Virtual screening method and system for pharmacophores based on virtual reality

Abstract

The invention relates to the technical field of medicine virtual screening, in particular to a virtual screening method and system for pharmacophores based on virtual reality. The method comprises the following steps: inputting a protein structure and a compound library into a VR scene as a screening object through a molecular resource loading module; analyzing a pocket area in the protein structure by using a pocket recognition module, and performing three-dimensional observation and position calibration on the pocket area in the VR scene to judge the actual position and range of the pocket area; wherein conformation consistency analysis is performed on the pocket region based on a conformation integration mechanism, and spatial fusion is performed on the pocket region according to a conformation consistency analysis result to form a conformation stable pocket region. According to the method, the whole process of protein structure, pocket recognition, pharmacophore construction and molecular docking is integrated in a virtual reality (VR) environment, and a user can perform three-dimensional observation on a conformational stable pocket area in an immersive manner.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of virtual screening of drugs, in particular to a virtual screening method and system of pharmacophore based on virtual reality. BACKGROUND

[0002] With the continuous development of computer-aided drug design (CADD) technology, pharmacophore template and virtual screening have become an important means of lead compound discovery. Traditional pharmacophore modeling mainly relies on two-dimensional structure analysis or static three-dimensional structure calculation, which is difficult to fully reflect the dynamic flexibility characteristics of protein targets and the spatial complexity of ligand-target interaction. At the same time, the existing screening platform is mostly based on desktop graphical interface, and users lack intuitive perception of the spatial relationship between molecular structure, binding pocket and pharmacophore characteristics, resulting in strong subjectivity of pharmacophore template construction, limited accuracy, and difficulty in efficient correction. In addition, the shape change of the pocket caused by protein conformation diversity is often ignored, so that virtual screening based on a single rigid structure is prone to false positives or missing real active molecules. Therefore, a virtual screening method and system of pharmacophore based on virtual reality are provided. SUMMARY

[0003] The present application aims to provide a virtual screening method and system of pharmacophore based on virtual reality to solve the problems of low screening accuracy, high false positive rate and poor human-computer collaboration efficiency caused by lack of protein conformation flexibility, insufficient three-dimensional spatial interaction perception and strong subjectivity of pharmacophore modeling in the traditional pharmacophore virtual screening method proposed in the background.

[0004] To achieve the above-mentioned purpose, the present application aims to provide a virtual screening method of pharmacophore based on virtual reality, comprising:

[0005] S1, inputting protein structure and compound library into VR scene as screening objects through a molecular resource loading module;

[0006] S2, analyzing the pocket region in the protein structure using a pocket recognition module, and performing three-dimensional observation and position calibration on the pocket region in the VR scene for judging the actual position and range of the pocket region;

[0007] Among them, the conformation consistency of the pocket region is analyzed based on the conformation integration mechanism, and the spatial fusion of the pocket region is performed according to the conformation consistency analysis result, forming a conformation stable pocket region;

[0008] S3, generating initial functional feature points based on the conformation stable pocket region using a pharmacophore analysis module, and performing interactive editing and correction on the spatial position, type and tolerance parameters of the pharmacophore in the VR environment to form a pharmacophore template;

[0009] S4, perform a pharmacophore screening process based on a pharmacophore template, match and score each candidate compound, and generate a preliminary screening molecule list according to the matching score;

[0010] S5, perform fine molecular docking on the preliminary screening molecule list based on the actual position and range of the conformationally stable pocket region, obtain the binding conformation and energy score of each molecule, and use it to screen the final candidate molecule.

[0011] As a further improvement of the technical solution, in S2, the pocket recognition module is used to analyze the pocket region in the protein structure, and the pocket region is observed and positioned in the VR scene, including the following steps:

[0012] S2.1, rotate, move and zoom the protein through three-dimensional operations

[0013] S2.2, perform a pocket recognition algorithm on the protein structure based on the pocket recognition module to generate potential binding sites, i.e. pocket regions, and mark the pocket regions in the VR scene in a highlighted manner;

[0014] S2.3, position the identified pocket region based on the geometric features and spatial parameters of the protein structure.

[0015] As a further improvement of the technical solution, in S2.2, the pocket recognition module is used to analyze the pocket region in the protein structure, and the pocket region is observed and positioned in the VR scene, including the following steps:

[0016] S2.21, convert the atomic coordinates of the protein structure into a three-dimensional surface mesh model, calculate the spatial coordinates and neighborhood atomic information of each mesh point;

[0017] S2.22, perform cavity scanning on the surface mesh of the three-dimensional surface mesh model, identify the protein surface depression area and the cavity volume that can accommodate ligands, and calculate the geometric center, volume and opening direction of each cavity, forming a preliminary candidate pocket region set;

[0018] S2.23, calculate the geometric features of each candidate pocket region to form a feature vector of the candidate pocket;

[0019] S2.24, evaluate the feature vector of the candidate pocket by constructing a preset scoring function, select the pocket region whose evaluation result exceeds the threshold n, and mark it as a potential binding site;

[0020] S2.25, perform conformation consistency analysis on the pocket region generated in step S2.24 based on the conformation integration mechanism, and perform spatial fusion on the pocket region according to the conformation consistency analysis result to form a conformationally stable pocket region.

[0021] As a further improvement of the technical solution, in the S2.25, the conformation stable pocket region is formed, comprising the following steps:

[0022] Based on the pocket region generated in step S2.24, normal mode analysis is performed on the protein structure to obtain the dominant motion direction of the protein structure under the low-frequency eigenmode, and a plurality of lightweight alternative conformations are generated within a reasonable perturbation range along the principal mode to form a conformation set;

[0023] For each pocket region, each conformation in the conformation set is repositioned and its corresponding spatial occupancy region is calculated to obtain the projection result of the pocket region under different conformations, forming a conformation mapping pocket set;

[0024] Cross-conformation consistency analysis is performed on the conformation mapping pocket set to generate a conformation consistency score ;

[0025] According to the conformation consistency score and a preset threshold , the candidate pocket region that meets is marked as a conformation stable candidate pocket;

[0026] The conformation stable pocket region is subjected to spatial fusion processing to obtain a fused conformation stable pocket region.

[0027] As a further improvement of the technical solution, the conformation stable pocket region is subjected to spatial fusion processing to obtain a fused conformation stable pocket region, involving the following steps:

[0028] The conformation mapping pocket set is repositioned and scaled to the original coordinate system;

[0029] Based on the geometric center offset range of the conformation mapping pocket after coordinate alignment and scale normalization, the boundary grid of each conformation pocket is subjected to Boolean fusion strategy for fusion;

[0030] The fused pocket region is subjected to voxel sampling and grid reconstruction to calculate the stable spatial volume commonly maintained under each conformation, forming a fused conformation stable pocket region.

[0031] As a further improvement of the technical solution, in the S3, the pharmacophore template is formed, comprising the following steps:

[0032] S3.1, the pharmacophore initial feature points are generated by the protein local structure, key residue distribution and pocket shape based on the pharmacophore analysis module;

[0033] S3.2, the pharmacophore initial feature points are displayed near the pocket region with different marks and superimposed with the protein structure for display, which is convenient for users to observe the geometric relationship with the pocket.

[0034] S3.3, checking the distance, direction and geometric layout between the pharmacophore point and the protein residue;

[0035] S3.4, after the completion of the inspection, the pharmacophore is saved as a pharmacophore template.

[0036] As a further improvement of the technical solution, in S4, the matching score of each candidate compound is calculated, and the preliminary screening molecule list is generated according to the matching score ranking, including the following steps:

[0037] S4.1, input the pharmacophore template as the matching rule into the screening module;

[0038] S4.2, set the screening conditions according to the type and tolerance of the pharmacophore feature point;

[0039] S4.3, pre-filtering based on the compound library information, output the candidate compound set after pre-filtering;

[0040] S4.4, traverse each molecule in the candidate compound set, and calculate the geometric alignment of the pharmacophore feature point between the three-dimensional structure of the molecule and the pharmacophore template;

[0041] S4.5, compare the matching degree between the functional group position of the compound and the pharmacophore template, and perform multi-dimensional energy weighted pharmacophore matching scoring to generate energy weighted matching score;

[0042] S4.6, sort all candidate compounds from high to low according to the energy weighted matching score, and output the preliminary screening molecule list.

[0043] As a further improvement of the technical solution, in S4.5, the multi-dimensional energy weighted pharmacophore matching score includes the following steps:

[0044] S4.51, construct the basic energy contribution coefficient for different pharmacophore points , and based on the chemical microenvironment of the protein, energy correct the basic energy contribution coefficient to generate the energy corrected contribution coefficient ;

[0045] S4.52, according to the spatial deviation between the matching points in the candidate molecule and the pharmacophore template points, construct a Gaussian type distance attenuation weight for each pharmacophore feature point;

[0046] S4.53, for directional pharmacophore features, introduce an angle-dependent weight factor to evaluate the consistency between the interaction direction of the candidate molecule and the direction of the pharmacophore template;

[0047] S4.54, the contribution coefficient The spatial weight and the angle-dependent weight factor are jointly integrated, a multi-dimensional weighted score is calculated for each pharmacophore feature point, and the energy weighted matching score of the whole molecule is obtained by summing up the weighted scores of all the pharmacophore feature points.

[0048] As a further improvement of the technical solution, in S5, the binding conformation and energy score of each molecule are obtained, including the following steps:

[0049] S5.1, based on the preliminary screening molecule list, each candidate molecule in the preliminary screening molecule list is introduced into the docking module, and the three-dimensional grid of the conformationally stable pocket region is taken as the constraint space;

[0050] S5.2, for each candidate molecule, perform conformational sampling and rotational translation optimization to generate multiple groups of initial binding poses that can enter the pocket region, and perform collision checking and geometric accessibility checking on each pose to generate the checked conformation;

[0051] S5.3, re-perform local energy minimization on the checked conformation, and generate the final comprehensive binding energy score by combining interaction strength, conformational stability and pocket space complementarity;

[0052] S5.4, sort all candidate molecules according to the comprehensive binding energy score, generate a candidate molecule set, and present the binding conformation, key interaction and docking search space of the protein and ligand in a three-dimensional manner through the docking visualization module.

[0053] On the other hand, the present application provides a virtual screening system for a pharmacophore based on virtual reality, comprising 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 virtual screening method for the pharmacophore based on virtual reality as described in any one of the above.

[0054] Compared with the prior art, the present application has the following advantages:

[0055] 1、In the virtual screening method and system for a pharmacophore based on virtual reality, the protein structure, pocket identification, pharmacophore construction and molecular docking are integrated in a virtual reality (VR) environment, and the user can immerse in the three-dimensional observation, spatial calibration and interactive editing of the conformationally stable pocket region. Especially in the pharmacophore template construction stage, the user can intuitively drag, modify the feature point type and tolerance parameter, and verify the spatial matching relationship with the key residues in real time, effectively overcoming the modeling deviation caused by the lack of spatial perception in the traditional two-dimensional interface. At the same time, the "conformationally stable pocket region" generated by the conformation integration mechanism significantly improves the adaptability of the pharmacophore template to the flexibility change of the protein, making the subsequent virtual screening more close to the real biological environment, thereby greatly improving the hit rate of preliminary screening and the efficiency of lead compound discovery.

[0056] 2、The virtual screening method and system of a pharmacophore based on virtual reality disclosed by the present application, which combines pre-physical and chemical rule filtering, multi-dimensional pharmacophore matching scoring with energy correction, and fine molecular docking based on conformational stability pocket area, forms a three-level progressive screening system of "coarse screening-fine screening-verification". Among them, the pharmacophore matching scoring introduces local chemical microenvironment correction (such as pH, charge density), Gaussian space attenuation weight and direction angle dependent factor, so that the scoring not only reflects the geometric matching degree, but also embodies the authenticity of the physical and chemical interaction; finally, the user can intuitively evaluate the ligand binding conformation, key interaction and docking box setting rationality through the VR visual docking result, support fast decision and structure optimization. The process significantly shortens the drug discovery cycle and reduces the experimental trial and error cost while ensuring the screening accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 The overall method flowchart of the present application. DETAILED DESCRIPTION

[0058] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0059] Embodiment 1: Please refer to Figure 1 The present embodiment provides a virtual screening method of a pharmacophore based on virtual reality, which comprises the following steps:

[0060] S1, input the protein structure and compound library into the VR scene as screening objects through a molecular resource loading module;

[0061] In the present embodiment, the molecular resource loading module includes a local molecular resource module, a website resource pulling module, a PDB molecular library pulling module, and a two-dimensional code loading module.

[0062] The local molecular resource module (Load From File) is used to load molecular resources from the user's local storage. The user places the.pdb format molecular file in the preset path, and the system reads the file and loads the molecular three-dimensional structure to the visualization window through the module, realizes the rapid import and visualization display of local molecular resources, and provides input data for subsequent pocket recognition and pharmacophore analysis;

[0063] The website pull resource module (Load From Internet) is used to automatically pull the corresponding molecular file from the online database based on the PDB ID (Protein Data Bank Identifier) provided by the user, and load it into the visualization window; the system provides virtual keyboard input and control functions, enabling users to conveniently input PDB ID and complete remote molecular resource acquisition, providing technical support for users to access structural data without the need for local storage;

[0064] The PDB molecular library pull module (Load From Resources) is used to select target molecules from the software built-in or preset molecular library, including drug molecules, target proteins, ligands, and other experimental or test compounds, and load the selected molecules into the visualization window for three-dimensional display; this module supports multiple categories of molecular resource management, ensuring that users can quickly call the sorted common molecular library, facilitating molecular screening and docking operations;

[0065] The two-dimensional code loading module (Load By QRCode) parses the target molecular file from the two-dimensional code containing the PDB file information and loads it into the visualization window, realizing fast acquisition of molecular resources; this module provides a convenient import solution based on two-dimensional codes for users, simplifying the molecular data acquisition process, and is particularly suitable for mobile or laboratory applications that quickly collect molecular information;

[0066] Users can select different display modes such as Hide (hidden), Wireframe (wireframe model), Stick (stick model), Ball (ball model), BallStick (ball stick model), Cartoon (ribbon model), Surface (surface model), AEO (atomic ellipsoid model (usually refers to anisotropic displacement ellipsoid)), AEO+ (enhanced atomic ellipsoid model) through the style module (Style), making the protein structure more suitable for observing and identifying potential binding pockets.

[0067] S2, analyze the pocket region in the protein structure using the pocket recognition module, and perform three-dimensional observation and position calibration of the pocket region in the VR scene, for judging the actual position and range of the pocket region;

[0068] Among them, the conformation consistency of the pocket region is analyzed based on the conformation integration mechanism, and the spatial fusion of the pocket region is performed according to the conformation consistency analysis result, to form a conformationally stable pocket region;

[0069] In this embodiment, the pocket recognition module is used to analyze the pocket region in the protein structure, and the pocket region is observed and calibrated in three dimensions in the VR scene, including the following steps:

[0070] S2.1, rotate, move and scale the protein through three-dimensional operations to view the surface grooves and internal cavity regions of the protein from different angles, providing spatial references for pocket identification;

[0071] wherein the three-dimensional operations include Rotate operation, Move operation, Scale operation, Auto operation; Rotate is a rotation operation for adjusting the orientation of the protein molecule structure in three-dimensional space to observe its surface grooves and internal cavity regions from different angles; Move is a translation operation for changing the position of the molecule in three-dimensional visualization space to adjust the observation angle center; Scale is a scaling operation for changing the display scale of the molecule in three-dimensional space to focus on the detail area of the protein structure, such as the active site, thereby providing a spatial reference for pocket identification and position calibration; Auto is an automatic control for controlling the molecule to be fixed or to rotate automatically;

[0072] S2.2, based on the pocket identification module, perform a pocket identification algorithm on the protein structure (i.e., use the pocket identification module (XPocket module) in the "Function" menu module, and the system automatically performs a pocket identification algorithm on the protein structure) to generate potential binding sites, i.e., pocket regions, and mark the pocket regions in the VR scene in a highlighted or boundary manner;

[0073] wherein the user observes the pocket region from a 360° free angle of view, and views the depth, volume, opening direction and internal composition of the pocket through VR environment immersion to confirm whether it meets the biological meaning of the binding site;

[0074] Further, the "Function" menu module includes Sequence module (sequence viewing function for viewing sequence information of the molecule), Water module (water molecule management function for performing related management operations on water molecules in the molecular structure), XPocket module (pocket identification function for identifying pocket regions in the molecular structure), XPharma module (pharmacophore feature, drug molecule labeling function for labeling drug molecules to distinguish and analyze), Atom module (atom-level operation function for supporting fine operations on atoms in the molecule), Dock Result module (docking result visualization function for presenting the results of molecular docking simulation in a visualized manner);

[0075] wherein the pocket identification algorithm based on the pocket identification module for the protein structure to generate potential binding sites, i.e., pocket regions, includes the following steps:

[0076] S2.21, convert the atomic coordinates of the protein structure into a three-dimensional surface mesh model, calculate the spatial coordinates of each mesh point and the neighborhood atomic information, so as to detect the subsequent groove and cavity; Specifically: based on the input atomic coordinates of the protein structure, a molecular surface potential field with atomic radius superposition is constructed, and a continuous three-dimensional surface mesh is generated by space sampling or Marching Cubes isosurface extraction algorithm; Then, for each generated mesh point, its three-dimensional coordinates are used to perform neighborhood query in the atomic KD-Tree or octree spatial index structure, all atoms within a predetermined radius from the mesh point are searched, and their atomic type, van der Waals radius and charge attribute are recorded; Finally, the spatial coordinates, normal vector and associated neighborhood atom set of each mesh point are output;

[0077] S2.22, cavity scanning of the surface mesh of the three-dimensional surface mesh model, identifying the protein surface depression area and the cavity volume that can accommodate ligands, and calculating the geometric center, volume and opening direction of each cavity, forming a preliminary candidate pocket area set; Specifically: first, based on the normal information of the mesh point and the relative distance between the meshes, the area where the surface appears to be depressed is identified, and the continuous gap inside the protein not occupied by atoms is searched by using the space bounding box or voxelization method; Then, by performing connectivity analysis on the mesh elements of these recessed areas and connected gaps, the voxel set of each independent cavity is extracted, and the geometric center (such as the centroid), the overall volume of the cavity and the opening direction along the mesh normal or the maximum opening section are calculated; Finally, the cavity area that can accommodate small molecule ligands is uniformly marked as a candidate pocket to form a preliminary set of pocket regions;

[0078] S2.23, calculate the geometric features of each candidate pocket region, including depth, curvature, pocket opening size, to form the feature vector of the candidate pocket; Specifically: when calculating the geometric features of each candidate pocket region, the system first extracts the basic data for describing the shape from the cavity inside and outside surface mesh, such as the normal of the mesh vertex, the boundary contour line and the shortest distance from the mesh to the surrounding atoms; Then, the depth of the pocket is calculated by measuring the shortest distance along the path from the deepest point inside the pocket to the opening on the surface, the principal curvature and average curvature of the recessed area are obtained by local surface fitting (such as quadratic surface fitting), and the opening size of the pocket is obtained by measuring the maximum cross-sectional area or the circumcircle radius of the pocket boundary; Finally, the above depth, curvature, caliber and other numerical values are combined into a feature vector in a predetermined order as a structured input for subsequent pocket scoring;

[0079] S2.24, evaluate the feature vector of the candidate pocket by constructing a predetermined scoring function, filter the pocket regions whose evaluation results exceed the threshold n, and mark them as potential binding sites;

[0080] Wherein, the predetermined scoring function is:

[0081] ;

[0082] wherein, is the evaluation result, i.e., the evaluation score, is the depth, is the curvature, is the pocket opening size, is the weight coefficient of the depth, is the weight coefficient of the curvature, is the weight coefficient of the pocket opening size, and the weight coefficients are all dimensionless and determined by expert experience, is the reference length (e.g., 10 angstroms or a characteristic size of the protein);

[0083] S2.25, performing conformation consistency analysis on the pocket region generated in step S2.24 based on a conformation integration mechanism, and performing spatial fusion on the pocket region according to the conformation consistency analysis result to form a conformationally stable pocket region (the protein structure in different conformational states (such as open and closed conformations or induced fit states) can cause significant changes in the volume, shape and opening direction of the pocket);

[0084] The conformation consistency analysis on the pocket region based on the conformation integration mechanism and the formation of the conformationally stable pocket region mainly solve the problem of unstable or inconsistent spatial position of the binding pocket caused by the conformational change (such as opening and closing, twisting or breathing movement) of the protein structure in virtual screening. Traditional methods usually rely on a single static protein structure for pocket recognition, which cannot accurately reflect the dynamic flexibility of the protein in the physiological state, and may identify false pockets that exist only in a specific conformation or ignore key binding sites, thereby reducing the accuracy and reliability of virtual screening. The conformation integration mechanism can capture the binding pocket that exists continuously in the dynamic behavior of the protein by generating multiple alternative conformations of the protein and analyzing the spatial consistency and stability of the pocket in different conformations, reducing false positive results caused by conformation fluctuations, and improving the robustness of pocket recognition. Its role is to provide a more realistic and stable binding environment for subsequent pharmacophore construction and molecular docking, ensuring that the candidate molecules screened not only match the static structure but also adapt to the conformational changes of the protein, thereby improving the success rate of virtual screening and the practical application value;

[0085] The conformation consistency analysis on the pocket region based on the conformation integration mechanism and the formation of the conformationally stable pocket region include the following steps:

[0086] Based on the pocket region generated in step S2.24, performing normal mode analysis on the protein structure to obtain the dominant motion direction of the protein structure in the low-frequency eigenmode, and generating multiple lightweight alternative conformations within a reasonable perturbation range along the normal mode to form a conformation set ​ The conformation set is used to simulate the conformation fluctuations that the protein can appear under natural flexibility, providing input for subsequent evaluation of the stability of the candidate pocket; Specifically: first, based on the input protein three-dimensional atomic coordinates, an elastic network model (ENM) is constructed, with amino acid residues as nodes and distance constraints as springs, to establish a simplified force field describing the overall flexibility of the protein: When constructing the elastic network model (ENM) based on the three-dimensional atomic coordinates of the protein, the following reproducible specific parameters are adopted: Node setting: Take the a-carbon atoms (Ca) of each amino acid residue as nodes, and the node coordinates are the original coordinates of the Ca atoms in the PDB file; Glycine and other Ca-free residues use the coordinates of the main chain carbon atoms (C) as nodes; Only the nodes are connected by springs with a spatial distance (Angstrom), and the initial force constant of the spring is uniformly ; Then, the Hessian matrix of the elastic network is calculated and its eigenvalues and eigenvectors are solved, from which the lowest frequency eigenmodes are selected as the dominant motion directions reflecting the overall opening and closing, twisting and breathing changes of the protein; Next, with the original structure as the reference, the atomic coordinates are continuously translated and deformed in the positive and negative directions of each low-frequency mode within a predetermined perturbation amplitude (such as ±1~3 normalized mode units), generating multiple slightly offset and energy-acceptable alternative conformations; Finally, all the alternative conformations generated along different modes are collected into a conformation set , which is used for subsequent conformation consistency analysis and pocket stability evaluation;

[0087] For each pocket region, the corresponding spatial occupancy region is relocated and calculated on each conformation in the conformation set, including the geometric center, pocket boundary grid and local surface shape, obtaining the projection results of the pocket region under different conformations, forming a conformation mapping pocket set , which is used to describe the spatial fluctuation characteristics of the pocket with the change of conformation; Wherein, for each pocket region, when mapping in the conformation set, the system first maps the pocket geometric center, boundary grid and surface description parameters in the original structure to the coordinate system of each conformation through coordinate alignment (such as rigid body superposition based on amino acid residues); Then, the local grid sampling and cavity reconstruction are re-executed on the three-dimensional surface of the conformation , and the spatial occupancy region of the pocket is updated according to the new position of the surrounding atoms, and the geometric properties of the pocket under this conformation, such as the geometric center, boundary contour, deepest point, local concave shape and opening direction, are recalculated; Through the above steps, the complete spatial projection of the pocket under the conformation is obtained, thereby forming the corresponding conformation mapping pocket ; Finally, repeat the above mapping process for all conformations in the conformation set to obtain the conformation mapping pocket set , for subsequent conformation consistency evaluation and stable pocket extraction;

[0088] Conformation consistency analysis is performed on the conformation mapping pocket set to generate a conformation consistency score , including overlap degree and center offset degree;

[0089] Among them, the overlap degree is (dimensionless):

[0090] ;

[0091] The center offset degree is (unit: angstrom):

[0092] ;

[0093] After normalization processing, the greater the score is, the more consistent it is:

[0094] ;

[0095] The consistency score is:

[0096] ;

[0097] In the formula, is the spatial occupation area of the th pocket in the conformation set, is the spatial occupation area of the th pocket in the conformation set, is the volume of the calculation space area, is the Euclidean distance between the geometric centers of the th and th pockets, is an adjustment parameter for controlling the influence strength of the offset on the score (positive number, set according to the pocket size), is the weight coefficient of the overlap degree, is the weight coefficient of the center offset degree, and the weight coefficients are dimensionless and determined by expert experience, is the maximum allowed offset distance, and the experience value is 5 angstroms;

[0098] According to the conformation consistency score and the preset threshold , the candidate pocket region that meets is marked as a conformation stable candidate pocket, otherwise it is considered as a non-stable pocket susceptible to conformation fluctuation and is removed;

[0099] The conformation stable pocket region is subjected to spatial fusion processing to obtain a fused conformation stable pocket region;

[0100] Further, obtaining the fused conformation stable pocket region involves the following steps:

[0101] Realigning the coordinates of the conformation mapping pocket set (to compare all pockets in the same reference frame) and scale normalization (to eliminate the deviation caused by rotation, translation or slight scale difference between different conformations, to ensure the comparability of subsequent fusion analysis) to make them in the same reference coordinate system used in step S2.25 consistency analysis;

[0102] Based on the geometric center offset range of the conformation mapping pocket after coordinate alignment and scale normalization, the boundary grid of each conformation pocket is executed Boolean fusion strategy for fusion (using the intersection strategy of the majority of conformations existing in the region, or the soft union method based on weight distribution, to integrate the pocket space information under different conformations, to obtain the preliminary fused pocket region); Specifically: when performing Boolean fusion on the conformation mapping pocket, the system first aligns the coordinates of each conformation pocket in the same reference coordinate system and normalizes the scale, so that the geometric center and the overall size can be directly compared; Then, according to the aligned geometric center position, calculate the offset range of each pocket, judge which grid points exist in the majority of conformations; Next, perform Boolean intersection operation (AND) on the boundary grid of each conformation pocket, and retain the space region occupied by all conformations at the same time, so as to remove the occasional or single conformation specific pocket part, and at the same time, the soft union processing can be combined with the weight strategy to process part of the region, to integrate the space information under different conformations; Finally, generate the fused conformation stable pocket region through grid reconstruction and voxel sampling, to ensure that the extracted pocket region exists and is geometrically stable under multiple conformations;

[0103] The fused pocket region is voxelized and reconstructed, the stable spatial volume commonly maintained in each conformation is calculated, and the conformationally stable pocket region after fusion is formed. This step is used to remove accidental or single-conformation-specific pocket parts. In this way, stable spatial features that exist in multiple conformations can be extracted to generate the final fused conformationally stable pocket region. When voxelizing and reconstructing the fused pocket region, the system first divides the fused boundary mesh into a regular three-dimensional voxel mesh. For each voxel, it determines its occupancy in each conformation mapping pocket and counts the frequency of each voxel appearing in all conformations. Then, according to the preset threshold, voxels occupied in most or all conformations are selected, and these common stable voxels are retained as the core pocket region. Next, the selected voxels are surface reconstructed (such as Marching Cubes algorithm) to generate a continuous three-dimensional boundary mesh, and the total volume, geometric center and surface shape of the fused pocket are calculated. Finally, the continuous mesh obtained is the conformationally stable pocket region after fusion, which ensures that the region exists in multiple conformations and has stable geometric features, and can be used for subsequent pharmacophore construction and molecular docking.

[0104] S2.3, the user can further adjust the viewing angle, based on the geometric shape features of the protein structure and the spatial parameters such as pocket boundary, center point and opening direction, to calibrate the position of the identified pocket region, to ensure that the identification result is consistent with the true geometric shape of the protein structure, for subsequent pharmacophore construction and docking box setting, and to confirm the key parameters of the pocket region, including center point position, boundary range and opening direction, etc. Information, which will be used in the subsequent docking step to set PocketCenter and Pocket Size.

[0105] S3, based on the conformationally stable pocket region, an initial functional feature point is generated using a pharmacophore analysis module, and the spatial position, type and tolerance parameters of the pharmacophore are interactively edited and corrected in the VR environment to form a pharmacophore template.

[0106] In this embodiment, based on the conformationally stable pocket region determined in step S2, the user enters the visualization viewing angle near the pocket to provide spatial positioning basis for pharmacophore identification.

[0107] The steps of forming a pharmacophore template include:

[0108] S3.1, based on the pharmacophore analysis module (XPharma module), generate initial feature points of the pharmacophore through protein local structure, key residue distribution and pocket shape, including aromatic ring, hydrogen bond donor, hydrogen bond acceptor, hydrophobic region, positive center (or cation), negative center (or anion); Specifically: first, analyze the protein local structure in the conformationally stable pocket region, including amino acid residue type, spatial arrangement of main chain and side chain and their contactable surface; Then identify the key residue functional groups (such as hydroxyl, amino, carboxyl, aromatic ring, etc.) and their distribution positions, combine the pocket geometric shape and depth information to determine the spatial region where hydrogen bond, hydrophobic interaction, ionic interaction or aromatic stacking occurs with the ligand; Then, automatically generate initial feature points of the pharmacophore in these regions, and identify them by type (such as hydrogen bond donor, hydrogen bond acceptor, aromatic ring, hydrophobic point, positive and negative charge center), while recording their three-dimensional coordinates, direction vectors and spatial tolerance, so as to be interactively adjusted and optimized in the VR environment later, forming a complete initial point set of the pharmacophore;

[0109] S3.2, display the initial feature points of the pharmacophore near the pocket region with different identifiers (such as balls, icons, color coding), and superimpose them with the protein structure for easy observation of their geometric relationship with the pocket by the user;

[0110] Wherein, the user uses VR three-dimensional operation tools such as "Rotate, Move, Scale" to observe from multiple angles whether the pharmacophore points are in reasonable spatial positions, for example, whether they are close to key residues or key functional groups; The user can select any pharmacophore point and modify its type to aromatic ring, hydrogen bond donor / acceptor, hydrophobic, anion / cation, etc. through the VR controller to correct the points that may not be accurately identified by the system; The user can directly grab the pharmacophore points and drag their spatial positions through the VR controller, so that they can more accurately fit the spatial features inside the protein pocket, such as being close to hydrogen bond residues or hydrophobic patches; The user can set matching tolerances for each pharmacophore point, such as identification radius, direction vector limit and other geometric parameters, which are used as spatial scoring standards in subsequent compound matching; If the automatically identified feature points do not meet the actual pocket requirements, the user can delete the redundant points and add new pharmacophore points based on the VR fine view to improve the matching accuracy;

[0111] S3.3, the user checks the distance, direction and geometric layout between the pharmacophore points and the protein residues in the VR scene to ensure that the generated template can truly reflect the best binding mode of the ligand in the pocket;

[0112] S3.4, after the check is completed, the user saves the pharmacophore as a pharmacophore template for subsequent virtual screening and molecular docking processes.

[0113] S4. Based on the pharmacophore template, a pharmacophore screening process is performed, each candidate compound is matched and scored, and a preliminary screening molecule list is generated according to the matching score;

[0114] In this embodiment, each candidate compound is matched and scored, and a preliminary screening molecule list is generated according to the matching score, including the following steps:

[0115] In this embodiment, the user loads all candidate compounds through "Load From File / Load From Internet / Load From Resources" and the like, so that they can be called one by one in the VR environment by the screening algorithm;

[0116] S4.1, the pharmacophore template is input as a matching rule into the screening module, wherein the screening module is used for automatically evaluating and sorting the candidate compound set based on the preset pharmacophore template and related screening conditions, so as to identify molecules with high matching degree with the target protein pocket; the module compares and calculates the three-dimensional structure of each candidate compound with the pharmacophore template according to parameters such as pharmacophore feature point type, spatial tolerance, geometric angle and missing matching tolerance, generates a matching score, and outputs a preliminary screening molecule list according to the score result, providing a candidate molecule set for subsequent fine molecule docking;

[0117] S4.2, screening conditions are set according to the type and tolerance of the pharmacophore feature points, at least including: matching tolerance radius (allowed spatial deviation range of each feature point), minimum number of matching points (minimum number of matching pharmacophore points), allowed proportion of missing matching points (such as allowing 1 point to be missing), geometric angle deviation limit (such as directional H-bond feature), and these screening conditions affect the final matching score of each compound;

[0118] S4.3, pre-filtering is performed based on the compound library information (such as molecular weight, polarity, structural complexity, Lipinski rule, etc.); the basic physicochemical parameters (such as molecular weight, cLogP, number of hydrogen bond donors / acceptors, polar surface area, structural complexity, number of rotatable bonds, etc.) of each compound in the compound library are extracted, and these parameters are compared with the preset drug compatibility rules one by one, such as the restrictions of Lipinski rule on the molecular weight, liposolubility and hydrogen bond characteristics of oral drugs, and the restrictions of Veber rule on the polar area and the number of rotatable bonds; at the same time, the system excludes molecules containing strong reactive groups, toxic warning fragments and unsynthesizable structures in combination with the structure rule library; only the compounds that meet all the filtering conditions are retained for the next stage, thereby quickly narrowing the candidate space and improving the efficiency and reliability of subsequent pharmacophore matching, and outputting the candidate compound set that can participate in pharmacophore matching after pre-filtering, for eliminating molecules that do not meet the pharmacophore structure requirements;

[0119] S4.4, for each molecule in the candidate compound set, perform a geometric alignment calculation of the three-dimensional structure of the molecule with the pharmacophore template; specifically: the system first generates or loads the three-dimensional conformation of each compound, and extracts the functional groups in the molecule that can be used as pharmacophore matching points (such as hydrogen donor atoms, hydrogen acceptor atoms, aromatic ring centers, hydrophobic carbon chains, positive and negative charge centers, etc.); then these functional points are paired with the corresponding feature points in the pharmacophore template according to the type, and the optimal spatial transformation matrix is calculated using a geometric alignment algorithm (such as least squares rigid body alignment) to make the feature point cloud of the compound and the three-dimensional feature point cloud of the pharmacophore template coincide in space position and direction to the greatest extent; during the alignment process, the distance deviation, direction vector matching degree and tolerance range are evaluated simultaneously, and the matching mode that cannot meet the geometric constraints is screened out; finally, the optimal alignment conformation of the compound and the pharmacophore template is obtained, providing a basis for subsequent matching score calculation;

[0120] S4.5, compare the matching degree between the functional group position of the compound and the pharmacophore template, and perform multi-dimensional energy-weighted pharmacophore matching scoring to generate energy-weighted matching scores;

[0121] The multi-dimensional energy-weighted pharmacophore matching scoring mainly solves the problem of oversimplified scoring function in traditional pharmacophore screening methods. Traditional methods usually only score based on the geometric matching (such as distance, direction) of feature points, ignoring the local chemical microenvironment (such as charge distribution, acidity and basicity) of the protein binding pocket and the difference in energy contribution of different interaction types, resulting in scoring results that cannot accurately reflect the true physical and chemical trend of molecular binding potential, thereby reducing the hit rate of virtual screening and the quality of lead compounds. Multi-dimensional energy-weighted matching scoring integrates basic energy contribution, local chemical microenvironment correction, spatial geometric deviation attenuation and directional interaction weight to build a scoring system that is closer to the real biological physical process. Its advantage is to upgrade simple geometric matching to multi-dimensional energy-weighted evaluation, so that the score can not only distinguish whether the molecule is "similar" to the pharmacophore template, but also predict its "quality" and "strength" of the binding interaction. Its role is to significantly improve the enrichment effect and reliability of the virtual screening results, and to preferentially select high-quality candidate molecules that are highly complementary to the target in terms of geometric shape and interaction energy;

[0122] The multi-dimensional energy-weighted pharmacophore matching scoring includes the following steps:

[0123] S4.51, construct a basic energy contribution coefficient for different pharmacophore points (such as hydrogen bond acceptor, hydrogen bond donor, aromatic ring, hydrophobic point, positive and negative charge point, etc.) , and based on the protein chemical microenvironment (such as local acidity and basicity, charge density), the basic energy contribution coefficient is energy corrected to generate the energy corrected contribution coefficient The pharmacophore score can reflect the influence of the real chemical environment of the protein on the strength of the interaction, such as local pH, charge density, thereby improving the expression ability of the matching score to the trend of the actual binding energy, and improving the accuracy of virtual screening;

[0124] After energy correction:

[0125] ;

[0126] ;

[0127] In the formula, is the number or index of the pharmacophore feature point, is the quantitative factor of the first class of chemical microenvironment characteristics, such as pH, charge density, is the weighting coefficient corresponding to the first class of microenvironment characteristics, used to adjust the influence strength of different environmental factors on energy correction, is the local pH parameter near the pharmacophore point, is the local charge density near the pharmacophore point, is the influence weight of local pH on energy correction, is the influence weight of local charge density on energy correction;

[0128] S4.52, according to the spatial deviation between the matching points in the candidate molecule and the pharmacophore template points (the final determination after VR editing of the three-dimensional coordinates), a Gaussian type distance attenuation weight is constructed for each pharmacophore feature point (the pharmacophore feature point here is the final pharmacophore feature point set after VR editing of the initial functional feature point), based on the spatial deviation between the pharmacophore template points and the corresponding points in the candidate molecule, a Gaussian type spatial attenuation weight is constructed to measure the “geometric position matching degree”, by punishing the spatial deviation, only the candidate molecule with close geometric position to the template can obtain a high score, so that the scoring system is more in line with the spatial specificity requirements of molecular recognition, and the structural rationality and target selectivity of the screening result are improved;

[0129] wherein the Gaussian type distance attenuation weight is:

[0130] ;

[0131] In the formula, is the spatial Gaussian weight of the pharmacophore point (i.e. the Gaussian type distance attenuation weight), used to describe the attenuation brought by the spatial deviation, is the three-dimensional spatial distance deviation between the matching points of the candidate molecule and the pharmacophore template points, in angstrom (Å), For the spatial tolerance parameter, it represents the degree of spatial deviation allowed for the corresponding pharmacophore type (the larger the tolerance is, the higher the tolerance is), and the unit is angstrom (Å);

[0132] S4.53, for the directional pharmacophore feature (hydrogen bond, ion interaction), an angle-dependent weight factor is introduced, for the hydrogen bond, ion interaction, these pharmacophore features with directional constraints, an angle-dependent weight factor is introduced, which is used to evaluate the consistency between the interaction direction of the candidate molecule and the template direction, to ensure that the interactions with directional requirements (such as hydrogen bond must be close to linear direction) can be accurately modeled, so that the molecules that do not meet the interaction geometry requirements are automatically given a lower score, thereby improving the physicochemical accuracy of the pharmacophore template and the reliability of the screening;

[0133] wherein the angle-dependent weight factor is introduced is (used to describe the influence of directional deviation on matching quality):

[0134] ;

[0135] In the formula, is the included angle between the interaction direction of the candidate molecule and the direction of the template pharmacophore, is the optimal interaction angle, such as the hydrogen bond directionality is commonly 120° to 180°, is the angle tolerance parameter, which represents the allowed angle deviation range;

[0136] S4.54, the contribution coefficient , the spatial weight and the angle-dependent weight factor are combined and integrated, the multi-dimensional weighted score is calculated for each pharmacophore feature point, and the sum of the weighted scores of all pharmacophore feature points is taken as the energy-weighted matching score of the whole candidate molecule;

[0137] wherein the multi-dimensional weighted score is calculated :

[0138] ;

[0139] The overall energy-weighted matching score is:

[0140] ;

[0141] S4.6, according to the energy-weighted matching score of all candidate compounds from high to low, a preliminary screening molecule list is output.

[0142] S5, based on the actual position and range of the conformational stable pocket region, the preliminary screening molecule list is fine-molecularly docked, and the binding conformation and energy score of each molecule are obtained, which are used for screening the final candidate molecules;

[0143] In this embodiment, the binding conformation and energy score of each molecule are obtained, including the following steps:

[0144] S5.1, based on the preliminary screening molecule list, each candidate molecule in the preliminary screening molecule list is introduced into the docking module, and the three-dimensional grid of the conformation stable pocket region is taken as the constraint space;

[0145] S5.2, for each candidate molecule, conformation sampling and rotational translation optimization are performed to generate multiple sets of initial binding poses that can enter the pocket region, and collision detection and geometric accessibility check are performed on each pose to generate the checked conformation; Specifically, based on the internal conformational freedom of the candidate molecule (such as the rotation key, ring conformation and side chain flip, etc.), several different molecular conformations are generated, and in the conformation stable pocket region, systematic translation and rotation scanning are performed on each conformation in three-dimensional space, so that the molecule tries to enter the pocket in different positions and directions, while collision detection and geometric accessibility check are performed (when checking the potential binding conformation of each candidate molecule, first, collision detection is performed on each atom in the conformation and the pocket grid and the residues in the pocket, to determine whether the distance between atoms or the distance between atoms and the pocket boundary is lower than the preset minimum tolerance threshold, if there is overlap or serious approach, mark the conformation as un-enterable; Then, the geometric accessibility check is performed, that is, the connectivity analysis is performed on the path from the pocket entrance to each functional group of the molecule to ensure that each key group can reach its target binding position smoothly, if the path is blocked or there is a space closed area, the conformation is removed), remove the conformations that have serious spatial conflicts with the protein atoms, so as to obtain multiple sets of initial binding poses that can enter the pocket region, and ensure sufficient coverage of the docking search space;

[0146] S5.3, re-perform local energy minimization on the checked conformation, generate the final comprehensive binding energy score combining interaction strength, conformation stability and pocket space complementarity; Specifically, first, energy minimization is performed on the atoms of the molecule in a local range, including adjusting bond length, bond angle, dihedral angle and non-bonding interaction, to eliminate local stress and obtain the optimal conformation state; Then, the interaction energy between the molecule and the pocket residues is calculated, including van der Waals force, electrostatic interaction, hydrogen bond and hydrophobic interaction, while combining the conformation stability evaluation of the molecule itself, such as internal strain energy and flexible group adaptability; On this basis, the complementarity of the molecule in the pocket space is analyzed, that is, the matching degree of the molecular volume and the pocket groove, and the proximity of the key group position and the key residue in the pocket; Finally, the above factors are weighted according to the preset weight to generate the final comprehensive binding energy score of each candidate molecule, which is used for sorting and selecting the optimal binding conformation;

[0147] S5.4、According to the comprehensive binding energy score, all candidate molecules are sorted to generate a candidate molecule set, and the binding conformation of protein and ligand, key interactions and docking search space (including the size and position of Dock Box) are presented in three dimensions through the docking visualization module (this step is realized through the Dock Result module, and the binding conformation of protein and ligand and the docking search space are presented in three dimensions through the docking visualization module. The center position, size and direction of the docking search space can be adjusted by the user. The candidate conformations are sorted from low to high according to the binding energy and displayed in an overlay manner. The key interaction sites (including hydrogen bonds, salt bridges and hydrophobic interactions) are identified by color mapping. The conformation selection, saving, exporting and further optimization functions are provided to assist the user in interactive analysis and structure optimization), to assist the user in interactive structure analysis.

[0148] Embodiment 2: The present embodiment provides a virtual screening system for pharmacophore based on virtual reality, comprising 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 realize the virtual screening method for pharmacophore based on virtual reality as described in any one of the above embodiments.

[0149] The basic principles, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only preferred examples of the present application and are not intended to limit the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application.

Claims

1. A method for virtual screening of pharmacophores based on virtual reality, characterized in that, The method comprises the following steps: S1, inputting the protein structure and the compound library into the VR scene as screening objects through a molecular resource loading module; S2, analyzing the pocket region in the protein structure using a pocket recognition module, and performing three-dimensional observation and position calibration on the pocket region in the VR scene, for judging the actual position and range of the pocket region; wherein, based on the conformation integration mechanism, the conformation consistency of the pocket region is analyzed, and the spatial fusion of the pocket region is performed according to the conformation consistency analysis result, to form a conformation stable pocket region; S3, generating initial functional feature points based on the conformation stable pocket region using a pharmacophore analysis module, and performing interactive editing and correction on the spatial position, type and tolerance parameters of the pharmacophore in the VR environment to form a pharmacophore template; S4, performing a pharmacophore screening process based on the pharmacophore template, matching and scoring the candidate compounds one by one, and generating a preliminary screening molecule list according to the matching score ranking; S5, performing fine molecular docking on the preliminary screening molecule list based on the actual position and range of the conformation stable pocket region, to obtain the binding conformation and energy score of each molecule, for screening the final candidate molecules.

2. The virtual screening method of pharmacophore based on virtual reality according to claim 1, characterized in that: In S2, the pocket recognition module is used to analyze the pocket region in the protein structure, and the three-dimensional observation and position calibration of the pocket region in the VR scene are performed, comprising the following steps: S2.1, rotating, moving and zooming in on the protein through three-dimensional operations S2.2, performing a pocket recognition algorithm on the protein structure based on the pocket recognition module to generate potential binding sites, i.e. pocket regions, and marking the pocket regions in the VR scene in a highlighted manner; S2.3, performing position calibration on the identified pocket regions based on the geometric feature and spatial parameters of the protein structure.

3. The virtual screening method of pharmacophore based on virtual reality according to claim 2, characterized in that: In S2.2, the pocket recognition algorithm is performed on the protein structure based on the pocket recognition module to generate potential binding sites, i.e. pocket regions, comprising the following steps: S2.21, converting the atomic coordinates of the protein structure into a three-dimensional surface mesh model, calculating the spatial coordinates and neighborhood atomic information of each mesh point; S2.22, performing cavity scanning on the surface mesh of the three-dimensional surface mesh model, identifying the protein surface depression region and the cavity volume that can accommodate ligands, and calculating the geometric center, volume and opening direction of each cavity to form a preliminary candidate pocket region set; S2.23, calculating the geometric features of each candidate pocket region to form a feature vector of the candidate pocket; S2.24, evaluating the feature vector of the candidate pocket by constructing a preset scoring function, screening the pocket regions whose evaluation results exceed a threshold n, and marking them as potential binding sites; S2.25, based on the conformation integration mechanism, the conformation consistency of the pocket region generated in step S2.24 is analyzed, and the spatial fusion of the pocket region is performed according to the conformation consistency analysis result, to form a conformation stable pocket region.

4. The virtual screening method of pharmacophore based on virtual reality according to claim 3, characterized in that: In S2.25, the conformation stable pocket region is formed, comprising the following steps: Based on the pocket region generated in step S2.24, normal mode analysis is performed on the protein structure to obtain the dominant motion direction of the protein structure under the low-frequency eigenmode, and a plurality of lightweight alternative conformations are generated within a reasonable perturbation range along the principal mode to form a conformation set; For each pocket region, the projection of each conformation in the conformation set The pocket region is repositioned and its corresponding spatial occupancy region is calculated to obtain the projection result of the pocket region under different conformations, forming a conformation mapping pocket set. Conformational mapping pocket set is subjected to cross-conformational consistency analysis to generate conformational consistency scores ; According to the conformation consistency score with a preset threshold value Screening is performed, and candidate pocket regions satisfying are marked as conformationally stable candidate pockets; The conformationally stable pocket region is subjected to spatial fusion processing to obtain a fused conformationally stable pocket region.

5. The virtual screening method of pharmacophore based on virtual reality according to claim 4, characterized in that: The spatial fusion processing of the conformationally stable pocket region to obtain the fused conformationally stable pocket region involves the following steps: The conformation mapping pocket set is re-aligned in coordinates and normalized in scale; Based on the geometric center offset range of the conformation mapping pocket after coordinate alignment and scale normalization, a Boolean fusion strategy is performed on the boundary grid of each conformation pocket to fuse; The fused pocket region is voxelized and reconstructed, and the stable spatial volume common to each conformation is calculated to form a fused conformationally stable pocket region.

6. The virtual reality-based pharmacophore virtual screening method according to claim 1, wherein: In S3, a pharmacophore template is formed, including the following steps: S3.1, generating initial pharmacophore feature points based on the protein local structure, key residue distribution and pocket shape by the pharmacophore analysis module; S3.2, displaying the initial pharmacophore feature points near the pocket region with different markers and superimposing them on the protein structure for easy observation of their geometric relationship with the pocket; S3.3, checking the distance, direction and geometric layout between the pharmacophore points and the protein residues; S3.4, after the inspection is completed, the pharmacophore is saved as a pharmacophore template.

7. The virtual reality-based pharmacophore virtual screening method of claim 1, wherein: In S4, each candidate compound is matched and scored, and an initial screening molecule list is generated according to the matching score ranking, including the following steps: S4.1, input the pharmacophore template as the matching rule into the screening module; S4.2, set the screening conditions according to the types and tolerances of the pharmacophore feature points; S4.3, pre-filtering based on the compound library information, output the candidate compound set after pre-filtering; S4.4, traverse each molecule in the candidate compound set, and calculate the geometric alignment of the pharmacophore feature points between the three-dimensional structure of the molecule and the pharmacophore template; S4.5, compare the matching degree between the functional group position of the compound and the pharmacophore template, and perform multi-dimensional energy-weighted pharmacophore matching scoring to generate energy-weighted matching scores; S4.6, sort the energy-weighted matching scores of all candidate compounds from high to low, and output the initial screening molecule list.

8. The virtual screening method of pharmacophore based on virtual reality according to claim 7, characterized in that: In S4.5, multi-dimensional energy-weighted pharmacophore matching scoring is performed, including the following steps: S4.51, constructing the base energy contribution coefficient for different pharmacophore points and energy-correcting the base energy contribution coefficient based on the protein chemical microenvironment to generate the energy-corrected contribution coefficient ; S4.52, according to the spatial deviation between the matching points in the candidate molecule and the pharmacophore template points, a Gaussian type distance attenuation weight is constructed for each pharmacophore feature point; S4.53, for directional pharmacophore features, an angle-dependent weight factor is introduced to evaluate the consistency between the interaction direction of the candidate molecule and the direction of the pharmacophore template; S4.54, the contribution coefficient The spatial weight and the angle-dependent weight factor are jointly integrated, the multi-dimensional weighted score is calculated for each pharmacophore feature point, and the weighted scores of all pharmacophore feature points are summed up as the energy-weighted matching score of the whole candidate molecule.

9. The virtual screening method of pharmacophore based on virtual reality according to claim 1, characterized in that: In S5, the binding conformation and energy score of each molecule are obtained, including the following steps: S5.1, based on the initial screening molecule list, each candidate molecule in the initial screening molecule list is imported into the docking module, and the three-dimensional grid of the conformationally stable pocket region is used as the constraint space; S5.2, performing conformational sampling and rotational-translation optimization on each candidate molecule to generate multiple sets of initial binding poses that can enter the pocket region, and performing collision checking and geometric accessibility checking on each pose to generate checked conformations; S5.3, re-performing local energy minimization on the checked conformations, and combining interaction strength, conformational stability and pocket space complementarity to generate final comprehensive binding energy scores; S5.4, ranking all candidate molecules according to the comprehensive binding energy scores to generate a candidate molecule set, and presenting the protein and ligand binding conformations, key interactions and docking search space in a three-dimensional manner through a docking visualization module.

10. A virtual screening system based on virtual reality pharmacophore comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor executes a computer program to implement the virtual screening method of a pharmacophore based on virtual reality as claimed in any one of claims 1-9.

Images