Protein pocket detection system and method and storage medium

Abstract

The invention relates to a protein pocket detection system and method and a storage medium, and the detection system comprises a conformation generation module, a pocket detection module, a task scheduling and management module and a visualization module: the task scheduling and management module obtains to-be-detected protein data; the conformation generation module is used for generating a conformation track of the protein data by utilizing molecular dynamics simulation based on the protein data acquired by the task scheduling and management module; the pocket detection module is used for determining pocket data based on the conformation track generated by the conformation generation module; wherein the pocket data is used for representing the stable data of the spatial position and the geometric shape of the protein data in the change process of the conformation trajectory; and the visualization module is used for rendering and outputting the conformation track generated by the conformation generation module and the pocket data determined by the pocket detection module. The problem that in the prior art, key drug targets generated in the dynamic process of protein cannot be efficiently and conveniently found is solved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of protein detection, and in particular to a protein pocket detection system, method and storage medium. BACKGROUND

[0002] Proteins are the core executors of life activities, and their functions are closely related to their three-dimensional structures, especially to the "pockets" or "binding sites" that bind ligands. Traditional drug discovery and target recognition rely heavily on static protein structures obtained through techniques such as X-ray crystallography and nuclear magnetic resonance. However, this method has inherent limitations: proteins are dynamic in physiological environments, and their structures are always in constant motion and conformational changes. Many crucial drug targets, such as allosteric pockets or cryptic pockets, are usually closed or invisible in static snapshots, and only form or open at specific conformations during their dynamic changes. Therefore, the traditional pocket detection method relying on a single static structure has the risk of missing a large number of potentially important drug targets.

[0003] Currently, the complete dynamic pocket analysis process requires researchers to manually operate multiple independent software tools to sequentially complete simulation, trajectory processing, sampling, pocket calculation and visualization analysis, which is complex, time-consuming and prone to errors. For biological or chemical researchers without a computing background, it is difficult to overcome technical challenges such as configuring a supercomputing environment, writing job scripts, and managing data flow.

[0004] In summary, the prior art cannot efficiently and conveniently find the key drug targets generated by proteins in the dynamic process. SUMMARY

[0005] To solve the above technical problems or at least partially solve the above technical problems, the present disclosure provides a protein pocket detection system, method and storage medium.

[0006] The present disclosure provides a protein pocket detection system, which comprises a conformation generation module, a pocket detection module, a task scheduling and management module, and a visualization module: The task scheduling and management module acquires protein data to be detected and coordinates the execution of the conformation generation module, the pocket detection module and the visualization module; The conformation generation module generates conformation trajectories of the protein data based on the protein data acquired by the task scheduling and management module using molecular dynamics simulation in response to the scheduling of the task scheduling and management module; wherein the conformation trajectories are used to represent the structural change process of the protein data during the execution of biological functions; The pocket detection module determines pocket data based on the conformation trajectory generated by the conformation generation module in response to scheduling of the task scheduling and management module; wherein the pocket data is used to represent data of stable spatial position and geometric shape of the protein data in the change process of the conformation trajectory. The visualization module renders and outputs the conformation trajectory generated by the conformation generation module and the pocket data determined by the pocket detection module in response to scheduling of the task scheduling and management module.

[0007] Optionally, the task scheduling and management module generates a conformation trajectory of the protein data based on the protein data obtained by the task scheduling and management module, including: determining the data type of the protein data; determining a molecular dynamics simulation method based on the data type; generating the conformation trajectory by using the molecular dynamics simulation.

[0008] Optionally, the determination of the molecular dynamics simulation method based on the data type includes: when the data type is a first data type, determining the molecular dynamics simulation method as a first molecular dynamics simulation method; wherein the first data type is used to represent protein sequence data; when the data type is a second data type, determining the molecular dynamics simulation method as a second molecular dynamics simulation method; wherein the second data type is used to represent protein structure data, and the first molecular dynamics simulation method and the second molecular dynamics simulation method are different in structure.

[0009] Optionally, the pocket detection module determines pocket data based on the conformation trajectory generated by the conformation generation module, including: extracting a plurality of conformation samples from the conformation trajectory; simultaneously performing pocket detection on a plurality of the conformation samples to obtain a plurality of pocket detection results; performing clustering and / or screening processing on the pocket detection results to determine the pocket data.

[0010] Optionally, the conformation generation module further includes: generating topological data of the protein data based on the protein data obtained by the task scheduling and management module.

[0011] Optionally, the visualization module further includes: in response to selection and / or query operation on the pocket data, extracting selected pocket data from the pocket data; The selected pocket data is highlighted and output, and attribute information of the selected pocket data is output; wherein the attribute information comprises at least one of pocket average volume, drug forming score, hydrophobicity parameter, and appearance frequency.

[0012] Optionally, the visualization module further comprises: During output of the conformation trajectory generated by the conformation generation module, pocket data corresponding to a current output frame is identified, and the pocket data corresponding to the current output frame is synchronously highlighted and output.

[0013] The present disclosure provides a protein pocket detection method, applied to a protein pocket detection system, and the method comprises: Obtaining protein data to be detected; Based on the protein data, a conformation trajectory of the protein data is generated by using molecular dynamics simulation; Based on the conformation trajectory, pocket data is determined; wherein the pocket data is used to represent data of stable spatial position and geometric shape of the protein data in a change process of the conformation trajectory; The conformation trajectory and the pocket data are rendered and output.

[0014] The present disclosure provides a computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the protein pocket detection method when executing the program.

[0015] The present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the program is executed by a processor to implement the steps of the protein pocket detection method.

[0016] The technical solution provided by the present disclosure has the following advantages compared with the prior art: the task scheduling and management module coordinates data flow, the conformation generation module generates a conformation trajectory reflecting structural changes in the process of protein biological function by using molecular dynamics simulation, the pocket detection module determines pocket data of stable spatial position and geometric shape from the conformation trajectory, and finally the visualization module realizes rendering and output of the conformation trajectory and the pocket data. The whole can efficiently capture stable pockets under dynamic conformations of proteins, intuitively present related data, efficiently and accurately identify stable pockets from dynamic conformations of proteins, and intuitively display through interactive visualization, thereby significantly improving the efficiency and accuracy of drug target discovery and greatly reducing the use threshold. BRIEF DESCRIPTION OF DRAWINGS

[0017] The accompanying drawings, which are incorporated into and form a part of the specification, illustrate one embodiment consistent with the present disclosure and, together with the description, serve to explain the principles of the disclosure.

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without creative effort.

[0019] Figure 1 The structural schematic diagram of the protein pocket detection system in the embodiments of the present disclosure is shown in Figure 2 The flowchart of the protein pocket detection method in the embodiments of the present disclosure is shown in Figure 3 The flowchart of the protein pocket detection method in the embodiments of the present disclosure is shown in Figure 4 The flowchart of the three-layer cooperative framework of the protein pocket detection in the embodiments of the present disclosure is shown in DETAILED DESCRIPTION

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without creative effort.

[0021] In the following description, many specific details are set forth in order to provide a thorough understanding of the present disclosure, but the present disclosure can also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some of the embodiments of the present disclosure, not all the embodiments.

[0022] Figure 1 The structural schematic diagram of the protein pocket detection system in the embodiments of the present disclosure is shown in Figure 1 As shown in the figure, the detection system 100 includes: a conformation generation module 101, a pocket detection module 102, a task scheduling and management module 103 and a visualization module 104.

[0023] The task scheduling and management module 103 acquires the protein data to be detected, and coordinates and controls the automatic execution of the conformation generation module 101, the pocket detection module 102 and the visualization module 104.

[0024] Specifically, the protein data to be detected can cover two core types: one is protein sequence data (e.g., FASTA format data), which directly reflects the amino acid composition and arrangement order of the protein; the other is protein three-dimensional structure data (e.g., PDB format data), which presents the spatial configuration of the protein molecule, including atomic position, chemical bond connection, etc. The two data types correspond to the selection basis of different molecular dynamics simulation methods of the subsequent conformation generation module 101, providing initial data support for subsequent targeted generation of conformation trajectories.

[0025] The task scheduling and management module 103 serves as the core control hub of the system, responsible for starting the entire protein pocket detection process. This module receives input protein data through the user interaction interface, which can be protein sequence information or three-dimensional structure data.

[0026] After receiving the data, the task scheduling and management module 103 can perform the following key operations: first, perform validity check on the input protein data to ensure the standardization and integrity of the data format; then create an independent task identifier and initialize the task state; then store the verified data in the system designated storage area for subsequent computing process.

[0027] The task scheduling and management module 103 also undertakes the intelligent allocation of computing resources. According to the characteristics of the protein data and the current load of the system, the usage strategy of the computing resources is dynamically planned, and the corresponding computing job configuration is generated. Through the application program interface with the high-performance computing platform, the computing task and related parameters are submitted to the computing cluster.

[0028] In the task execution phase, the task scheduling and management module 103 continuously monitors the state changes of the computing job, including the queuing state, running progress and completion condition, etc. This state monitoring mechanism enables users to understand the task processing progress in real time, and provides the system with the ability to perceive abnormal situations in a timely manner.

[0029] For example, the backend defines a set of RESTful APIs, and the frontend submits tasks through POST requests. The request body contains input files, selected molecular dynamics simulation methods (bio-emu or AI 2BMD), parameters, etc. The backend has built-in supercomputing job script templates for bio-emu and fpocket parallel computing. Upon receiving a task, the backend program fills in the template with the specific parameters of the task (e.g., input file path, CPU / GPU node request) to generate an executable job script. Through the unified API of the supercomputer, the backend submits the generated script to the computing cluster. After submission, the backend immediately returns a task ID to the frontend and begins to periodically poll the status of the task in the supercomputer queue. The task status is updated in real time to the backend database, and the frontend can query the latest status through the task ID.

[0030] When the computing task is completed, the task scheduling and management module 103 automatically obtains the result data from the computing platform, performs standardized processing, and updates the task status to complete. Finally, the computing result is transmitted to the visualization module 104 through a standardized data interface, and the entire data processing process is completed.

[0031] Through this series of coordination and management work, the task scheduling and management module 103 effectively realizes the decoupling of user interaction and backend complex computing, provides a simple and clear operation experience for users, and at the same time realizes the optimized utilization of computing resources and efficient management of task flow at the bottom.

[0032] The conformation generation module 101 as the core computing unit of the system responds to the scheduling of the task scheduling and management module 103, and generates the conformation trajectory of the protein data based on the protein data obtained by the task scheduling and management module 103 using molecular dynamics simulation; wherein the conformation trajectory is used to represent the structural change process of the protein data in the execution of biological functions.

[0033] Specifically, the conformation generation module 101 adopts an intelligent model selection strategy, first identifies the type of the input protein data, and automatically selects the most suitable molecular dynamics simulation method according to the data type characteristics. Specifically, when the protein data is the first data type, i.e., protein sequence information, the system enables the first molecular dynamics simulation method for processing; when the protein data is the second data type, i.e., protein three-dimensional structure information, the second molecular dynamics simulation method with different structures is used for analysis. This differentiated molecular dynamics simulation method selection mechanism ensures that each type of input data can obtain accurate and reliable conformation trajectory generation effect; wherein the first data type can correspond to protein sequence data (e.g., FASTA format data reflecting the arrangement order of amino acids), and the second data type can correspond to protein structure data (e.g., PDB format data presenting the spatial configuration of molecules), the difference between the two data types directly determines the selection direction of the subsequent molecular dynamics simulation.

[0034] In the trajectory generation process, this module performs a fine calculation process: standardization preprocessing of input protein data, including atom naming normalization, structure integrity check and other necessary steps; according to the selected molecular dynamics simulation parameter configuration, deep sampling and generation calculation of protein conformation space are performed; through specific algorithm optimization calculation, the generated conformation trajectory can not only fully reflect the natural dynamic characteristics of the protein, but also maintain physical rationality.

[0035] For example, when the protein data is the first data type, i.e., protein sequence information, the system enables the bio-emu model for processing. This model is based on deep learning methods, uses a pre-trained AlphaFold2 sequence encoder to extract features from the input sequence, and combines multiple sequence alignment (MSA) to generate representation information. Subsequently, a coarse-grained structure representation module is used to retain only the heavy atoms of the protein backbone and establish a local coordinate system through orthogonalization processing, thereby obtaining an efficient representation of the backbone framework. On this basis, bio-emu gradually removes noise through a diffusion condition generation model to generate protein conformations that conform to the target distribution. At the same time, the Score model is introduced to score multidimensional information to ensure the accuracy and stability of the predicted conformation. Finally, the system can output trajectory files ( *. xtc) and topology files ( *. pdb) containing multiple candidate conformations. This process avoids the need for long physical simulations, significantly reducing the calculation period. Bio-emu is mainly suitable for structure sampling of monomeric proteins, while for multimers or complexes containing multiple chains, the system can use linker tricks or template-based modeling.

[0036] For example, when the protein data is the second data type, i.e., protein three-dimensional structure information, the system will preferentially call the AI 2 BMD model. This model first standardizes the input structure, including hydrogen supplementation, adding caps (ACE and NME) at the N and C terminals, and adjusting atom names through tools to ensure compatibility. Subsequently, the protein data will undergo solvation and energy minimization, and can select FF19SB or AMOEBA force field schemes for pre-equilibration as needed, to obtain a reasonable initial conformation. In the core simulation stage, AI 2 BMD will disassemble the protein into units through fragmentation, and use a machine learning potential function based on VisNet design to calculate the total energy and atomic force of the protein data at each time step, thereby driving molecular dynamics evolution with DFT-level ab initio accuracy. In addition to the default fragmentation mode, AI 2BMD also supports the overall ViSNet model, which can directly model the complete protein molecule, but requires user pre-training and provides corresponding potential energy parameters. Finally, the process can generate ASE format trajectory files (.traj), which can be converted to the commonly used.dcd format by the provided tools for further analysis and visualization. Compared with traditional molecular mechanics methods, AI 2 BMD exhibits higher quantum chemical accuracy in energy and atomic force calculations, and can maintain good consistency with experimental data on the dynamics and thermodynamics levels.

[0037] The conformation generation module 101 also includes generating topological data of the protein data based on the protein data obtained by the task scheduling and management module 103; the crane, the topological data covers the basic structural information of the protein molecule, for example, the connection mode of atoms, the positional relationship of amino acid residues, the characteristics of the molecular skeleton, etc., which can be used as supplementary data for the conformation trajectory; on the one hand, the topological data can provide a basic framework for the subsequent pocket detection module to analyze the spatial position and geometric shape of the pocket; on the other hand, it can also provide basic information for the visualization module to render the protein structure, ensuring the complete presentation of dynamic trajectories and static structural characteristics.

[0038] The pocket detection module 102 is a key analysis unit of the system, which responds to the scheduling of the task scheduling and management module 103 and determines the pocket data based on the conformation trajectory generated by the conformation generation module 101; wherein the pocket data is used to represent the spatial position and geometric shape stable data of the protein data in the change process of the conformation trajectory.

[0039] Specifically, the pocket detection module 102 extracts a plurality of conformation samples from the conformation trajectory; at the same time, the plurality of conformation samples are subjected to pocket detection to obtain a plurality of pocket detection results; the pocket detection results are clustered and / or filtered to determine the pocket data.

[0040] The execution process of the pocket detection module 102 includes three core links: first, a plurality of representative conformation samples are systematically extracted from the conformation trajectory, these samples are not randomly selected, but are uniformly sampled according to a predetermined time interval or frame number (for example, one structure is selected every fixed simulation step), to ensure that the selected samples can cover the main conformation change state of the protein, avoiding both one-sided analysis due to insufficient sample quantity and increased computational burden due to sample redundancy.

[0041] For the extracted multiple conformation samples, the pocket detection module 102 performs pocket detection in a parallel processing manner. This processing manner is not to detect a single sample one by one, but to distribute the detection tasks of multiple conformation samples to different computing units by means of computing resources, and simultaneously start the detection operation. Through parallel detection, the task that originally needs to be processed in series for several hours can be shortened to be completed within several minutes, greatly improving the detection efficiency; at the same time, the detection of each sample is based on the unified pocket recognition logic, ensuring the consistency and comparability of the detection results of different samples, and finally obtaining the pocket detection result corresponding to each conformation sample.

[0042] Due to the differences in the structures of different conformation samples, the detected pockets may have overlaps or differences in position and shape. In order to screen out the pockets with stable spatial position and geometric shape in the dynamic change process of the protein, the pocket detection module 102 will further process all the pocket detection results: on the one hand, the pocket detection results with similar spatial position and geometric characteristics are classified into one class through clustering algorithm, and this kind of pocket can be regarded as the same pocket in different conformations; on the other hand, the pocket categories with low appearance frequency, large shape fluctuation or poor drugability are removed by screening in combination with the appearance frequency, volume stability, hydrophobicity parameter, drugability score and other indicators. Through the cooperative processing of clustering and screening, the finally determined pocket data can accurately represent the stable pocket characteristics of the protein in the dynamic conformation change, and provide reliable target reference for subsequent visualization analysis and drug design.

[0043] The visualization module 104, as an important interactive interface of the system, responds to the scheduling of the task scheduling and management module 103, and is responsible for rendering and outputting the conformation trajectory generated by the conformation generation module 101 and the pocket data determined by the pocket detection module 102.

[0044] The visualization module 104 will first perform basic rendering processing on the conformation trajectory and the pocket data. For the conformation trajectory, the cartoon mode or the ribbon mode will be used to present the protein main chain structure, and the trajectory will be played in a dynamic form to completely show the structural changes of the protein in the process of performing biological functions, such as the movement of the molecular skeleton, the opening and closing of the domain, etc.; for the pocket data, a semi-transparent surface mode will be used to superimpose on the protein structure, and different colors will be used to distinguish different pockets, so that users can quickly locate the spatial distribution of the pockets on the protein molecule and intuitively perceive the geometric shape of the pockets. Through this rendering mode, the originally complex structure data is converted into visual images, reducing the understanding threshold of users.

[0045] The visualization module 104 further comprises: in response to selection and / or query operation on the pocket data, extracting the selected pocket data from the pocket data; outputting the selected pocket data in highlight mode, while outputting attribute information of the selected pocket data; wherein the attribute information comprises at least one of pocket average volume, drugability score, hydrophobicity parameter, and occurrence frequency.

[0046] Specifically, to meet the user's targeted analysis needs, the visualization module 104 supports interactive operation on the pocket data. When the user selects or queries specific pocket data through the interface, the visualization module 104 will respond to these operations in real time: on the one hand, the selected pocket data is extracted from the overall pocket data and presented in highlight mode (e.g., increased opacity, changed edge color) in the 3D view, which is clearly distinguished from other pockets, helping the user focus on the target pocket; on the other hand, the attribute information of the pocket data is output synchronously, which covers the pocket average volume, drugability score, hydrophobicity parameter, occurrence frequency, etc., providing data support for the user to judge the research value of the pocket in depth.

[0047] The visualization module 104 further comprises: in the process of outputting the conformation trajectory generated by the conformation generation module 101, identifying the pocket data corresponding to the current output frame, and synchronously outputting the pocket data corresponding to the current output frame in highlight mode.

[0048] Specifically, considering the dynamic characteristics of the conformation trajectory, the visualization module 104 also has the ability to synchronously display the trajectory and the pocket data. In the process of playing the conformation trajectory, the visualization module 104 will identify the pocket data corresponding to the current output frame (i.e., the protein structure at a certain time) in real time; for example, a specific pocket formed by the opening of the protein conformation in a certain frame, the visualization module 104 will immediately highlight the pocket in the 3D view, allowing the user to intuitively observe the correlation between protein structure changes and pocket appearance / disappearance / morphology adjustment. This synchronous highlighting design can help users clearly capture the rules of pocket changes with the dynamic conformation of the protein, such as some pockets appearing only in certain conformational states or maintaining stable morphology during the trajectory playback process, providing direct reference for understanding the dynamic mechanism of drug binding.

[0049] In this embodiment, after the task scheduling and management module 103 obtains the protein data, the data is transmitted to the conformation generation module 101; the conformation generation module 101 generates the conformation trajectory and topological data, and transmits the conformation trajectory to the pocket detection module 102; the pocket detection module 102 determines the pocket data based on the conformation trajectory, and then transmits the conformation trajectory and the pocket data to the visualization module 104; the visualization module 104 renders and interactively displays the two types of data, realizing comprehensive analysis and presentation of the dynamic pockets of the protein.

[0050] Based on the above protein pocket detection system, Figure 2 The flowchart of the protein pocket detection method of the embodiments of the present disclosure is applied to the above protein pocket detection system, as shown in the figure. Figure 2 The protein pocket detection method includes the following steps: S201, obtaining protein data to be detected.

[0051] Specifically, the system receives the user-submitted protein information to be analyzed through the task scheduling and management module. The system supports two main data input forms: one is protein sequence data, which records the amino acid composition and arrangement order of the protein, and is the core basis for subsequent sequence-based prediction of protein dynamic structure; the other is protein structure data, which presents the three-dimensional spatial configuration of the protein molecule, including atomic position, chemical bond connection, and amino acid residue spatial distribution details, providing a basis for optimizing and simulating dynamic changes based on existing structures.

[0052] After receiving the data, the system will check the integrity and format specification of the data, and create an independent identifier and record for the analysis task, and include it in the task queue for unified management.

[0053] S202, based on the protein data, generating conformational trajectories of the protein data using molecular dynamics simulation.

[0054] Specifically, the protein data obtained in S201 is identified by type to determine whether it is protein sequence data or protein structure data. The structural feature difference between the two data types directly determines the selection of the molecular dynamics simulation method.

[0055] According to the data type, the corresponding molecular dynamics simulation method is matched. If the protein data is protein sequence data, a molecular dynamics simulation method suitable for sequence analysis (for example, bio-emu model) is selected; if the data is protein structure data, a molecular dynamics simulation method based on optimization of existing structure (for example, AI 2 BMD model) is selected. The two types of molecular dynamics simulation methods have different structures and can achieve efficient conformation generation according to the data characteristics.

[0056] Through the selected molecular dynamics simulation operation, conformational trajectories (for example, xtc, traj format files) containing structural snapshots of the protein at different time nodes or functional states are generated. The trajectory can fully present the dynamic structural changes of the protein, including main chain movement, domain opening and closing, local conformation adjustment, etc., providing a comprehensive structural change basis for subsequent detection of dynamic pockets; at the same time, topological data (for example,.pdb format files containing atomic connection mode and molecular skeleton characteristics) are also generated based on the protein data as supplementary information for the conformational trajectory to assist subsequent analysis.

[0057] S203, determining pocket data based on the conformational trajectory.

[0058] The pocket data is used to represent data of stable spatial position and geometric shape of the protein data in the process of change of the conformational trajectory.

[0059] Specifically, a plurality of conformation samples covering the main conformational change state of the protein are selected from the conformational trajectory generated in S202 by uniform sampling at a preset time interval or frame number (for example, 100 conformations are extracted), so as to avoid one-sided analysis caused by too few samples and to avoid increasing the calculation burden caused by sample redundancy.

[0060] With the aid of high-performance computing resources, the pocket detection task of the plurality of conformation samples extracted is distributed to different computing units, and the pocket detection operation is started at the same time (for example, the pocket is identified by analyzing the geometric features, hydrophobicity, etc. of the concave area on the surface of the protein), which greatly shortens the detection time and obtains the pocket detection result corresponding to each conformation sample.

[0061] All pocket detection results are processed in multiple dimensions to determine the final pocket data. First, the clustering algorithm is used to classify the pocket detection results with similar spatial positions and geometric features into one category, which is regarded as the performance of the same pocket in different conformations; then, the appearance frequency, volume stability, hydrophobicity parameter, drugability score and other indicators of the pocket are combined for screening to remove the pocket categories with low appearance frequency, large morphological fluctuation and poor drugability, and finally the pocket data obtained can accurately represent the pocket features of stable spatial position and geometric shape of the protein in the process of change of the conformational trajectory.

[0062] For example, from the.xtc and.traj conformational trajectory files generated by the dynamic conformation generation module, a series of representative protein three-dimensional structure snapshots are obtained by uniform sampling at a preset time interval or frame number (for example, 100 conformations are extracted); the pocket detection task of the N structure snapshots sampled is parallelized by using a job scheduling system. Figure 3 The flowchart of the pocket detection method of the embodiment of the present disclosure is shown in FIG. 1. Figure 3 As shown in FIG. 1, the user provides protein data, the cloud server initiates a task to the supercomputing system, the supercomputing system predicts the dynamic structure of the protein after delivering the task, and then detects the pocket in parallel through a plurality of Fpocket tools, and finally the result is fed back to the user to form a complete closed loop.

[0063] The specific implementation is that the main job script submits N independent fpocket sub-tasks to the scheduling system, each sub-task is allocated to a different computing core or node, and a structure snapshot is calculated simultaneously. This large-scale parallel processing mode shortens the calculation task which originally takes several hours to complete in several minutes. After all the fpocket sub-tasks are completed, an aggregation script automatically collects the pocket detection results of all conformations. Through an algorithm, the pockets appearing in different frames are spatially clustered, and according to the frequency of appearance, volume, hydrophobicity, drugability score and other indicators, a comprehensive ranking is performed, and finally a group of "consensus pockets" (Consensus Pockets) with the highest probability, the strongest stability and the best drugability are selected, and each pocket is labeled with which conformations it appears in.

[0064] S204, rendering and outputting the conformation trajectory and pocket data.

[0065] Specifically, a professional visualization engine is used to perform basic rendering on the conformation trajectory and pocket data. For the conformation trajectory, the protein main chain structure is displayed in cartoon mode or ribbon mode, and the trajectory is played in a dynamic form to completely present the structural change process of the protein. For the pocket data, a semi-transparent surface mode is used to superimpose the protein structure, and different colors are used to distinguish different pockets, helping users quickly locate the spatial distribution of the pockets and intuitively perceive the geometric shape of the pockets.

[0066] For example, when the user requests to view the completed task results, the front end obtains the URL of the result file from the back end. The Mol component can simultaneously load the topological structure of the protein (*.pdb), the dynamic trajectory (*.xtc, *.traj), and the integrated pocket data (for example, a PDB file or geometric data containing all instances of each consensus pocket is generated for each consensus pocket).

[0067] At the same time, the user can select or query the pocket data. When the user triggers the operation, the selected pocket data is extracted from the pocket data in real time, and is presented in a highlighted form in the 3D view, and the attribute information of the pocket is output, providing data support for in-depth analysis of the research value of the pocket.

[0068] During the playback of the conformation trajectory, the pocket data corresponding to the current output frame is identified in real time, and these pocket data are synchronously highlighted and output, allowing the user to intuitively observe the correlation between the protein structure change and the appearance, disappearance, and shape adjustment of the pockets, and clearly capture the rules of the pockets changing with the conformations, providing intuitive reference for understanding the drug binding mechanism and target selection.

[0069] In the above scheme, the diversified protein data covering sequence and structure are acquired to lay the foundation for analysis, the topological data are generated by combining the data type with the dedicated molecular dynamics simulation method, and the conformation trajectory reflecting the structural changes of the biological functional processes of the protein is generated, the spatial position and geometric shape stable pocket data are determined through conformation sample extraction, parallel detection and clustering screening, the intuitive presentation of the conformation trajectory and the pocket data is realized through basic rendering, interactive display and dynamic synchronous highlighting, the overall process not only efficiently captures the stable pockets in the dynamic conformation of the protein, but also greatly reduces the use threshold of non-computing background personnel, and can provide clear and valuable reference for drug target discovery, drug action mechanism research and drug design, and significantly improves the efficiency and practicability of protein pocket analysis.

[0070] The present disclosure is further illustrated below through specific embodiments.

[0071] Figure 4 The flowchart of the three-layer cooperative framework for protein pocket detection of the embodiments of the present disclosure is as shown in Figure 4 The framework includes an application layer, a scheduling layer and a computing layer.

[0072] Specifically, the application layer provides a user interaction interface, including task submission and preview, dynamic interaction of protein molecules, and multi-source data loading function, which is the entrance for user interaction with the system, responsible for data import, task initiation and result preview, supports dynamic visual analysis of protein molecules, and meets the loading demand of multiple types of data.

[0073] The application layer is constructed based on Web technology, and can be accessed by the user through a browser. The user can submit a protein sequence (such as FASTA format) or a three-dimensional structure file (such as PDB format). After task submission, the page does not need to wait, and the user can check the real-time state (in queue, in calculation, completed) of the task in the “task management” interface. For the completed task, an integrated 3D preview window is provided, and the Mol* engine is used to load and render the result.

[0074] The scheduling layer is deployed on a cloud server, and is responsible for submitting tasks, interacting with the front end, and managing the task queue, which is the task hub of the system, responsible for receiving task requests from the application layer, and realizing data interaction with the front-end interface, and at the same time, queuing and priority management of the tasks are performed to ensure the orderly execution of the tasks; at the same time, the user file temporary storage service is provided.

[0075] The scheduling layer uses the Django REST Framework architecture to provide standard API interfaces for the front-end to call. This layer does not perform computationally intensive tasks, but is responsible for logical processing and task scheduling. When a task is received, the backend will dynamically generate a job script submitted to the supercomputer according to the input type (sequence or PDB) and parameters, and initiate submission through the API. This layer will periodically query the task status and update the database for the front-end to query.

[0076] The computing layer is composed of a supercomputer computing cluster. It has parallel computing and GPU accelerated computing capabilities, and is the computing core of the system. It improves large-scale data processing efficiency through parallel computing technology, and further strengthens the support for computationally intensive tasks such as protein dynamics simulation and pocket detection using GPU acceleration, including using bio-emu or AI 2 BMD for dynamic conformation generation, and parallel fpocket pocket detection, providing efficient computing power support for upper-layer functions, The computing layer pre-deploys bio-emu, AI 2 BMD, fpocket and other core computing software on the supercomputer platform. The scheduling layer starts the calculation by submitting a customized job script (such as a Slurm script). The computing task is executed on the CPU or GPU nodes of the supercomputer, making full use of its powerful parallel computing capabilities. After the calculation is completed, the result files (such as xtc, pdb and pocket information files) are stored in the shared file system of the supercomputer, waiting for the scheduling layer to access and obtain through the API.

[0077] The following is a further description of the present disclosure by analyzing a dynamic pocket of a kinase target, and the specific implementation process is as follows: The user logs in to the Web platform and enters the task submission page. Select "Submit by sequence" and paste the FASTA sequence of the kinase into the input box. Click the "Submit" button.

[0078] The front-end sends the sequence to the back-end (scheduling layer). The back-end verifies the input, creates a new task, and submits a bio-emu or AI 2 BMD calculation job through the API.

[0079] The supercomputer (computing layer) receives the job and runs bio-emu or AI 2 BMD on the GPU node to generate the dynamic trajectory of the protein.

[0080] The job script then extracts 100 conformations from the trajectory and starts 100 parallel fpocket sub-tasks for pocket detection.

[0081] After all subtasks are completed, the aggregation script clusters and ranks the results, generating the final pocket data file and a summary report.

[0082] The user sees the task status change to "Completed" on the Task Management page and clicks "View Results".

[0083] The front-end loads the Mol* visualization component and fetches the URL of the results file from the back-end.

[0084] Mol* renders the dynamic structure of the kinase and the top 5 high-probability pockets (shown with different colored surfaces) simultaneously.

[0085] The user can click the play button to observe the opening and closing of the pockets during the kinase's motion; click "Pocket 1" in the pocket list to make the corresponding pocket in the view opaque and highlighted, and the list displays the average volume, drugability score, etc. of the pocket; rotate and zoom the view to carefully observe the fine structure of the pocket and the surrounding amino acid residues from different angles.

[0086] In summary, the present disclosure builds a full-automatic dynamic pocket analysis solution from input to visualization through technical innovation and process integration, significantly reduces the technical threshold, improves analysis efficiency, and provides a powerful computing tool for drug discovery.

[0087] The embodiment also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the program to implement the steps of the protein pocket detection method.

[0088] The embodiment also provides a computer readable storage medium, which stores computer program codes in the computer readable storage medium (including but not limited to disk memory, compact disc read-only memory (CD-ROM), optical memory, etc.), and the computer program codes make the computer execute the related method steps to implement the protein pocket detection method provided in the above embodiment when the computer program codes are executed on the computer.

[0089] The beneficial effects of the above embodiments can refer to the beneficial effects of the corresponding methods provided above, and will not be repeated here.

[0090] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0091] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A protein pocket detection system characterized by, The detection system comprises a conformation generation module, a pocket detection module, a task scheduling and management module and a visualization module: The task scheduling and management module acquires protein data to be detected and coordinates the execution of the conformation generation module, the pocket detection module and the visualization module; The conformation generation module generates a conformation trajectory of the protein data by using molecular dynamics simulation in response to the scheduling of the task scheduling and management module based on the protein data acquired by the task scheduling and management module; wherein the conformation trajectory is used to represent the structural change process of the protein data in the execution of biological functions; The pocket detection module determines pocket data based on the conformation trajectory generated by the conformation generation module in response to the scheduling of the task scheduling and management module; wherein the pocket data is used to represent the data of stable spatial position and geometric shape in the change process of the conformation trajectory; The visualization module renders and outputs the conformation trajectory generated by the conformation generation module and the pocket data determined by the pocket detection module in response to the scheduling of the task scheduling and management module.

2. The protein pocket detection system of claim 1, wherein, The task scheduling and management module generates a conformation trajectory of the protein data by using molecular dynamics simulation based on the protein data acquired by the task scheduling and management module, comprising: determining the data type of the protein data; determining the molecular dynamics simulation method based on the data type; generating the conformation trajectory by using the molecular dynamics simulation.

3. The protein pocket detection system of claim 2, wherein, The determination of the molecular dynamics simulation method based on the data type comprises: when the data type is a first data type, determining the molecular dynamics simulation method as a first molecular dynamics simulation method; wherein the first data type is used to represent protein sequence data; when the data type is a second data type, determining the molecular dynamics simulation method as a second molecular dynamics simulation method; wherein the second data type is used to represent protein structure data, and the first molecular dynamics simulation method and the second molecular dynamics simulation method are different in structure.

4. The protein pocket detection system of claim 1, wherein, The pocket detection module determines pocket data based on the conformation trajectory generated by the conformation generation module, comprising: extracting a plurality of conformation samples from the conformation trajectory; simultaneously performing pocket detection on a plurality of the conformation samples to obtain a plurality of pocket detection results; clustering and / or screening the pocket detection results to determine the pocket data.

5. The protein pocket detection system of claim 1, wherein, The conformation generation module further comprises: generating topological data of the protein data based on the protein data acquired by the task scheduling and management module.

6. The protein pocket detection system of claim 1, wherein, The visualization module further comprises: in response to selection and / or query operation on the pocket data, extracting selected pocket data from the pocket data; highlighting and outputting the selected pocket data, and outputting attribute information of the selected pocket data; wherein the attribute information comprises at least one of pocket average volume, drugability score, hydrophobicity parameter and frequency of occurrence.

7. The protein pocket detection system of claim 1, wherein, The visualization module further comprises: During outputting the conformation trajectory generated by the conformation generation module, pocket data corresponding to a current output frame is identified, and the pocket data corresponding to the current output frame is synchronously highlighted and outputted.

8. A method of protein pocket detection, characterized by, The method is applied to a protein pocket detection system, and the method comprises: Obtaining protein data to be detected; Based on the protein data, a conformation trajectory of the protein data is generated by using molecular dynamics simulation; Based on the conformation trajectory, pocket data is determined; wherein the pocket data is used to represent data of stable spatial position and geometric shape of the protein data in the change process of the conformation trajectory; The conformation trajectory and the pocket data are rendered and outputted.

9. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the steps of the protein pocket detection method in claim 8.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps of the protein pocket detection method in claim 8.

Images