the binding energy data of the target virus and the host molecule during the interaction process; and track the conformational changes of the virus structure through time series atomic coordinate analysis to obtain conformational change data; a three-dimensional reconstruction unit, configured to perform three-dimensional Gaussian splashing processing on the point cloud data to generate an in-situ visualized three-dimensional model of the target virus; wherein, in the Gaussian splashing processing, the shape, direction and weight of the Gaussian kernel are dynamically adjusted according to the binding energy data and conformational change data.

[0013] According to a third aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the steps of any one of the methods according to the first aspect are implemented.

[0014] According to a fourth aspect, an electronic device is provided, comprising: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions, which, when read and executed by the one or more processors, execute the steps of the method described in any one of the above-mentioned first aspects.

[0015] According to the specific embodiments provided in this application, this application discloses the following technical effects: This method achieves a visual modeling effect that is more consistent with the behavior of real biomolecules by introducing binding energy data and conformational change data into the three-dimensional Gaussian splattering process and dynamically adjusting the shape, direction, and weight of the Gaussian kernel. This not only improves the continuity and authenticity of the viral three-dimensional structural reconstruction, but also enhances the ability to express key interaction regions, highlighting high-energy binding regions and their conformational change characteristics, allowing the model to more accurately reflect the interaction mechanism between the virus and the host at the spatial level. It has higher expression accuracy and practical value in displaying biological properties and molecular dynamics, facilitating in-depth research on viral structure and visual analysis of interaction mechanisms.

[0016] Of course, any product implementing the present application does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0018] Figure 1 This is a system architecture diagram applicable to the embodiments of the present application; Figure 2 This is a flow chart of a method for three-dimensional virus reconstruction based on three-dimensional Gaussian splattering provided in an embodiment of the present application; Figure 3 A structural block diagram of a virus 3D reconstruction system based on 3D Gaussian splattering provided in an embodiment of the present application; Figure 4 A schematic block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of this application.

[0020] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "an", "the" and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise.

[0021] It should be understood that the term "and / or" as used herein is merely a description of the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0022] The word "if," as used herein, may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0023] There are already some three-dimensional reconstruction technologies for viruses. These methods mainly rely on cryo-electron microscopy (Cryo-EM) or X-ray crystallography to generate static high-resolution structures, but it is difficult to capture the dynamic conformational changes and energy distribution of viruses during interactions.

[0024] In view of this, the present application provides a new approach. To facilitate understanding of the present application, the system architecture on which the present application is based is first described. Figure 1 An exemplary system architecture to which the embodiments of the present application can be applied is shown. Figure 1 As shown in , the system architecture may include a user device and a 3D reconstruction system located on the server side.

[0025] Users can input stereoscopic data and point cloud data through their user devices, which then send them to a server-side 3D reconstruction system. The 3D reconstruction system can employ the methods provided in the embodiments of this application to perform 3D reconstruction of the target virus sample, generating an in-situ 3D model. The server-side can then send the 3D reality model to the user terminal, which then uses the 3D reality model for rendering, generating 2D or 3D images.

[0026] User devices include, but are not limited to, smart mobile terminals, smart home devices, wearable devices, and personal computers (PCs). Smart mobile devices include mobile phones, tablets, laptops, PDAs (Personal Digital Assistants), and internet-connected cars. Smart home devices include smart TVs and smart refrigerators. Wearable devices include smart watches and smart glasses.

[0027] The 3D reconstruction system can be set up as an independent server, a server group, or a cloud server. A cloud server, also known as a cloud computing server or cloud host, is a host product in the cloud computing service system. It solves the problems of difficult management and weak service scalability in traditional physical hosts and virtual private servers (VPS). Figure 1 In addition to the shown architecture, the 3D reconstruction system can also be set up on a computer terminal with strong computing capabilities.

[0028] It should be understood that Figure 1 The user equipment and 3D reconstruction system in the figure are merely illustrative. Any number of user equipment and 3D reconstruction systems may be provided as required.

[0029] Figure 2 Flowchart of the virus 3D reconstruction method based on 3D Gaussian splattering provided in the embodiment of the present application. The method can be performed by Figure 1 The virus 3D reconstruction system based on 3D Gaussian splattering in the system shown is executed. Figure 2 As shown in , the method may include the following steps: Step 201: Acquire three-dimensional point cloud data including target viruses, wherein the point cloud data includes spatial coordinates, geometric features, and biological attributes.

[0030] Step 202: Perform molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model, calculate the energy distribution of the binding site between the target virus and the host molecule, identify high-energy interaction regions, and obtain binding energy data between the target virus and the host molecule during the interaction process; and track the conformational changes of the virus structure through time series atomic coordinate analysis to obtain conformational change data.

[0031] Step 203: Perform three-dimensional Gaussian splash processing on the point cloud data to generate an in-situ visualized three-dimensional model of the target virus; wherein, in the Gaussian splash processing, the shape, direction and weight of the Gaussian kernel are dynamically adjusted according to the binding energy data and conformational change data.

[0032] As can be seen from the above process, the present invention achieves a visual modeling effect that is more consistent with the behavior of real biological molecules by introducing binding energy data and conformational change data into the three-dimensional Gaussian splattering process and dynamically adjusting the shape, direction and weight of the Gaussian kernel. This not only improves the continuity and authenticity of the reconstruction of the three-dimensional structure of the virus, but also enhances the ability to express key interaction regions, highlighting high-energy binding regions and their conformational change characteristics, so that the model can more accurately reflect the interaction mechanism between the virus and the host at the spatial level. It has higher expression accuracy and practical value in displaying biological properties and molecular dynamic information, which is conducive to in-depth research on viral structure and visual analysis of interaction mechanisms.

[0033] The following describes in detail each step of the above process and the effects that can be further produced, in conjunction with the embodiments. It should be noted that the "first" and "second" definitions involved in this disclosure do not have limitations in terms of size, order, or quantity, but are only used to distinguish them in name. For example, "first preset threshold" and "second preset threshold" are used to distinguish two thresholds.

[0034] First, the above step 201, namely "obtaining three-dimensional point cloud data including target viruses, wherein the point cloud data includes spatial coordinates, geometric features and biological properties" is described in detail in conjunction with the embodiment.

[0035] Point cloud data is a discrete three-dimensional representation composed of a large number of points, each carrying multidimensional information, used to describe the surface structure and properties of the target virus. Point cloud data not only contains the virus's geometric shape but also incorporates biological information, ensuring that subsequent rendering reflects both the virus's physical structure and biological functions. For example, for the novel coronavirus spike protein, point cloud data can represent its surface morphology and functional properties, providing comprehensive data support for dynamic visualization.

[0036] Point cloud data consists of three components: spatial coordinates, geometric features, and biological properties. Spatial coordinates refer to the position (x, y, z) of each point in three-dimensional space, measured in nanometers (nm), accurately describing the geometric position of the virus surface. Geometric features can include information such as normal vectors and curvature. The normal vector represents the surface orientation of the point cloud point and is used to adjust the orientation of the Gaussian kernel; the curvature reflects changes in surface shape and is used to assist in rendering details. Biological properties encompass functional properties such as immunogenicity and hydrophilicity. For example, an RBD binding site might have an immunogenicity score of 0.7 and a hydrophilicity score of 0.5. These properties provide a basis for visualizing interaction regions and functional areas.

[0037] Point cloud data can be obtained in a variety of ways, for example, using cryo-electron microscopy (Cryo-EM), X-ray crystallography, atomic force microscopy, scanning electron microscopy, transmission electron microscopy scanning, assisted by three-dimensional laser scanning or structured light scanning, to collect three-dimensional point cloud data of the target virus surface.

[0038] The following describes in detail the above step 202, namely, "performing molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model, calculating the energy distribution of the binding sites between the target virus and the host molecule, identifying high-energy interaction regions, and obtaining binding energy data between the target virus and the host molecule during the interaction process; and tracking the conformational changes of the virus structure through time series atomic coordinate analysis to obtain conformational change data," in conjunction with an embodiment.

[0039] This application uses molecular dynamics simulation technology to generate dynamic data of the target virus and host molecules during the interaction process. Molecular dynamics simulation uses numerical methods to solve the motion equations of molecular systems, simulate the dynamic behavior of atoms or molecules, and output time series data for analyzing the structural changes and interaction characteristics of the virus. This application combines coarse-grained models to calculate the energy distribution of the binding sites between viruses and host molecules, identify high-energy interaction areas, and track conformational changes to provide binding energy data and conformational change data for dynamic rendering of three-dimensional models. For example, for the interaction between the new coronavirus spike protein and the ACE2 receptor, simulation can reveal the energy distribution and conformational dynamics of the binding site, providing a basis for the study of infection mechanisms.

[0040] The coarse-grained molecular dynamics model is the core technical approach of this application. Compared to all-atom molecular dynamics simulations, the coarse-grained model simplifies multiple atoms into a single "bead" or pseudo-atom, reducing the system's degrees of freedom and thus significantly reducing the amount of computation. This patent uses a coarse-grained model to simulate the target virus, which can be implemented using a simulation platform. For example, the novel coronavirus spike protein consists of a viral protein of approximately 3,000 residues, a host molecule of 600 residues, and surrounding water molecules and ions.

[0041] Binding site energy distribution and interaction region identification are important outputs of molecular dynamics simulations. This feature calculates the binding site energy distribution between the virus and host molecules using a coarse-grained model, identifies high-energy interaction regions, and obtains binding energy data. Binding sites are regions on the viral surface where direct contact occurs with host molecules, such as the binding interface between the spike protein's receptor-binding domain and the ACE2 receptor. Binding energies are calculated using the molecular mechanics-Poisson-Boltzmann surface area method, with typical values ​​ranging from -10 to -15 kcal / mol. High-energy interaction regions typically have high absolute binding energies, such as the binding site of the receptor-binding domain, which can reach -15 kcal / mol, significantly higher than the -5 kcal / mol of non-interacting regions. This data is used in subsequent Gaussian kernel adjustments to enhance rendering effects for regions that interact with each other.

[0042] Time-series atomic coordinate analysis is used to track conformational changes in the viral structure and generate conformational change data. Molecular dynamics simulations output trajectory files containing atomic coordinates for each frame. Time-series atomic coordinate analysis processes these coordinates to quantify dynamic changes in the viral structure. For example, the conformational change from closed to open in the spike protein's receptor-binding domain can be calculated using a root mean square deviation (RMS) with a typical value of 2 angstroms. The analysis process involves extracting coordinates, calculating normal vectors, and updating point cloud data. Normal vectors are generated from coordinates using principal component analysis or curvature analysis and reflect changes in surface geometry. For example, the normal vector of the receptor-binding domain changes from [0, 0, 1] to [0.7, 0.7, 0]. Conformational change data drives point cloud updates, generating a dynamic point cloud with approximately 100,000 points per frame, providing geometric input for Gaussian splattering.

[0043] The above-mentioned step 203, i.e., "performing three-dimensional Gaussian splash processing on the point cloud data to generate an in-situ visualized three-dimensional model of the target virus; wherein, in the Gaussian splash processing, the shape, direction and weight of the Gaussian kernel are dynamically adjusted according to the combined energy data and conformational change data" is described in detail below in conjunction with an embodiment.

[0044] This application uses 3D Gaussian splattering technology to convert discrete point cloud data into a continuous 3D model to dynamically display the structure and interaction characteristics of the target virus. 3D Gaussian splattering is an advanced point cloud rendering method that generates a smooth surface effect by assigning an ellipsoidal Gaussian kernel to each point cloud point. It is suitable for efficient and high-fidelity 3D reconstruction. This application uses combined energy data and conformational change data to dynamically adjust the Gaussian kernel parameters to ensure that the model accurately reflects the dynamic changes in the virus's interaction regions, such as the spike protein receptor binding domain, such as the conformational transition from closed to open, and the visual prominence of high-energy binding sites, thereby supporting virology research and drug design.

[0045] This technical feature plays a crucial role in the patent. It integrates point cloud data, combined energy data, and conformational change data into a dynamic three-dimensional model, achieving high-fidelity visualization of the interaction characteristics of the virus. In Gaussian splashing, each point cloud point is represented as a three-dimensional Gaussian kernel. The three-dimensional model is essentially composed of a series of Gaussian kernels, which can be mapped to a specific perspective during the subsequent rendering of the in-situ three-dimensional model to generate an image. These Gaussian kernels are represented by Gaussian distribution, color, and opacity, and are usually encoded as vector representations respectively. Then a series of Gaussian kernels are actually encoded as Gaussian distribution matrices, color matrices, and opacity matrices respectively. After these matrices are initialized, they are continuously optimized through a series of training iterations. The resulting Gaussian kernel constitutes the in-situ three-dimensional model.

[0046] To meet the needs of 3D reconstruction of viruses, this application proposes a dynamic adjustment mechanism that dynamically adjusts the shape, orientation, and weight of the Gaussian kernel based on combined energy data and conformational change data. For example, in areas on the virus surface where there is strong binding energy with host molecules or where significant conformational changes occur, a sharper, more concentrated Gaussian kernel (i.e., a more elongated shape and higher weight) can be used to highlight these functional areas in the visualization model; while in areas where stability or less change occurs, a flatter, lower-weighted Gaussian kernel can be used to reduce visual clutter.

[0047] As an implementable method, the present application converts each point in the point cloud data into a Gaussian kernel, and the parameters of the Gaussian kernel include the Gaussian distribution representation, color representation and opacity representation corresponding to the point in the point cloud; wherein the Gaussian distribution representation includes: shape information and direction information of the Gaussian kernel; according to the combined energy distribution data, the shape information of the Gaussian kernel corresponding to each point in the point cloud is adjusted; according to the conformational change data, the direction information in the Gaussian distribution representation is adjusted so that the shape of the Gaussian kernel is along the normal vector direction of the interaction area; according to the biological properties of the interaction area between the target virus and the host molecule, the weight of each Gaussian kernel is dynamically adjusted.

[0048] Specifically, the Gaussian kernel parameters include a Gaussian distribution representation, a color representation, and an opacity representation, providing multi-dimensional control for rendering. The Gaussian distribution representation defines the geometric properties of the Gaussian kernel, including shape and orientation information. The shape information is determined by the eigenvalues ​​of the covariance matrix and controls the size of the kernel. The orientation information is determined by the eigenvectors of the covariance matrix and reflects the orientation of the Gaussian kernel, which is typically aligned with the normal vector of the point cloud point.

[0049] First, three-dimensional point cloud data of the target virus is acquired, and each point is assigned an initial Gaussian distribution, color, and opacity parameters. Molecular dynamics simulations are then performed to obtain binding energy distributions and conformational change data. In high-energy binding regions, the corresponding Gaussian kernel shape is adjusted to a more prolate or more concentrated form. For example, in the binding region of the SARS-CoV-2 spike protein and the human ACE2 receptor, point A in the point cloud is located at a key contact site between the RBD (receptor binding domain) and ACE2 within this high-energy binding region. Molecular dynamics simulations show that the binding energy value of this point is significantly higher than that of the surrounding area, indicating its important role in viral infection. To enhance the visual prominence and structural expression of this point in the 3D model, the corresponding Gaussian kernel shape can be adjusted to a more concentrated or prolate form. For example, by default, the covariance matrix of the Gaussian kernel is the identity matrix, indicating uniform expansion in all directions, i.e., a spherical Gaussian kernel distribution with a standard deviation of 1.0. After adjustment, the Gaussian kernel of point A can be set to have a smaller standard deviation along the surface normal direction (such as the z-axis), for example, a standard deviation of 0.5, and slightly stretched in the transverse direction (such as the x-axis and y-axis), for example, the x-axis standard deviation = the y-axis standard deviation = 0.8, so that it forms an ellipsoid that is more concentrated along the direction perpendicular to the contact surface.

[0050] Based on the results of the conformational change analysis, the displacement direction or surface normal of each point during the simulation is extracted, and the orientation of the Gaussian kernel is set to better reflect the actual conformational trends in the interaction region. Continuing with the previous example, in the receptor binding domain of the SARS-CoV-2 spike protein, during molecular dynamics simulations, a point C in the point cloud, located at an amino acid residue at the tip of the RBD, undergoes an average displacement of 0.8 nanometers along the z-axis during the conformational transition from "downward" to "upward," exhibiting a clear vertical opening trend. The unit vector of this displacement is (0, 0, 1). To reflect this conformational trend in the three-dimensional Gaussian splatter model, the Gaussian kernel of point C is adjusted from the default spherical distribution to an ellipsoid stretched along the z-axis. For example, the Gaussian kernel originally has a standard deviation of 1 in all directions, and the principal axis is oriented along the z-axis by default. The simulation analysis results show that the displacement direction of this point is 45 degrees tilted toward the xy plane direction (0.7, 0.7, 0). Therefore, the main axis direction is rotated to the direction of (0.7, 0.7, 0) through the rotation matrix to complete the direction adjustment.

[0051] Finally, the weight of each Gaussian kernel is dynamically adjusted based on biological properties such as the importance of the functional region in the viral life cycle, thereby completing a three-dimensional visualization reconstruction that is more accurate in both spatial morphology and biological significance.

[0052] Preferably, the dynamic adjustment of the weights of each Gaussian kernel in the present application can be achieved by the following method: obtaining the biological properties of the high-energy interaction region, wherein the biological properties include binding affinity and immunogenicity; using a quantitative mapping function to calculate the weight of each Gaussian kernel according to the binding affinity and immunogenicity, wherein the interaction region with high binding affinity is assigned a higher weight, and the interaction region with high immunogenicity is further weighted.

[0053] Specifically, the weight of the Gaussian kernel determines its opacity, controlling the visual intensity of the interaction region in the rendering. For example, high-weight regions appear brighter or more contrasted. This feature captures the biological properties of high-energy interaction regions and uses binding affinity and immunogenicity to calculate weights. Regions with high binding affinity are prioritized and weights are further increased based on high immunogenicity. For example, the receptor binding domain (RBD) of the SARS-CoV-2 spike protein, due to its high binding affinity and immunogenicity, is highlighted in deep red in the model, with a frame rate of 35 frames per second, supporting infection mechanism analysis. Biological properties, including binding affinity and immunogenicity, are core inputs to the weight calculation. Binding affinity reflects the strength of the interaction between the virus and host molecules and is calculated through molecular dynamics simulations to calculate the binding energy. For example, the binding site energy of the ACE2 receptor binding domain is -15 kilocalories per mole, normalized to 0.9. Immunogenicity indicates the ability of the interaction region to be recognized by the immune system. Predicted by tools such as the IEDB, it typically has a score between 0 and 1, with a score of 0.7 for the RBD epitope. These attributes are extracted from high-energy interaction regions, focusing on regions with absolute binding energies above 10 kcal / mol, ensuring that weight adjustments are concentrated on key interaction sites, such as the receptor binding domain, rather than non-interacting regions.

[0054] The quantitative mapping function is the mathematical basis of weight adjustment and is used to convert biological properties into Gaussian kernel weights. This application adopts a linear mapping function, and the interaction regions with high binding affinity are assigned higher weights, and the interaction regions with high immunogenicity are further improved in weight. That is, this application first assigns higher weights to the interaction regions with high binding affinity, and then further improves the weights for the interaction regions with high immunogenicity, with binding affinity being the dominant factor in weight. Specifically, the weight calculation formula can be a weight equal to the normalized value of binding affinity multiplied by a coefficient of 0.6 plus an immunogenicity score multiplied by a coefficient of 0.4, wherein the weight contribution of binding affinity is greater, accounting for 60%, and immunogenicity is an auxiliary factor, accounting for 40%. For example, the receptor binding domain binding affinity is normalized to 0.9, the immunogenicity score is 0.7, and the weight calculation is 0.6 times 0.9 plus 0.4 times 0.7, which is equal to 0.82, corresponding to an opacity of 0.82. If the immunogenicity is low, such as 0.3, the weight drops to 0.66, indicating that binding affinity is the dominant factor and immunogenicity only further increases the weight based on high binding affinity.

[0055] In order to save computing resources and improve the efficiency of three-dimensional reconstruction, this application pre-calculates multiple key states of the virus and host molecules during the interaction process when performing molecular dynamics simulation of the target virus based on a coarse-grained molecular dynamics model. The key states include binding energy and conformational change data of different infection states. This application efficiently obtains dynamic data of the interaction between the virus and host molecules by pre-calculating key states, providing input for subsequent pseudo-real-time rendering. Key states refer to representative time point data during the interaction between the virus and host molecules, including binding energy and conformational change data.

[0056] The present application can also use a predictive optimization algorithm based on Kalman filtering to predict the conformational changes and energy distribution of the next infection state and pre-adjust the parameters of the Gaussian kernel. Kalman filtering is a recursive estimation method that predicts future states and optimizes estimates by combining historical observation data and system dynamic models. Kalman filtering treats the trajectory data of molecular dynamics simulation as a time series, and the state variables include point cloud coordinates, normal vectors, and binding energy. The filter predicts the conformational changes of the next infection state through a state transition model, such as the trend of root mean square deviation changes, and corrects the prediction through the observation model, integrating the actual data of the current frame. The prediction mechanism is specifically aimed at predicting conformational changes and energy distribution. Conformational changes are quantified by changes in point cloud coordinates and normal vectors. For example, when the receptor binding domain changes from closed to open, the root mean square deviation increases from 1 angstrom to 2 angstroms, and the normal vector changes from the initial direction to the new direction. The Kalman filter predicts the point cloud position and direction of the next frame based on the coordinate and normal vector sequences of the previous frames. Energy distribution predicts binding energy trends, such as an increase in binding site energy from -10 kcal / mol to -15 kcal / mol in a receptor binding domain. The filter then estimates the energy value for the next frame. Parameter pre-adjustment optimizes Gaussian kernel parameters based on predicted data. Kernel parameters include the eigenvalues ​​and eigenvectors of the covariance matrix, weights, and colors. Predicted conformational changes adjust the kernel orientation. For example, the normal vector is predicted to have a new orientation, eigenvectors are pre-aligned, and the kernel shape is stretched along the surface.

[0057] Preferably, when the present application generates an in-situ visualized three-dimensional model through three-dimensional Gaussian splattering, it can only render the key Gaussian kernel, thereby saving computing resources. Specifically, the key Gaussian kernel is retained, other Gaussian kernels are deleted, and the key Gaussian kernel is used to generate a continuous three-dimensional rendering model; wherein, the method for judging the importance includes: according to the binding energy data, calculating the binding energy value of the point cloud points corresponding to each Gaussian kernel, and obtaining a first Gaussian kernel set whose absolute value of the binding energy is higher than a first preset threshold; according to the conformational change data, evaluating the conformational change amplitude of each point cloud point, and obtaining a second Gaussian kernel set whose conformational change is higher than a second preset threshold; according to the biological properties, obtaining a third Gaussian kernel set with high immunogenicity or high hydrophilicity; calculating the union of the first Gaussian kernel set, the second Gaussian kernel set and the third Gaussian kernel set, and obtaining the key Gaussian kernel.

[0058] Key Gaussian kernel screening is based on three criteria, each of which is used to generate three kernel sets and then take the union of them. The binding energy criterion calculates the binding energy value of each point cloud point based on the binding energy data. Kernels with absolute values ​​above a first preset threshold, such as 10 kcal / mol, are selected to form the first Gaussian kernel set. For example, a receptor binding domain binding site with an energy of -15 kcal / mol, far exceeding the threshold, is included in the first set, reflecting the strength of the interaction. The conformational change criterion assesses the magnitude of conformational change in the point cloud points, quantified by the root mean square deviation (RMS), and selects kernels with a second preset threshold, such as 1 angstrom, to form the second Gaussian kernel set. For example, a receptor binding domain transitioning from closed to open with an RMS deviation of 2 angstroms is included in the second set, reflecting dynamic geometry. The biological property criterion selects kernels with high immunogenicity (e.g., a score of 0.7) or high hydrophilicity (e.g., a score of 0.5), forming the third Gaussian kernel set to highlight epitope regions. Finally, the union of the three sets is taken to obtain the key Gaussian kernels, ensuring coverage of all important regions.

[0059] When performing 3D Gaussian splattering on point cloud data to generate an in-situ visualized 3D model of the target virus, this application also optimizes the visualization of the functional regions of the virus 3D model by adjusting the color and opacity of the Gaussian kernel of the functional regions based on immunogenicity and hydrophilicity. This includes adjusting the color representation and opacity representation of the Gaussian kernel corresponding to the functional regions of the target virus based on the biological properties of the point cloud data; wherein the biological properties include immunogenicity or hydrophilicity; the functional regions include immune epitopes or structurally stable regions; the color representation is mapped to a specific color channel based on immunogenicity, and the opacity representation is dynamically adjusted based on the hydrophilicity value.

[0060] Specifically, functional regions refer to areas on the viral surface with specific biological significance, including immune epitopes and structurally stable regions. Immune epitopes are regions on the viral surface that are recognized by antibodies, and their high immunogenicity requires them to be highlighted in the model. Structurally stable regions maintain the integrity of the viral structure, and their high hydrophilicity is generally associated with stability. Color representation is mapped to a specific color channel based on immunogenicity. For example, epitope regions with an immunogenicity score above 0.7 are mapped to green with RGB values ​​of 0, 255, 0, while regions below 0.3 are mapped to gray with RGB values ​​of 128, 128, 128. Opacity representation is dynamically adjusted based on the hydrophilicity value. For example, structurally stable regions with a hydrophilicity value above 0.5 are assigned an opacity of 0.8, while hydrophilic regions below -0.5 are assigned an opacity of 0.4. The mapping function is implemented using a linear or exponential form, for example, opacity equals 0.5 plus 0.3 multiplied by the hydrophilicity value.

[0061] The above-mentioned method provided in the embodiment of this application can be applied to a variety of application scenarios, including but not limited to: infection mechanism research, through dynamic visualization of the interaction area between the virus and host molecules, such as the binding process of the novel coronavirus spike protein receptor binding domain and the ACE2 receptor, revealing conformational changes and binding energy distribution; drug design, highlighting high-energy binding sites, supporting target optimization of neutralizing antibodies or small molecule drugs; vaccine development, using immunogenicity mapping to render immune epitopes, analyze epitope exposure dynamics, and assist in immunogen design; virus structure analysis, showing the hydrophilicity and hydrophobicity properties of structurally stable regions such as the capsid, and evaluating virus stability. These scenarios improve analysis accuracy and interactivity through efficient, pseudo-real-time three-dimensional model rendering, providing powerful tools for antiviral research.

[0062] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0063] According to another embodiment, a three-dimensional virus reconstruction system based on three-dimensional Gaussian splattering is provided. Figure 3 FIG. 1 is a schematic block diagram of a virus 3D reconstruction system based on 3D Gaussian splattering according to an embodiment. Figure 3 As shown, the system 300 includes: The point cloud data acquisition unit 301 is configured to acquire three-dimensional point cloud data including the target virus, where the point cloud data includes spatial coordinates, geometric features, and biological attributes.

[0064] The dynamics simulation unit 302 is configured to perform molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model, calculate the energy distribution of the binding site between the target virus and the host molecule, identify high-energy interaction regions, and obtain the binding energy data between the target virus and the host molecule during the interaction process; and track the conformational changes of the virus structure through time series atomic coordinate analysis to obtain conformational change data.

[0065] The three-dimensional reconstruction unit 303 is configured to perform three-dimensional Gaussian splash processing on the point cloud data to generate an in-situ visualized three-dimensional model of the target virus; wherein, in the Gaussian splash processing, the shape, direction and weight of the Gaussian kernel are dynamically adjusted according to the binding energy data and conformational change data.

[0066] As an implementable method, the three-dimensional reconstruction unit 303 can be configured to: convert each point in the point cloud data into a Gaussian kernel when dynamically adjusting the shape, direction and weight of the Gaussian kernel according to the combined energy data and conformational change data during Gaussian splattering processing, and the parameters of the Gaussian kernel include the Gaussian distribution representation, color representation and opacity representation corresponding to the point in the point cloud data; wherein the Gaussian distribution representation includes: shape information and direction information of the Gaussian kernel; according to the combined energy distribution data, adjust the shape information of the Gaussian kernel corresponding to each point in the point cloud; according to the conformational change data, adjust the direction information in the Gaussian distribution representation so that the shape of the Gaussian kernel is along the normal vector direction of the interaction area; according to the biological properties of the interaction area between the target virus and the host molecule, dynamically adjust the weight of each Gaussian kernel.

[0067] As an implementable method, the three-dimensional reconstruction unit 303 can be configured to dynamically adjust the weights of each Gaussian kernel according to the biological properties of the interaction area between the target virus and the host molecule as follows: obtain the biological properties of the high-energy interaction area, the biological properties including binding affinity and immunogenicity; use a quantitative mapping function to calculate the weights of each Gaussian kernel according to the binding affinity and immunogenicity, wherein, on the basis of assigning higher weights to the interaction areas with high binding affinity, the weights of the interaction areas with high immunogenicity are further increased.

[0068] As an implementable manner, the dynamics simulation unit 302 can be configured as follows: when performing molecular dynamics simulation on the target virus based on the coarse-grained molecular dynamics model, pre-calculating multiple key states of the target virus and host molecules during the interaction process, wherein the key states include binding energy and conformational change data of different infection states. The three-dimensional reconstruction unit 303 can be configured as follows: in the Gaussian splashing process, dynamically adjusting the shape, direction and weight of the Gaussian kernel according to the binding energy data and conformational change data includes: using an interpolation algorithm to generate smoothly transitioned point cloud data and kernel parameters between the key states to form a pseudo-real-time demonstration effect; based on the pseudo-real-time data, dynamically adjusting the shape, direction and weight of the Gaussian kernel to generate a continuous three-dimensional rendering model.

[0069] As an implementable manner, the dynamic simulation unit 302 can be configured to: modify the parameters of the point cloud area where the conformational changes and energy distribution are significant based on the shape, direction and weight of the Gaussian kernel of the previous infection state; and / or; use a predictive optimization algorithm based on Kalman filtering to predict the conformational changes and energy distribution of the next state and pre-adjust the parameters of the Gaussian kernel.

[0070] As an implementable manner, the three-dimensional reconstruction unit 303 can be configured to: retain the key Gaussian kernel, delete other Gaussian kernels, and use the key Gaussian kernel to generate a continuous three-dimensional rendering model when dynamically adjusting the weights of each Gaussian kernel according to the biological properties of the interaction area between the target virus and the host molecule; wherein the method for judging the degree of importance includes: calculating the binding energy value of the point cloud points corresponding to each Gaussian kernel according to the binding energy data, and obtaining a first Gaussian kernel set whose absolute value of the binding energy is higher than a first preset threshold; evaluating the conformational change amplitude of each point cloud point according to the conformational change data, and obtaining a second Gaussian kernel set whose conformational change is higher than a second preset threshold; obtaining a third Gaussian kernel set with high immunogenicity or high hydrophilicity according to the biological properties; and calculating the union of the first Gaussian kernel set, the second Gaussian kernel set, and the third Gaussian kernel set to obtain the key Gaussian kernel.

[0071] As an implementable manner, the three-dimensional reconstruction unit 303 can be configured to: adjust the color representation and opacity representation of the Gaussian kernel corresponding to the functional area of ​​the target virus based on the biological properties of the interaction area between the target virus and the host molecule based on the biological properties of the point cloud data; wherein, the biological properties include immunogenicity or hydrophilicity; the functional area includes an immune epitope or a structurally stable area; the color representation is mapped to a specific color channel according to the immunogenicity, and the opacity representation is dynamically adjusted according to the hydrophilicity value.

[0072] Each embodiment in this specification is described in a progressive manner. The same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiment. The system embodiment described above is only exemplary, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without making any creative efforts.

[0073] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0074] In addition, an embodiment of the present application further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps of any one of the methods in the aforementioned method embodiments are implemented.

[0075] And an electronic device comprising: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions, which, when read and executed by the one or more processors, execute the steps of the method described in any one of the aforementioned method embodiments.

[0076] The present application also provides a computer program product, comprising a computer program, which implements the steps of any one of the methods described in the aforementioned method embodiments when executed by a processor.

[0077] in, Figure 4 The electronic device architecture is shown as an example, and may include a processor 410, a video display adapter 411, a disk drive 412, an input / output interface 413, a network interface 414, and a memory 420. The processor 410, the video display adapter 411, the disk drive 412, the input / output interface 413, the network interface 414, and the memory 420 may be communicatively connected via a communication bus 430.

[0078] The processor 410 may be implemented as a general-purpose CPU, a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and may be used to execute relevant programs to implement the technical solutions provided in this application.

[0079] The memory 420 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 420 can store an operating system 421 for controlling the operation of the electronic device 400 and a basic input and output system (BIOS) 422 for controlling the low-level operations of the electronic device 400. In addition, it can also store a web browser 423, a data storage management system 424, and a virus 3D reconstruction system 425 based on 3D Gaussian splattering, etc. The above-mentioned virus 3D reconstruction system 425 based on 3D Gaussian splattering can be the application program that specifically implements the aforementioned steps in the embodiment of the present application. In short, when the technical solution provided by the present application is implemented through software or firmware, the relevant program code is stored in the memory 420 and is called and executed by the processor 410.

[0080] The input / output interface 413 is used to connect to input / output modules to enable information input and output. The input / output modules can be configured as components within the device (not shown) or externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, and various sensors, while output devices may include a display, speaker, vibrator, indicator light, and the like.

[0081] The network interface 414 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.).

[0082] The bus 430 comprises a pathway for transmitting information between the various components of the device, such as the processor 410 , the video display adapter 411 , the disk drive 412 , the input / output interface 413 , the network interface 414 , and the memory 420 .

[0083] It should be noted that although the above device only shows a processor 410, a video display adapter 411, a disk drive 412, an input / output interface 413, a network interface 414, a memory 420, a bus 430, etc., in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may also include only the components necessary to implement the solution of the present application, and does not necessarily include all the components shown in the figure.

[0084] Through the description of the above embodiments, it can be seen that those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer program product. The computer program product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments of the present application or certain parts of the embodiments.

[0085] The above is a detailed introduction to the technical solutions provided by this application. Specific examples are used herein to illustrate the principles and implementation methods of this application. The description of the above embodiments is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the contents of this specification should not be understood as limiting this application.

Claims

1. A three-dimensional virus reconstruction method based on three-dimensional Gaussian splattering, characterized in that: The method comprises: Acquiring three-dimensional point cloud data including target viruses, wherein the point cloud data includes spatial coordinates, geometric features, and biological properties; Performing molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model to calculate the energy distribution of the binding sites between the target virus and the host molecule, identify high-energy interaction regions, and obtain binding energy data during the interaction between the target virus and the host molecule; and tracking conformational changes in the virus structure through time series atomic coordinate analysis to obtain conformational change data; The point cloud data is subjected to three-dimensional Gaussian splash processing to generate an in-situ visualized three-dimensional model of the target virus; wherein, in the Gaussian splash processing, the shape, direction and weight of the Gaussian kernel are dynamically adjusted according to the binding energy data and the conformational change data.

2. The method according to claim 1, characterized in that In the Gaussian splattering process, dynamically adjusting the shape, direction, and weight of the Gaussian kernel according to the binding energy data and the conformational change data includes: Converting each point in the point cloud data into a Gaussian kernel, wherein the parameters of the Gaussian kernel include a Gaussian distribution representation, a color representation, and an opacity representation corresponding to the point in the point cloud data; wherein the Gaussian distribution representation includes shape information and direction information of the Gaussian kernel; Adjusting shape information of the Gaussian kernel corresponding to each point in the point cloud according to the combined energy distribution data; Adjusting the direction information in the Gaussian distribution representation according to the conformational change data so that the shape of the Gaussian kernel is along the normal vector direction of the interaction region; The weights of each Gaussian kernel are dynamically adjusted according to the biological properties of the interaction region between the target virus and the host molecule.

3. The method according to claim 2, characterized in that The dynamically adjusting the weights of the Gaussian kernels according to the biological properties of the interaction region between the target virus and the host molecule includes: obtaining biological properties of the high-energy interaction region, wherein the biological properties include binding affinity and immunogenicity; A quantitative mapping function is used to calculate the weight of each Gaussian kernel according to the binding affinity and immunogenicity, wherein the weight of the interaction region with high immunogenicity is further increased on the basis of assigning a higher weight to the interaction region with high binding affinity.

4. The method according to claim 1, wherein The method further comprises: When performing molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model, multiple key states of the target virus and host molecules during the interaction process are pre-calculated, wherein the key states include binding energy and conformational change data at different infection times; In the Gaussian splattering process, dynamically adjusting the shape, direction, and weight of the Gaussian kernel according to the binding energy data and the conformational change data includes: An interpolation algorithm is used to generate smoothly transitioned point cloud data and kernel parameters between the key states, thereby forming a pseudo real-time demonstration effect; Based on the pseudo real-time data, the shape, direction and weight of the Gaussian kernel are dynamically adjusted to generate a continuous three-dimensional rendering model.

5. The method according to claim 4, characterized in that The method further comprises: Based on the shape, orientation, and weight of the Gaussian kernel of the previous infection state, the parameters of the point cloud regions with significant conformational changes and energy distribution are modified; and / or; A predictive optimization algorithm based on Kalman filtering is used to predict the conformational changes and energy distribution of the next infection state and pre-adjust the parameters of the Gaussian kernel.

6. The method according to claim 3, characterized in that Performing three-dimensional Gaussian splattering on the point cloud data to generate an in-situ visualized three-dimensional model of the target virus includes: retaining a key Gaussian kernel, deleting other Gaussian kernels, and generating a three-dimensional rendering model using the key Gaussian kernel; The method for determining the importance includes: Calculating the binding energy values ​​of the point cloud points corresponding to the respective Gaussian kernels according to the binding energy data to obtain a first Gaussian kernel set whose binding energy absolute value is greater than a first preset threshold; According to the conformational change data, the conformational change amplitude of each point cloud point is evaluated to obtain a second Gaussian kernel set whose conformational change is higher than a second preset threshold; According to the biological properties, a third Gaussian kernel set with high immunogenicity or high hydrophilicity is obtained; A union of the first Gaussian kernel set, the second Gaussian kernel set, and the third Gaussian kernel set is calculated to obtain the key Gaussian kernel.

7. The method according to claim 1, characterized in that When performing three-dimensional Gaussian splattering on the point cloud data to generate an in-situ visualized three-dimensional model of the target virus, the method further includes: Adjusting the color representation and opacity representation of the Gaussian kernel corresponding to the functional region of the target virus based on the biological attributes of the point cloud data; wherein the biological attributes include immunogenicity or hydrophilicity; and the functional region includes an immune epitope or a structurally stable region; The color representation is mapped to a specific color channel according to immunogenicity, and the opacity representation is dynamically adjusted according to the hydrophilicity value.

8. A virus 3D reconstruction system based on 3D Gaussian splattering, characterized in that: The system comprises: a point cloud data acquisition unit configured to acquire three-dimensional point cloud data including target viruses, wherein the point cloud data includes spatial coordinates, geometric features, and biological attributes; a dynamics simulation unit configured to perform molecular dynamics simulation on the target virus based on a coarse-grained molecular dynamics model, calculate the energy distribution of the binding site between the target virus and the host molecule, identify high-energy interaction regions, and obtain binding energy data during the interaction between the target virus and the host molecule; and track conformational changes of the virus structure through time series atomic coordinate analysis to obtain conformational change data; The three-dimensional reconstruction unit is configured to perform three-dimensional Gaussian splash processing on the point cloud data to generate an in-situ visualized three-dimensional model of the target virus; wherein, in the Gaussian splash processing, the shape, direction and weight of the Gaussian kernel are dynamically adjusted according to the binding energy data and the conformational change data.

9. An electronic device, characterized in that: include: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions, wherein the program instructions, when read and executed by the one or more processors, execute the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images