Protein-ligand molecule analysis system and analysis method

Abstract

The application discloses a protein-ligand molecule analysis system and an analysis method. The protein-ligand molecule analysis system comprises a storage module, a preprocessing module, an analysis simulation module and a retrieval module. The storage module is used for storing automatically collected crystal structures and user-initiated uploaded structure data. The preprocessing module is used for data classification and preprocessing of the stored structure data to obtain re-docking conformations. The analysis simulation module is used for molecular dynamics simulation based on the re-docking conformations. The retrieval module is used for inputting a target protein or a target ligand molecule and searching at least a molecular dynamics simulation result corresponding to the target protein or the target ligand molecule. The visualization module is used for dynamically displaying at least the molecular dynamics simulation result of the target protein or the molecular dynamics simulation result of the target ligand molecule. The application can be dynamically displayed, and the free energy change of the conformation of the molecule under different binding paths can be intuitively understood, thereby providing a new direction and strategy for drug research and development.

Description

Technical Field

[0001] This invention relates to the field of bioinformatics technology, and in particular to a protein-ligand molecular analysis system and method. Background Technology

[0002] With the rapid development of bioinformatics and computational biology, protein structure research has ushered in unprecedented opportunities. Currently, researchers typically obtain protein crystal structures from protein data banks (PDBs) and conduct in-depth studies using molecular dynamics simulations (MDs). MD simulations can reproduce the motion state of proteins under physiological conditions, revealing their conformational changes and internal dynamic behavior. This is crucial for understanding protein function and its interaction mechanisms with ligands, and can significantly improve drug screening efficiency and shorten the development cycle. However, when using PDB databases, static crystal structures are mostly observed using Pymol tools, making it difficult to present the dynamic binding process between proteins and ligands. If medicinal chemists and others want to observe these dynamic processes, they need to use a large amount of computational resources and complex software for simulation and result interpretation, greatly increasing the complexity of research and application, and hindering further improvements in research efficiency. Summary of the Invention

[0003] The purpose of this application is to overcome the shortcomings of the prior art and provide a protein-ligand molecular analysis system and method.

[0004] In a first aspect, this application provides a protein-ligand molecular analysis system, the protein-ligand molecular analysis system comprising: The storage module is used to store the structure data of automatically collected crystal structures and structures actively uploaded by users; The preprocessing module is used to classify and preprocess the stored structural data to obtain the reconnection conformation. An analysis and simulation module is used to perform molecular dynamics simulations based on the re-docking conformation; The retrieval module is used to input a target protein or target ligand molecule and at least search for the molecular dynamics simulation results corresponding to the target protein or target ligand molecule; A visualization module is used to dynamically display at least the molecular dynamics simulation results of the target protein or the molecular dynamics simulation results of the target ligand molecule.

[0005] Optionally, the preprocessing module includes: A classification unit is used to classify the stored structured data; The protein structure preprocessing unit is used to preprocess the protein structures in the classified structural data. The ligand molecule preprocessing unit is used to preprocess the ligand molecules in the classified structural data; The re-docking processing unit is used to obtain a conformation group based on the pretreated protein structure and the pretreated ligand molecules, and select the conformation with the smallest root mean square deviation from the conformation group as the re-docking conformation.

[0006] Optionally, the analysis and simulation module includes: A molecular dynamics simulation unit is used to perform molecular dynamics simulations based on the re-docking conformation to obtain trajectory files and topology files; The analysis unit is used to analyze the trajectory file and the topology file to obtain the stability of the ligand molecule, the stability of the protein structure, the hydrogen bonding between the ligand molecule and the protein structure, the free energy of the ligand molecule binding to the protein, and the free energy potential diagram of the ligand molecule and the protein structure.

[0007] Optionally, the preprocessing module further obtains the reconnected protein-compound complex structure or protein cavity structure; the retrieval module is further used to search for the protein-compound structure or protein cavity structure corresponding to the target protein or the target ligand molecule; the visualization module is further used to translate, rotate, zoom in, zoom out, or click on the protein-compound structure or protein cavity structure corresponding to the target protein and the target ligand molecule for static display or dynamic simulation.

[0008] Optionally, the protein-ligand molecular analysis system further includes: The system includes a customization module for submitting custom protein-compound structures; a preprocessing module for preprocessing the custom protein-compound structures; an analysis and simulation module for performing molecular dynamics simulations on the preprocessed custom protein-compound structures; and a visualization module for dynamically displaying the molecular dynamics simulation results of the custom protein-compound structures and statically displaying the preprocessed custom protein-compound structures.

[0009] Secondly, this application also provides a protein-ligand molecule analysis method, the protein-ligand molecule analysis method comprising: Acquire and store the structure data of automatically collected crystal structures and structures actively uploaded by users; The stored structural data is classified and preprocessed to obtain the reconnection conformation; Molecular dynamics simulations were performed based on the re-docking conformation. Input the target protein or target ligand molecule, and search for the corresponding molecular dynamics simulation results; The molecular dynamics simulation results of the target protein or the target ligand molecule are dynamically displayed.

[0010] Optionally, the step of classifying and preprocessing the stored structural data to obtain the reconnection conformation includes: The stored structured data is classified; Preprocess the protein structures in the classified structural data; Preprocessing of ligand molecules in the classified structural data; Based on the pretreated protein structure and the pretreated ligand molecules, a conformational group was obtained, and the conformation with the smallest root mean square deviation was selected from the conformational group as the re-docking conformation.

[0011] Optionally, after performing molecular dynamics simulations based on the re-docking conformation, a trajectory file and a topology file are obtained; after performing molecular dynamics simulations based on the re-docking conformation, the process further includes: The trajectory file and the topology file are analyzed to obtain the stability of the ligand molecule, the stability of the protein structure, the hydrogen bonding between the ligand molecule and the protein structure, the free energy of the ligand molecule binding to the protein, and the free energy potential diagram of the ligand molecule and the protein structure.

[0012] Optionally, after classifying and preprocessing the stored structural data, the reconnected protein-compound complex structure or protein cavity structure is obtained; the protein-ligand molecule analysis method further includes: Search for the protein-compound structure or protein cavity structure corresponding to the target protein or the target ligand molecule; translate, rotate, enlarge, shrink, or click on the protein-compound structure or protein cavity structure corresponding to the target protein or the target ligand molecule for static display or dynamic simulation.

[0013] Optionally, the protein-ligand molecular analysis method further includes: Submit custom protein-compound structures; Preprocess custom protein-compound structures; Molecular dynamics simulations were performed on the pre-processed custom protein-compound structures; The molecular dynamics simulation results of custom protein-compound structures are dynamically displayed, and the preprocessed custom protein-compound structures are statically displayed.

[0014] This application provides a protein-ligand molecule analysis system and method. It acquires structural data from automatically collected crystal structures and actively uploaded structures, classifies and preprocesses the structural data, and then performs molecular dynamics simulations. After inputting a target protein or ligand molecule, the system dynamically displays the molecular dynamics simulation results. Through free energy potential diagrams, it provides a visual understanding of the free energy changes of molecules under different binding pathways and conformations, identifying key binding pathways and energy barriers. Through searching, it can find compounds with similar geometric features and their bound proteins in a database, thereby identifying potential multi-target or off-target risks and providing new directions and strategies for drug development.

[0015] To make the above-mentioned features and advantages of the invention more apparent and understandable, specific embodiments are described below, and detailed descriptions are provided in conjunction with the accompanying drawings. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a structural block diagram of a protein-ligand molecular analysis system provided in one embodiment of this application.

[0018] Figure 2 This is a structural block diagram of the preprocessing module in a protein-ligand molecular analysis system provided in one embodiment of this application.

[0019] Figure 3 This is a structural block diagram of the analysis simulation module in a protein-ligand molecular analysis system provided in one embodiment of this application.

[0020] Figure 4 This is a structural block diagram of a protein-ligand molecular analysis system provided in another embodiment of this application.

[0021] Figure 5 This is a flowchart of a protein-ligand molecular analysis method provided in another embodiment of this application.

[0022] Figure 6 This is a flowchart of step S20 in a protein-ligand molecular analysis method provided in another embodiment of this application. Detailed Implementation

[0023] To make the objectives and technical solutions of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the described embodiments of this application without creative effort are within the scope of protection of this application.

[0024] In one embodiment, see Figure 1 This application provides a protein-ligand molecule analysis system, which may include: a storage module 10, a preprocessing module 20, an analysis simulation module 30, a retrieval module 40, and a visualization module 50; wherein, the storage module 10 is used to store structural data of automatically collected crystal structures and structures actively uploaded by users; the preprocessing module 20 is used to classify and preprocess the stored structural data to obtain the re-docking conformation; the analysis simulation module 30 is used to perform molecular dynamics simulation based on the re-docking conformation; the retrieval module 40 is used to input a target protein or target ligand molecule and at least search for the molecular dynamics simulation results corresponding to the target protein or the target ligand molecule; the visualization module 50 is used to dynamically display at least the molecular dynamics simulation results of the target protein or the target ligand molecule.

[0025] As an example, automatic collection mainly involves downloading protein-ligand molecules and cavity protein structures with already resolved crystal structures from the open-source PDB crystal database. After downloading them by category according to the protein gene name, they are stored in a temporary folder with the same protein gene name in storage module 10 for subsequent preprocessing.

[0026] As an example, proactive uploads primarily target popular drug targets whose crystal structures have not been resolved, including marketed drugs and molecules reported in literature / patents. This data is mainly collected manually and can be plotted using Chemdraw software, with protein-molecule binding structures obtained through molecular docking technology. The molecular docking technology discussed here can be found in the applicant's invention, the heatmap docking method (software registration number 2023SR036433), mentioned later; it will not be repeated here. Most of the structures proactively uploaded by users are docking structures, but crystal structures are also acceptable. Storage module 10 stores the structural data of automatically collected crystal structures and the structural data of structures proactively uploaded by users.

[0027] As an example, please combine Figure 1 See Figure 2The preprocessing module 20 may include: a classification unit 201, a protein structure preprocessing unit 202, a ligand molecule preprocessing unit 203, and a re-docking processing unit 204; wherein, the classification unit 201 is used to classify the stored structural data; the protein structure preprocessing unit 202 is used to preprocess the protein structures in the classified structural data; the ligand molecule preprocessing unit 203 is used to preprocess the ligand molecules in the classified structural data; and the re-docking processing unit 204 is used to obtain a conformation group based on the preprocessed protein structures and preprocessed ligand molecules, and select the conformation with the smallest root mean square deviation (RMSD) from the conformation group as the re-docking conformation.

[0028] As an example, the specific method by which classification unit 201 classifies the stored structural data can be as follows: First, it subdivides the data according to protein families and different subfamilies; then, it categorizes the data by the mutation types of protein residue sequences within each subfamilies, dividing them into wild-type and various specific mutant types. Therefore, the classification for proteins (with or without molecules) is protein family + subfamilies + wild-type / mutant type. Actively uploaded, unparsed, self-uploaded structures are also classified according to this category.

[0029] As an example, the specific method for preprocessing protein structures in the classified structural data by the protein structure preprocessing unit 202 can be as follows: For the protein structures obtained from the structural data, PDBFixer software can be used for preprocessing, including completing missing amino acids, adding hydrogen atoms and protonation, and removing water and solvent molecules to obtain protein structures that conform to the physiological environment. Simultaneously, the residue sequences within a 5 Å range surrounding the compound in the protein structure can be extracted and stored along with the protein sequence as corresponding PDB protein files. The path of these protein files is then stored in the protein seq and 5ÅResidues information columns of a data table created in a database (e.g., a PostgreSQL database). The PDB ID, along with the protein family name, wild / mutant type, UniportID, and species information, is also recorded in the PDBID, Family, Model Type, UniportID, and Species columns. This allows users to quickly compare existing crystals to find crystal structures with binding pockets similar to the target protein.

[0030] As an example, the specific method for preprocessing ligand molecules in the classified structural data by the ligand molecule preprocessing unit 203 is as follows: The three-dimensional coordinate information of the ligand molecules is extracted from the crystal structure, and hydrogen atoms are added and protonated to the ligand molecules using Open Babel software. Then, the SMILES format representation of the ligand molecules is stored in the ligand data column of the PostgreSQL database table, indexed by the corresponding PDB ID. Simultaneously, the Morgan fingerprint of the molecule is calculated using the GetMorganFingerprint() function in RDKit, and the USRCAT value of the ligand molecule is obtained using the GetUSRCAT() function. Subsequently, the similarity and geometric similarity differences between the user-input ligand molecules and existing ligand molecules in the protein-ligand molecule analysis system of this application can be compared.

[0031] It should be noted that the PostgreSQL database used for storage in this application can all be the storage module 10 of this application.

[0032] As an example, the re-docking processing unit 204 is used to obtain a conformation group based on the pre-treated protein structure and the pre-treated ligand molecules, and to select the conformation with the smallest root mean square deviation from the conformation group as the re-docking conformation. The specific method is as follows: Due to the limitations of crystal structure analysis methods, there are some interatomic conflicts between the compound and protein residues, or some groups may have incorrect conformations due to unclear analysis. Existing kinetic trajectory databases often use crystal structures directly for kinetic simulation, ignoring the impact of these errors, which may lead to incorrect compound-protein binding modes. Therefore, taking into full account the impact of this part, this application adds a re-docking step to reduce this impact. Specifically, the applicant's heatmap docking method (software registration number 2023SR036433) can be used. First, the protein structure and ligand molecular structure processed in the preprocessing step are used as input structures. Then, the point-to-point coordinate addition method and rotation search method are used to realize the molecular placement pocket operation. Then, the conformation group in the crystal structure is obtained by performing local energy optimization, and the RMSD between the conformation in the conformation group and the conformation of the ligand in the initial crystal is calculated. The conformation with the lowest RMSD is selected from the conformation group as the re-docking conformation, which is then used as the input structure for dynamic simulation. At the same time, the re-docking structure path is stored in the complex data column of the data table.

[0033] As an example, please combine Figure 1 and Figure 2 See Figure 3The analysis and simulation module 30 may include a molecular dynamics simulation unit 301 and an analysis unit 302. The molecular dynamics simulation unit 301 is used to perform molecular dynamics simulations based on the re-docking conformation to obtain trajectory files and topology files, etc. The analysis unit 302 is used to analyze the trajectory files and topology files to obtain the stability of ligand molecules, the stability of protein structures, the hydrogen bonding between ligand molecules and protein structures, the free energy of ligand molecules binding to proteins, and the free energy potential diagram of ligand molecules and protein structures, etc.

[0034] As an example, the analysis and simulation module 30 can also customize the selection of system characteristic variables such as the distance between specific atoms (domains), RMSD, number of atomic contacts, angle, dihedral angle, gyro radii, principal components, or select default parameters to calculate and draw a free potential landscape using internal scripts. Through the drawn two-dimensional free potential landscape, the differences in protein conformational change paths introduced by different ligands binding to the same protein or the same ligand binding to different proteins, as well as the energy magnitude between different paths, can be observed. At the same time, by clicking on the data points in the figure, the binding conformation of the protein and ligand at that point can be displayed in real time, deepening the user's understanding of the drug's efficacy, and the relevant information can also be applied to drug design.

[0035] As an example, the visualization module 50 is mainly used for displaying the dynamic trajectory animation of the target protein (with / without ligand molecules, i.e., with or without ligand molecules), comparing multi-window trajectories with / without ligand molecules, and displaying calculation result parameters. Specifically, the visualization module 50 can include a dynamic trajectory animation playback unit (not shown), a multi-window comparison unit (not shown), and a calculation result parameter display unit (not shown). For the dynamic trajectory animation, clicking the dynamic trajectory animation playback unit will play the simulation of the protein-compound or cavity protein at the corresponding time period. Clicking the multi-window comparison unit will compare the simulated conformation of the target protein-ligand molecule with that of the target protein in its cavity state, helping to explore the influence of compound binding on the conformation. The calculation result parameter display unit is mainly used to display RMSD, RMSF (Root Mean Square Fluctuation), hydrogen bond binding ratio, free energy calculation results, and free energy potential plot results. Clicking on each unit will display the corresponding results retrieved from the corresponding columns of the data tables in the corresponding PostgreSQL database.

[0036] As an example, for the free potential map, you can click on the highlighted extreme points in the map. After clicking, the visualization module 50 will jump to the protein (with / without ligand molecules) corresponding to that point's coordinates, and compare the free energy differences between different coordinate points in the free potential map. That is, after clicking, the visualization module 50 can jump to the simulation results of the cavity structure of the protein without ligand molecules, or the simulation results of the complex structure of the protein with ligand molecules, and compare the free energy differences between different coordinate points in the free potential map.

[0037] As an example, the search module 40 can input kinase, inhibitor name, PDBID, similar chemical structures, geometries, or pharmacophores to perform a search.

[0038] As an example, target protein searches can be performed in the search module 40. Directly searching for the protein gene name or PDB ID will match the PostgreSQL protein database and redirect to the corresponding kinetic trajectory and analysis results, displayed through the visualization module 50. Searching for the complete amino acid sequence of the protein or the 5 Å sequence surrounding the ligand will perform a search and match in the PostgreSQL protein database. After the search, options for 40%, 50%, 60%, 70%, 80%, 90%, and 100% similarity results will be output for the user to choose from. After the user selects a similarity result, the system will automatically link to the corresponding kinetic trajectory and analysis results, displayed through the visualization module 50. If the user draws a structural formula online or inputs a SMILE format in the search module 40, the system will first calculate the molecule's Morgan fingerprint and USRCAT value, comparing them with data in the PostgreSQL small molecule database. By comparing the Tanimoto similarity coefficient and the USRCAT value calculated using the GetUSRScore function, similarity results in the 50%-100% range will be output for the user to choose from. After the user selects a similarity result, the system will link to the corresponding kinetic trajectory and analysis results, displayed through the visualization module 50.

[0039] As an example, the preprocessing module 20 also obtains the reconnected protein-compound complex structure or protein cavity structure; the retrieval module 40 is also used to search for the protein-compound structure or protein cavity structure corresponding to the target protein or target ligand molecule; the visualization module 50 is also used to translate, rotate, zoom in, zoom out, or perform static display or dynamic simulation of the protein-compound structure or protein cavity structure corresponding to the target protein or target ligand molecule after clicking. Specifically, after clicking on the protein-compound structure, the visualization module 50 displays the types of protein-compound interaction forces, etc., to facilitate further understanding of the structure.

[0040] As an example, in addition to displaying the above-mentioned information, visualization module 50 also statically displays the compound's structural formula, common name, trade name, molecular formula, molecular weight, and IC50. 50 (half maximal inhibitory concentration) / EC 50 (half maximal effective concentration) and specific information about the structural origin, etc.

[0041] In another embodiment, please refer to Figures 1 to 3 See Figure 4 The protein-ligand molecular analysis system also includes: a customization module 60, which is used to submit custom protein-compound structures; a preprocessing module 20, which is used to preprocess the custom protein-compound structures; an analysis and simulation module 30, which is used to perform molecular dynamics simulations on the preprocessed custom protein-compound structures; and a visualization module 50, which is used to dynamically display the molecular dynamics simulation results of the custom protein-compound structures, and to statically display and analyze the preprocessed custom protein-compound structures.

[0042] As an example, the customization module 60 can submit at least one of the following: user input file, force field selection, simulation duration, NPT / NVT system selection, and reaction coordinate selection. It can also input an email address for receiving simulation results.

[0043] As an example, the storage module 10, preprocessing module 20, analysis simulation module 30, retrieval module 40, and visualization module 50 can all be integrated into the Biotrajectory online website as the protein-ligand molecular analysis system of this application.

[0044] As an example, the protein-ligand molecular analysis system of this application can be built based on the Python language. Specifically, Python language can be used to build dedicated scripts for system construction, task submission (e.g., retrieval tasks or customized tasks), analysis simulation, and visualization of results, enabling rapid large-scale molecular dynamics simulation processes.

[0045] The aforementioned protein-ligand research system acquires structural data from automatically collected crystal structures and user-uploaded structures. After classifying and preprocessing the structural data, molecular dynamics simulations are performed. Upon inputting a target protein or ligand molecule, the system dynamically displays the molecular dynamics simulation results. Through free potential diagrams, the system provides an intuitive understanding of free energy changes in molecules under different binding pathways and conformations, identifying key binding pathways and energy barriers. Searching allows the system to find compounds with similar geometric features and their bound proteins in the database, thereby identifying potential multi-target or off-target risks and providing new directions and strategies for drug development.

[0046] In yet another embodiment, please refer to Figures 1 to 4 See Figure 5 This application also provides a protein-ligand molecular analysis method, which may include the following steps: S10~S50.

[0047] S10: Acquire and store the structure data of automatically collected crystal structures and structures actively uploaded by users.

[0048] S20: Perform data classification and preprocessing on the stored structured data to obtain the reconnection conformation.

[0049] S30: Molecular dynamics simulations based on re-docking conformations.

[0050] S40: Input the target protein or target ligand molecule and search for the corresponding molecular dynamics simulation results.

[0051] S50: Dynamically display the molecular dynamics simulation results of the target protein or the target ligand molecule.

[0052] In the protein-ligand molecular analysis method of this application, sufficient data sources are ensured by acquiring automatically collected crystal structures and user-uploaded structures. The structural data is classified and screened through classification and preprocessing of protein and ligand molecular structures. Molecular dynamics simulations and free energy potential diagrams provide a visual understanding of free energy changes in molecules under different binding pathways and conformations, identifying key binding pathways and energy barriers. This allows for the reproduction of protein motion under physiological conditions, facilitating in-depth exploration of protein behavior and interaction details in dynamic processes. Furthermore, a search function in the database can locate compounds with similar geometric features and their bound proteins, identifying potential multi-target or off-target risks. This provides new directions and strategies for drug development, facilitating research in drug development, protein function studies, and other fields, and promoting the progress and translation of related research findings.

[0053] As an example, the protein-ligand molecular analysis method in this embodiment may be, but is not limited to, based on the above. Figures 1 to 4 The protein-ligand molecular analysis system in the embodiments is used for execution.

[0054] In step S10, please refer to Figure 5 In step S10, the structure data of automatically collected crystal structures and user-uploaded structures are acquired and stored.

[0055] As an example, automatic collection mainly involves downloading protein-molecule and cavity protein structures with resolved crystal structures from the open-source PDB crystal database. The structures are then categorized and downloaded according to the protein gene name and stored in a temporary folder with the same protein gene name for subsequent structure preprocessing.

[0056] As an example, proactive uploading primarily targets popular drug targets whose crystal structures have not been resolved, including marketed drugs and molecules reported in literature / patents. This data is mainly collected manually, plotted using Chemdraw software, and then used to obtain protein-molecule binding structures using molecular docking technology. Most of the structures proactively uploaded by users are docking structures, but crystal structures are also possible. In step S10, the acquired and stored structures are the automatically collected crystal structure data and the structure data of the user-proactively uploaded structures.

[0057] In step S20, please refer to Figure 5 In step S20, the stored structural data is classified and preprocessed to obtain the reconnection conformation.

[0058] As an example, please combine Figure 5 See Figure 6 Step S20 may include the following steps: S201~S204.

[0059] S201: Classify the stored structured data.

[0060] S202: Preprocess the protein structures in the classified structural data.

[0061] S203: Preprocess the ligand molecules in the classified structural data.

[0062] S204: Based on the pretreated protein structure and the pretreated ligand molecules, a conformational group is obtained, and the conformation with the smallest root mean square deviation is selected from the conformational group as the re-docking conformation.

[0063] As an example, in step S201, the automatically collected data of resolved crystal structures (with or without molecules) are further subdivided according to protein families and different subfamilies. Then, they are tagged and classified according to the mutation types of protein residue sequences within each subfamilies, resulting in wild-type and various specific mutant types. Therefore, for proteins (with or without molecules), they can be classified as protein family + subfamilies + wild-type / mutant. For actively uploaded structures, they are also classified using the same method as protein family + subfamilies + wild-type / mutant.

[0064] As an example, in step S202, preprocessing the protein structures in the classified structural data may include completing missing amino acids, adding hydrogen atoms and protonation, removing water molecules, and removing solvent molecules to obtain protein structures that conform to physiological environments. Simultaneously, the residue sequence within a 5 Å radius surrounding the compound in the protein structure is extracted and stored along with the protein sequence in the protein database of the background database. For example, it can be stored in the protein seq and 5ÅResidues information columns of the data table created by the protein database, and its PDB ID, along with the protein family name, wild / mutant type, Uniport ID, and species information, is also recorded in the PDB ID, Family, Model Type, Uniport ID, and Species columns. This allows users to quickly compare existing crystals to find crystal structures with binding pockets similar to the target protein. Using the PDB ID, protein gene name, and wild / mutant type as indexes, users can quickly compare existing crystals to find crystal structures with binding pockets similar to the target protein.

[0065] As an example, the preprocessing can be implemented using PDB Fixer software, and the backend database can be a PostgreSQL database, which may include: a protein database, a small molecule database, and a kinetic trajectory database.

[0066] As an example, in step S203, the three-dimensional coordinate information of the ligand molecule is obtained from the crystal structure, and the molecule is subjected to hydrogen atom addition and protonation processing using Open Babel software. The preprocessed molecule's SMILES format representation is then stored in the small molecule database of a PostgreSQL database, with the corresponding PDB ID set as the index. Simultaneously, the molecular similarity and geometric similarity of the molecule are calculated. The GetMorganFingerprint() function in RDKit can be used to calculate the molecule's Morgan fingerprint, and the GetUSRCAT() function can be used to obtain the molecule's USRCAT value, which is stored in the PostgreSQL database. This value can then be used to compare the similarity and geometric similarity between the user-input molecule and existing molecules in the database, helping to identify potential multi-target or off-target risks.

[0067] As an example, due to limitations in crystal structure analysis methods, there may be some interatomic conflicts between compounds and protein residues, or incorrect conformations of some functional groups due to unclear analysis. Existing kinetic trajectory databases often use crystal structures directly for kinetic simulations, ignoring the impact of these errors, which may lead to incorrect compound-protein binding patterns. Adding a re-docking step can reduce this impact.

[0068] As an example, in step S204, the applicant's invention, the heatmap docking method (software registration number 2023SR036433), can be used. First, the pre-processed protein structure and ligand molecular structure are used as input structures for re-docking. The point-to-point coordinate addition method and rotation search method are used to realize the molecular placement pocket operation. Then, the conformation group in the crystal structure is obtained through local energy optimization, and the RMSD between the conformation and the conformation of the ligand in the initial crystal is calculated. The conformation with the lowest RMSD is selected from the conformation group as the re-docking conformation, which is then used as the input structure for dynamic simulation. At the same time, the re-docking structure path is stored in the complex data column of the data table.

[0069] In step S30, please refer to Figure 5 The S30 step in the process involves molecular dynamics simulations based on the re-docking conformation.

[0070] As an example, in step S30, after performing molecular dynamics simulation based on the re-docking conformation, files such as trajectory files and topology files are obtained; after performing molecular dynamics simulation based on the re-docking conformation, the method further includes: analyzing the trajectory files and the topology files to obtain the stability of the ligand molecules, the stability of the protein structure, the hydrogen bonding between the ligand molecules and the protein structure, the free energy of the ligand molecules binding to the protein, and the free energy potential diagram of the ligand molecules and the protein structure.

[0071] Specifically, step S30 may include the following steps: S301~S304.

[0072] S301: A dynamic system built using the Python language.

[0073] S302: Input the reconnected conformation into the dynamic system, perform molecular dynamics simulation, and obtain trajectory files, topology files, and other files.

[0074] S303: MDtraj tool is used to analyze trajectory files and topology files to obtain the stability of ligand molecules, the stability of protein structures, the hydrogen bonding between ligand molecules and protein structures, and the free energy of ligand-protein binding. The analysis results are then stored in the background database.

[0075] S304: Based on trajectory files and topology files, construct a free potential map, output the corresponding protein-ligand conformation files, and store them in the background database.

[0076] As an example, we first use Python to build a dynamic system, which can quickly perform large-scale molecular dynamics simulations, solving the problems of cumbersome and inefficient system simulation parameter setup when performing molecular dynamics simulations.

[0077] As an example, the re-docking conformation obtained in step S204 is input into the kinetic system for kinetic simulation. The main script first calls the pdb2gmx tool in Gromacs (gmx) to process the input structure, obtaining the topology and trajectory files. Then, it calls the editconf tool to create a simulation box and adds the solvent and corresponding ions. Next, it sequentially calls the system minimization, heating, system pre-equilibrium, and formal kinetic simulation subscripts to perform three parallel 300ns simulations. It should be noted that the simulation time is not limited to the aforementioned 300ns; the simulation time can be adjusted according to actual needs.

[0078] As an example, if the process log file contains a specific field, such as "error," the process sub-script will send the file with an error message to the specified email address.

[0079] As an example, after the molecular dynamics simulation in step S302 is completed, in step S303, the simulated trajectory is exported by calling the trjconv command in gmx. Then, the obtained trajectory file and topology file are automatically read by the MDtraj tool. The RMSD, RMSF, and Baker_hubbard commands are called in sequence to calculate the stability results of ligands and proteins, as well as the formation and changes of hydrogen bonds between ligands and proteins. The results are then visualized using Matplotlib. At the same time, the mmgbsa command is called to calculate the binding free energy between ligand molecules and proteins. The above results are stored in the dynamic trajectory database of the PostgreSQL database, with the PDB ID as the index.

[0080] As an example, in addition to exporting the above parameters, such as in step S304, a free potential map can also be incorporated into the process. Based on trajectory files and topology files, a free potential map can be constructed to demonstrate and compare the energy differences in protein conformational transitions caused by different ligands. Specifically, for a specific research system, reaction coordinates (or multiple coordinates) are first specified as the dimensions of the free potential map during the simulation. For example, in the study of kinase systems, the RMSD of the ligand and the KE salt bridge distance of the kinase can be used as two-dimensional reaction coordinates. Then, the compute_distance, compute_displacements, compute_angles, and compute_dihedrals commands of the MDtraj tool are used to read the trajectory file and topology file, calculate the corresponding parameters, and after selecting the reaction coordinates, the frequency distribution of the sampling points is converted into free energy values ​​using statistical mechanical relationships. The statistical mechanical relationship can be a Boltzmann distribution, and the free energy value can be expressed as: in, For free energy, Boltzmann constant, For temperature, Reaction coordinates The probability distribution on.

[0081] Furthermore, the calculated free energy distribution is mapped to the reaction coordinate space, and a Matplotlib heatmap is used to represent the free energy distribution. The horizontal and vertical axes represent the reaction coordinates, and the vertical axis represents the free energy value, which is represented by color or height. Simultaneously, the reaction coordinates of the energy minimum (steady state) and energy barrier (transition state) points in the free energy potential map are input into the corresponding files of the kinetic trajectory data in the PostgreSQL database. Based on these reaction coordinates, the corresponding protein-ligand conformation files are output by reading the trajectory and topology files, and are also stored in the corresponding files of the kinetic trajectory database. By integrating the free energy potential map function, the changes in free energy during molecular interactions are presented in a dynamic and visual form. This allows for an intuitive understanding of the free energy changes of molecules under different binding pathways and conformations, identification of key binding pathways and energy barriers, and provides a deeper understanding of the mechanisms involved in molecular design and optimization, enhancing the practicality and intuitiveness of the technology.

[0082] As an example, the system's characteristic variables, such as the distance between specific atoms (domains), RMSD, number of atomic contacts, angle, dihedral angle, gyrorotation radius, principal components, or default parameters, can be customized and plotted using an internal script to calculate and draw a free potential map. The resulting two-dimensional free potential map allows observation of the differences in protein conformational change paths and the energy levels between different paths after different ligands bind to the same protein or the same ligand binds to different proteins. By clicking on data points in the map, the binding conformation of the protein and ligand at that point can be displayed in real time, deepening the user's understanding of drug efficacy and enabling the application of this information in drug design.

[0083] In step S40, please refer to Figure 5 In step S40, input the target protein or target ligand molecule and search for the corresponding molecular dynamics simulation results.

[0084] As an example, in step S40, you can enter the name of the kinase, inhibitor, PDB ID, similar chemical structures, geometries, or pharmacophores to perform a search.

[0085] As an example, when the name of the target protein is entered, it is matched in the background database to obtain the corresponding kinetic trajectory and analysis results; when the target protein sequence is entered, it is matched in the background database to output kinetic trajectories and analysis results with different similarities; when the SMILE format is entered and online plotting is performed, the molecular similarity and geometric similarity of the molecules are calculated and compared with the data in the background database to output the kinetic trajectory and analysis results with similarity in the 50%-100% range.

[0086] As an example, to facilitate medicinal chemistry experts or related practitioners in further understanding the protein-compound simulation results and reduce operational complexity, all simulation results stored in the PostgreSQL database can be integrated to build the Biotrajectory online website. Users can then use the website to search online and jump to a webpage to view the target protein-ligand kinetic trajectory and the visualization analysis of the results.

[0087] Specifically, when a user searches for a target protein on the website, directly entering the protein gene name and PDB ID will match the PostgreSQL protein database and redirect to the corresponding kinetic trajectory and analysis results. Entering the complete amino acid sequence of the protein or the 5 Å sequence surrounding the ligand will search the PostgreSQL protein database. After the search, options for 40%, 50%, 60%, 70%, 80%, 90%, and 100% similarity results will be provided for the user to choose from. After selecting the corresponding similarity result, the system will automatically link to the corresponding kinetic trajectory and analysis results. If the user draws a structural formula online or enters a SMILE format, the system will first calculate the molecule's Morgan fingerprint and USRCAT value, and compare them with data in the PostgreSQL small molecule database. By comparing the Tanimoto similarity coefficient and the USRCAT value calculated using the GetUSRScore function, similarity results in the 50%-100% range will be output for the user to choose from. After selecting the corresponding similarity result, the system will link to the kinetic trajectory and analysis results corresponding to the user's selection. By setting different search conditions as described above, users can obtain not only the binding trajectory of the target protein-ligand, but also information on proteins with similar binding pockets to the target protein and ligands with similar geometry to the target ligand. This can help users determine whether the target compound has off-target effects or multiple targets acting simultaneously.

[0088] As an example, after classifying and preprocessing the stored structural data, the reconnected protein-compound complex structure or protein cavity structure is obtained; the protein-ligand molecule analysis method may also include: searching for the protein-compound structure or protein cavity structure corresponding to the target protein or target ligand molecule; and performing static display or dynamic simulation of the protein-compound structure or protein cavity structure corresponding to the target protein or target ligand molecule by translating, rotating, enlarging, shrinking, or clicking.

[0089] As an example, in addition to displaying the above-mentioned information, the compound's structural formula, common name, trade name, molecular formula, molecular weight, and IC50 can also be statically displayed. 50(half maximal inhibitory concentration) / EC 50 (half maximal effective concentration) and specific information about the structural origin, etc.

[0090] As an example, the protein-ligand molecular analysis method may further include the following steps: submitting a custom protein-compound structure; preprocessing the custom protein-compound structure; performing molecular dynamics simulation on the preprocessed custom protein-compound structure; dynamically displaying the molecular dynamics simulation results of the custom protein-compound structure, and statically displaying the preprocessed custom protein-compound structure.

[0091] As an example, in addition to directly searching the data already available on the Biotrajectory website, users can also submit their own protein-compound structures for simulation. Specifically, users upload their input files, set parameters such as force field selection, simulation duration, and NPT / NVT series selection, and enter their email address to receive the results. The background program automatically reads the input files and parameters, runs the scripts in steps S20 and S30 for file processing, and returns the calculation time by reading the mdinfo file during the process. After the calculation is completed, the user is notified of the simulation completion status via email. Furthermore, users can view the simulation results visualization page on the webpage.

[0092] As an example, when uploading input files, users can also choose whether to include coordinates.

[0093] In one implementation, the crystal structure of the kinase domain of epidermal growth factor receptor (EGFR) (2ITO) can be selected. The inhibitory effect of the inhibitor (gefitinib) on the activity of the EGFR kinase domain crystal structure (2ITO) can be observed online, its kinetic trajectory can be observed, and a kinetic database of the inhibitor (gefitinib) can be constructed and related analyses can be performed. First, the PDBID of one of the EGFR kinase domain crystal structures is determined to be 2ITO, and its protein crystal structure is downloaded from the open-source database RCSB PDB. Then, the structure of the protein kinase 2ITO is sequence-completed through preprocessing operations, and the charge of its inhibitor (gefitinib) is calculated. Specifically, the crystal structure is downloaded from the open-source database website RCSB PDB and completed using PDB Fixer, and then the am1-bcc charge is calculated using gmx. Then, the preprocessed kinase (2ITO) and inhibitor (gefitinib) are subjected to three sets of 300 ns kinetic trajectory processing. Specifically, docking software, such as heatmap docking, is used to dock proteins and small molecules. The docking results can be set to 50 conformations. The conformation with the smallest RMSD compared to the initial structure is selected to ensure its stability. Then, the docked conformation is input into the kinetic system for three 300 ns molecular dynamics simulations. Further, the obtained trajectory and topology files are uploaded to a backend database. Based on this database, an online website, Biotrajectory, is built. Users can search by entering the name of a kinase or inhibitor, its similar chemical structure, or the geometry of a small molecule with PDB ID 2ITO. The backend data matches the corresponding structure and displays it on the webpage. Users can select the desired structure, and the webpage displays its 900 ns kinetic trajectory. Users can also choose to analyze the kinetic trajectory by selecting specific parameters and calling corresponding scripts to calculate and display the data results, such as RMSD and RMSF. Free energy can also be calculated, and the results are displayed on the webpage. Alternatively, by selecting the C-helix domain RMSD and ligand RMSD as characteristic parameters, the background will automatically process them and return a free energy potential map. Users can click on the points in the map to observe their specific binding conformations.

[0094] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0095] In the protein-ligand molecular analysis method of this application, the data can be refined and optimized through classification and preprocessing of protein and ligand molecular structures. By using the heatmap docking method to re-dock the preprocessed protein and ligand molecular structures, the influence of interatomic conflicts or incorrect conformations caused by the limitations of crystal structure analysis methods on subsequent dynamic simulations can be reduced, thereby improving the accuracy and reliability of the simulation. By constructing a free energy potential landscape, the changes in free energy during molecular interactions are presented in a dynamic and visual form, which can help users intuitively understand the changes in free energy of molecules under different binding pathways and conformations, identify key binding pathways and energy barriers, and provide a deeper understanding of the mechanisms for molecular design and optimization. By setting multiple search methods, the needs of different users can be met, reducing tediousness, and helping users to judge the off-target effects or multi-target interactions of target compounds.

[0096] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0097] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0098] Although this application has been disclosed above with reference to embodiments, it is not intended to limit this application. Anyone skilled in the art may make some modifications and refinements without departing from the spirit and scope of this application. Therefore, the scope of protection of this application shall be determined by the appended claims.

Claims

1. A protein-ligand molecular analysis system, characterized in that, The protein-ligand molecular analysis system includes: The storage module is used to store the structure data of automatically collected crystal structures and structures actively uploaded by users; The preprocessing module is used to classify and preprocess the stored structural data to obtain the reconnection conformation. An analysis and simulation module is used to perform molecular dynamics simulations based on the re-docking conformation; The retrieval module is used to input a target protein or target ligand molecule and at least search for the molecular dynamics simulation results corresponding to the target protein or target ligand molecule; A visualization module is used to dynamically display at least the molecular dynamics simulation results of the target protein or the molecular dynamics simulation results of the target ligand molecule.

2. The protein-ligand molecular analysis system according to claim 1, characterized in that, The preprocessing module includes: A classification unit is used to classify the stored structured data; A protein structure preprocessing unit is used to preprocess the protein structures in the classified structural data. The ligand molecule preprocessing unit is used to preprocess the ligand molecules in the classified structural data; The re-docking processing unit is used to obtain a conformation group based on the pretreated protein structure and the pretreated ligand molecules, and select the conformation with the smallest root mean square deviation from the conformation group as the re-docking conformation.

3. The protein-ligand molecular analysis system according to claim 2, characterized in that, The analysis and simulation module includes: The molecular dynamics simulation unit is used to perform molecular dynamics simulations based on the re-docking conformation to obtain trajectory files and topology files; The analysis unit is used to analyze the trajectory file and the topology file to obtain the stability of the ligand molecule, the stability of the protein structure, the hydrogen bonding between the ligand molecule and the protein structure, the free energy of the ligand molecule binding to the protein, and the free energy potential diagram of the ligand molecule and the protein structure.

4. The protein-ligand molecular analysis system according to claim 3, characterized in that, The preprocessing module also obtains the re-connected protein-compound complex structure or protein cavity structure; the retrieval module is also used to search for the protein-compound structure or protein cavity structure corresponding to the target protein or the target ligand molecule; the visualization module is also used to translate, rotate, zoom in, zoom out, or click on the protein-compound structure or protein cavity structure corresponding to the target protein and the target ligand molecule for static display or dynamic simulation.

5. The protein-ligand molecular analysis system according to claim 4, characterized in that, The protein-ligand molecular analysis system also includes: The customization module is used to submit custom protein-compound structures; the preprocessing module is also used to preprocess the custom protein-compound structures; the analysis and simulation module is also used to perform molecular dynamics simulations on the preprocessed custom protein-compound structures; the visualization module is also used to dynamically display the molecular dynamics simulation results of the custom protein-compound structures and statically display the preprocessed custom protein-compound structures (the customization module should also include the analysis module of claim 3 above).

6. A protein-ligand molecular analysis method, characterized in that, The protein-ligand molecular analysis method includes: Acquire and store the structure data of automatically collected crystal structures and user-uploaded structures; The stored structural data is classified and preprocessed to obtain the reconnection conformation; Molecular dynamics simulations were performed based on the re-docking conformation. Input the target protein or target ligand molecule, and search for the corresponding molecular dynamics simulation results; The molecular dynamics simulation results of the target protein or the target ligand molecule are dynamically displayed.

7. The protein-ligand molecular analysis method according to claim 6, characterized in that, The process of classifying and preprocessing the stored structural data to obtain the reconnection conformation includes: The stored structured data is classified; Preprocess the protein structures in the classified structural data; Preprocessing of ligand molecules in the classified structural data; Based on the pretreated protein structure and the pretreated ligand molecules, a conformational group was obtained, and the conformation with the smallest root mean square deviation was selected from the conformational group as the re-docking conformation.

8. The protein-ligand molecular analysis method according to claim 7, characterized in that, After performing molecular dynamics simulations based on the re-docking conformation, trajectory files and topology files are obtained; the molecular dynamics simulations based on the re-docking conformation further include: The trajectory file and the topology file are analyzed to obtain the stability of the ligand molecule, the stability of the protein structure, the hydrogen bonding between the ligand molecule and the protein structure, the free energy of the ligand molecule binding to the protein, and the free energy potential diagram of the ligand molecule and the protein structure.

9. The protein-ligand molecular analysis method according to claim 8, characterized in that, After data classification and preprocessing of the stored structural data, the reconnected protein-compound complex structure or protein cavity structure is obtained; the protein-ligand molecule analysis method further includes: Search for the protein-compound structure or protein cavity structure corresponding to the target protein or the target ligand molecule; translate, rotate, enlarge, shrink, or click on the protein-compound structure or protein cavity structure corresponding to the target protein or the target ligand molecule for static display or dynamic simulation.

10. The protein-ligand molecular analysis method according to claim 9, characterized in that, The protein-ligand molecular analysis method further includes: Submit custom protein-compound structures; Preprocess custom protein-compound structures; Molecular dynamics simulations were performed on the pre-processed custom protein-compound structures; The molecular dynamics simulation results of custom protein-compound structures are dynamically displayed, and the preprocessed custom protein-compound structures are statically displayed.

Images