Protein model quality evaluation method and device, computer equipment and storage medium

Abstract

The embodiment of the invention belongs to the technical field of artificial intelligence, and relates to a protein model quality evaluation method and device, computer equipment and a storage medium, and the method comprises the steps that a PDB file input by a user terminal is received, and the PDB file comprises the characteristics of each atom in protein and RNA structure scoring; according to the characteristics of each atom in the protein and RNA structure score and a preset algorithm, constructing an Atom-ProteinQA model; a quality evaluation request sent by a user terminal is received, and the quality evaluation request at least comprises to-be-evaluated protein structure data; and inputting the protein structure data into the Atom-ProteinQA model to carry out quality evaluation operation so as to obtain a quality evaluation result. According to the method, the LDDT score of the protein and RNA structure scoring atom level is used as supervision, accurate IDDL score prediction of the atom level is effectively carried out, and the scoring structure change of the protein and RNA structure is evaluated on one atom level.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a protein model quality assessment method, apparatus, computer equipment and storage medium. Background Art

[0002] Proteins are ubiquitous in almost all biological processes. Determining their structure and function helps to understand and potentially control these processes. However, although the determination of protein sequences is now a routine procedure, it is often very difficult to use this information to extract relevant functional knowledge of the system under study. In fact, the function of a protein relies on a combination of its chemical and mechanical properties, which are determined by its structure. Therefore, identifying a protein structure from its sequence is of great importance, albeit a difficult task. Experimental structure identification is not feasible in all cases and is often very tedious and expensive. Therefore, computational methods have emerged in the past that attempt to predict protein structure from its sequence. Most of these methods combine a sampling step of protein conformations, where the former generates protein conformations, and the latter scores these conformations to select protein conformations that are as close as possible to the native structure, with a model quality assessment step. Protein model quality assessment is an important and urgent problem in structural bioinformatics.

[0003] There are existing methods for protein model quality assessment that solve the scoring problem by using energy functions derived from general physical principles, such as CHARMM, AMBER, OPLS, GROMOS, derived from different known protein and RNA structure scoring sets, such as GOAP, RW, DFIRE, or both, such as Rosetta. They usually estimate a specific similarity score between the computational model and the reference structure, such as LDDT, CAD, GDT, TMscore, etc. Recently, various learning-based methods have been developed. They perform similarity prediction by utilizing artificial neural networks, support vector machines, CNN, LSTM, GCN, and 3D models.

[0004] However, the applicant found that none of them can predict the ProteinQA results at the atomic level, because the atomic inference for scoring all protein and RNA structures is very computationally intensive. In addition, they cannot simultaneously consider the topological and geometric relationships between atoms, resulting in the problem of low assessment accuracy of traditional protein model quality assessment methods. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose a protein model quality assessment method, apparatus, computer equipment and storage medium to solve the problem of low assessment accuracy in traditional protein model quality assessment methods.

[0006] In order to solve the above technical problems, the present application embodiment provides a protein model quality assessment method, which adopts the following technical solution:

[0007] Receiving a PDB file input by a user terminal, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring;

[0008] Constructing an Atom-ProteinQA model based on the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm;

[0009] Receiving a quality assessment request sent by a user terminal, wherein the quality assessment request at least includes protein structure data to be assessed;

[0010] The protein structure data is input into the Atom-ProteinQA model to perform a quality assessment operation to obtain a quality assessment result.

[0011] Furthermore, the preset algorithm first searches for a radius of P n The atom P n [i,j]'s neighbors, then select P by the preset index m and P n The neighboring atoms in P are calculated separately. m [i,j] and P n The respective distances of [i,j], P m and P n The difference between the individual distances in will be passed through the mapping function converted to predefined scores, and finally, the LDDT scores at the atomic, residue, and global levels were obtained by normalizing by the total number of neighbors.

[0012] Furthermore, the Atom-ProteinQA model includes a geometric perception module, and the step of inputting the protein structure data into the Atom-ProteinQA model for quality assessment to obtain a quality assessment result specifically includes the following steps:

[0013] Performing a feature extraction operation on the protein structure data to obtain voxelized Cartesian coordinates of each residue atom in the protein structure data;

[0014] The geometric features of the voxelized Cartesian coordinates are captured by the geometric perception module to generate a first atomic-level prediction result.

[0015] Furthermore, the Atom-ProteinQA model includes a topology perception module, and the step of inputting the protein structure data into the Atom-ProteinQA model for quality assessment to obtain a quality assessment result specifically includes the following steps:

[0016] constructing an atomic-level topological map corresponding to the protein structure data according to the topological perception module;

[0017] A second atomic-level prediction result is output according to the atomic-level topology map and the message passing mechanism.

[0018] Furthermore, the Atom-ProteinQA model includes a geometry perception module, a topology perception module and a cross-model fusion module. The step of inputting the protein structure data into the Atom-ProteinQA model for quality assessment to obtain a quality assessment result specifically includes the following steps:

[0019] Acquire atomic feature data of the protein structure data according to the geometric perception module;

[0020] Acquire an initial topological map of the protein structure data according to the topological perception module;

[0021] Performing a feature fusion operation on the atomic features and the topological map according to the cross-model fusion module to obtain an updated topological map;

[0022] A quality prediction operation is performed on the updated topology map according to the topology perception module and the message passing mechanism to obtain a third atomic-level prediction result.

[0023] In order to solve the above technical problems, the present application also provides a protein model quality assessment device, which adopts the following technical solution:

[0024] A file acquisition module, used to receive a PDB file input by a user terminal, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring;

[0025] A model building module, used to build an Atom-ProteinQA model according to the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm;

[0026] A request acquisition module, configured to receive a quality assessment request sent by a user terminal, wherein the quality assessment request at least includes protein structure data to be assessed;

[0027] The quality assessment module is used to input the protein structure data into the Atom-ProteinQA model to perform a quality assessment operation to obtain a quality assessment result.

[0028] Furthermore, the Atom-ProteinQA model includes a geometric perception module, and the quality assessment module includes:

[0029] A feature extraction submodule, used for performing a feature extraction operation on the protein structure data to obtain voxelized Cartesian coordinates of each residue atom in the protein structure data;

[0030] The first prediction submodule is used to capture the geometric features of the voxelized Cartesian coordinates according to the geometric perception module to generate a first atomic-level prediction result.

[0031] Furthermore, the Atom-ProteinQA model includes a topology perception module, and the quality assessment module includes:

[0032] A topology map construction submodule, used to construct an atomic-level topology map corresponding to the protein structure data according to the topology perception module;

[0033] The second prediction submodule is used to output a second atomic-level prediction result according to the atomic-level topology map and the message passing mechanism.

[0034] Furthermore, the Atom-ProteinQA model includes a geometric perception module, and the calculation module includes:

[0035] An atomic feature acquisition submodule, used for acquiring atomic feature data of the protein structure data according to the geometric perception module;

[0036] An initial topology map acquisition submodule, used for acquiring an initial topology map of the protein structure data according to the topology perception module;

[0037] A feature fusion submodule, used for performing a feature fusion operation on the atomic features and the topological map according to the cross-model fusion module to obtain an updated topological map;

[0038] The third prediction submodule is used to perform a quality prediction operation on the updated topology map according to the topology perception module and the message passing mechanism to obtain a third atomic-level prediction result.

[0039] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0040] The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the protein model quality assessment method as described above when executing the computer-readable instructions.

[0041] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0042] The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the protein model quality assessment method described above are implemented.

[0043] The present application provides a protein model quality assessment method, including: receiving a PDB file input by a user terminal, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring; constructing an Atom-ProteinQA model according to the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm; receiving a quality assessment request sent by the user terminal, wherein the quality assessment request at least includes the protein structure data to be evaluated; inputting the protein structure data into the Atom-ProteinQA model for quality assessment operation to obtain a quality assessment result. Compared with the prior art, the present application uses the LDDT score at the atomic level of protein and RNA structure scoring as supervision, effectively performs accurate atomic-level IDDL score prediction, and evaluates the structural changes of protein and RNA structure scoring at an atomic level. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0046] Figure 2 is a flowchart of the protein model quality assessment method provided in the embodiment of the present application;

[0047] Figure 3 It is a schematic diagram of an embodiment of the Atom-ProtinQA model architecture provided in an embodiment of the present application;

[0048] Figure 4 It is a schematic diagram of an embodiment of the algorithm structure provided in the embodiment of the present application;

[0049] Figure 5 It is a schematic diagram of an embodiment of the comparison of the qualitative results of the model provided in the embodiments of the present application;

[0050] Figure 6is a schematic diagram of the structure of a protein model quality assessment device provided in an embodiment of the present application;

[0051] Figure 7 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0052] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0053] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0054] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0055] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0056] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0057] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to the laptop computer 1011, the tablet computer 1012 or the mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0058] The server 103 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal device 101 .

[0059] It should be noted that the protein model quality assessment method provided in the embodiments of the present application is generally executed by a server / terminal device, and accordingly, the protein model quality assessment device is generally arranged in the server / terminal device.

[0060] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0061] Continue to refer Figure 2 , shows a flow chart of an embodiment of a protein model quality assessment method according to the present application. The protein model quality assessment method comprises: step S201, step S202, step S203 and step S204.

[0062] In step S201, a PDB file input by a user terminal is received, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring;

[0063] In step S202, an Atom-ProteinQA model is constructed according to the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm;

[0064] In step S203, a quality assessment request sent by a user terminal is received, wherein the quality assessment request at least includes protein structure data to be assessed;

[0065] In step S204, the protein structure data is input into the Atom-ProteinQA model for quality assessment to obtain a quality assessment result.

[0066] In the embodiment of the present application, in the model of the present application, a complete protein and RNA structure scoring atom set It is represented by three-dimensional coordinates and K-dimensional features. The K-dimensional feature is the amino acid type of each atom. The goal of the Atom-ProteinQA model is to predict the global score (S) of a given bait protein. G ), the shape is The residue score (S R ) and the shape is The atomic score (S A ). R Contains the M residue scores for protein and RNA structure scoring, S A Contains N atomic scores for protein and RNA structure scoring. The prediction score is between 0 and 1 and is used to measure the similarity between the decoy and native structures. If the global score, residue score, and atom score are close to 1, it means that the protein and RNA structure scoring model is of higher quality. When the score is close to 0, the quality decreases. The above description can be expressed as:

[0067]

[0068] like Figure 3 As shown, atomic-level ground truth is generated as supervision for atomic-level predictions. The first set of input features of the network are the voxelized Cartesian coordinates of atoms of each residue. The GP module takes the entire atoms of protein and RNA structure scores as input and generates atomic-level predictions. It focuses on capturing the geometric information of protein and RNA structure scores and outputs fine-grained atomic features. In contrast, the TP module considers the topological relationships between atoms of protein and RNA structure scores, uses chemical bonds as edges to construct an undirected graph, and utilizes message passing to aggregate atomic neighborhood features. Between these two modules, cross-model feature fusion is performed to enhance mutual representation. Finally, by combining the above components with multi-level supervision, Atom-ProteinQA can make accurate predictions at both residue and atomic levels simultaneously.

[0069] In the training phase, the atomic-level LDDT scores are calculated with native structures as supervision. In the GP module, the sparse convolution-based U-Net updates the features and feeds them to the topology-aware (TP) module. For each protein and RNA structure score, it builds a topology graph using the chemical bonds and distances within the threshold of the protein and RNA structure scores. Afterwards, the TP module is applied to update the graph features through message passing to obtain the final atomic-level predictions.

[0070] In some optional implementations of the embodiments of the present application, the preset algorithm first searches for a radius of P n The atom P n[i,j]'s neighbors, then select P by the preset index m and P n The neighboring atoms in P are calculated separately. m [i,j] and P n The respective distances of [i,j], P m and P n The difference between the individual distances in will be passed through the mapping function converted to predefined scores, and finally, the LDDT scores at the atomic, residue, and global levels were obtained by normalizing by the total number of neighbors.

[0071] In the embodiments of the present application, the present application directly generates atomic-level labels as ground-truth supervision. In previous work, the LDDT metric is most widely used as a local accuracy assessment. Specifically, it is calculated on all pairs of atoms in the reference structure that are closer than a predefined threshold. Please note that there is no open source code for LDDT calculations, and Mariani et al. 10 only provides an online server to generate residue-level LDDT scores. Therefore, the present application first inputs a PDB file, and the resulting output contains the features of each atom in the protein and RNA structure scoring. These features include 21 residue types (A, R, N, D, C, Q, E, G, H, I, L, K, M, F, P, S, T, W, Y, V, and unknown), 5 atom types (C, N, O, S, and unknown), and three coordinates for each atom. Subsequently, the present application adapted the original algorithm outlined in their paper and modified it to generate atomic-level LDDT, such as Figure 4 In the algorithm, this application uses to represent the atomic coordinates of the protein and RNA structure scoring models and their native structures. Each protein and RNA structure scoring model contains M residues, and the mth residue includes N m atom This application uses P[i,j] to represent the jth atom of the i-th residue in the protein and RNA structure score P. More specifically, the algorithm first searches for a region with a radius of P n The atom P n The neighbors of [i,j] are selected, and the indices of these neighbors are retained. Then, P is selected by the preset index. m and P n and calculate their neighboring atoms to P m [i,j] and P n The respective distances of [i,j]. Then, P m and P n The difference between the individual distances in will be passed through the mapping function Finally, by normalizing the total number of neighbors, this application can obtain LDDT scores at the atomic, residue, and global levels (i.e. and ).

[0072] In an embodiment of the present application, a protein model quality assessment method is provided, including: receiving a PDB file input by a user terminal, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring; constructing an Atom-ProteinQA model according to the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm; receiving a quality assessment request sent by the user terminal, wherein the quality assessment request at least includes the protein structure data to be assessed; inputting the protein structure data into the Atom-ProteinQA model for quality assessment operation to obtain a quality assessment result. Compared with the prior art, the present application uses the LDDT score at the atomic level of protein and RNA structure scoring as supervision, effectively performs accurate atomic-level IDDL score prediction, and evaluates the structural changes of protein and RNA structure scoring at an atomic level.

[0073] In some optional implementations of the embodiments of the present application, the Atom-ProteinQA model includes a geometric perception module, and the protein structure data is input into the Atom-ProteinQA model for quality assessment operation to obtain the quality assessment result, which specifically includes the following steps:

[0074] Perform feature extraction on the protein structure data to obtain voxelized Cartesian coordinates of each residue atom in the protein structure data;

[0075] The geometric features of voxelized Cartesian coordinates are captured by the geometry perception module to generate the first atomic-level prediction results.

[0076] In an embodiment of the present application, a geometry perception (GP) module is designed to capture geometric metric information of protein and RNA structure scoring structures. The protein and RNA structure scoring structures are first represented as 3D point clouds based on their atomic coordinates. Previous work has been unable to learn effective geometric representations because they use traditional 3D-CNNs to extract features. However, normal voxelization in 3D space usually produces a large proportion of empty voxels, and convolution of these invalid voxels brings a huge computational burden. Therefore, the resolution of 3D voxels used in these methods is usually low, resulting in the loss of geometric details in the protein and RNA structure scoring structures. In this article, the present application innovatively utilizes 3D sparse convolution to solve the above problems. The present application converts the coordinates of the original atoms (i.e., point clouds) into sparse volume representations. Specifically, 3D point clouds can be represented by x={x k}={(p k ,f k )} format, where p k is the 3D coordinate of the kth point, f k is its corresponding feature. First, the application moves all points to a local coordinate system with the geometric center as the origin. Then, the application rotates these points along a randomly selected axis (x, y or z axis) at a random angle α∈[0,π). Point feature {f k} remains unchanged, this application expresses the transformed coordinates as Afterwards, this application converts the above converted point cloud into a sparse voxel representation with a resolution of r:

[0077]

[0078] in is the rounding operation, yes A binary indicator of whether it belongs to the mth voxel grid. N m is the number of points in the mth voxel, and this application takes the average of the features of these points. Only non-empty voxels (N m >0), these voxels are saved in a hash table. Then, the convolution operation is performed only on these non-empty voxels. In this way, the application can represent point clouds with a larger volume resolution while maintaining computational efficiency. The difference between traditional voxels using 3D-CNN and sparse voxels with sparse convolution is illustrated.

[0079] In the embodiment of the present application, the use of sparse representation can effectively reduce memory usage and calculation, which has greater advantages when processing higher resolution and a large number of point clouds. Finally, the sparse voxels will be fed into the U-Net based on sparse convolution, and the results will be converted back to the atomic level through nearest neighbor interpolation. In this way, the present invention can not only learn more efficient geometric representations and higher resolutions, but also speed up network calculations.

[0080] In some optional implementations of the embodiments of the present application, the Atom-ProteinQA model includes a topology perception module, and the steps of inputting protein structure data into the Atom-ProteinQA model for quality assessment operation to obtain a quality assessment result specifically include the following steps:

[0081] Constructing an atomic-level topological map corresponding to the protein structure data according to the topological awareness module;

[0082] The second atomic-level prediction result is output according to the atomic-level topological map and the message passing mechanism.

[0083] In the present application embodiment, compared with 3D-CNN, which excels in capturing geometric details through regular grids, graph convolution is more suitable for unstructured data, and protein and RNA structure scores are naturally unstructured. Therefore, the present invention designs a topology-aware (TP) module to capture the protein and RNA structure score topology by learning atomic-level chemical relationships, thereby enhancing features. Specifically, protein and RNA structure score molecules can be represented as graphs in and ε are vertex sets and edge sets respectively. The present invention uses the coordinates of atoms as nodes and constructs the edges of the graph through chemical bonds. Then graph convolution is used to update the features of atoms:

[0084]

[0085] where v i and is the feature of the i-th node before and after graph convolution. The i-th and j-th nodes (W ij ) is calculated by the distance and chemical bond between the two atoms. Specifically, B is only ij is equal to 1, otherwise it is equal to 0. In addition, this application uses the distance between two nodes (D ij ) as features and gives the spatial relationship of the model. In the equation, [·;·] is a series operation, is a learnable nonlinear transformation. This application uses the symbol N i To represent the adjacent index of node i.

[0086] By stacking multiple graph convolution layers, the present invention can gradually enhance the feature representation of atoms in protein and RNA structure scoring. After passing through several layers, a maximum pooling operation is used to aggregate the atomic features in each residue and obtain residue-level features. These residue-level features will be further updated through more graph convolution layers to ultimately generate residue-level LDDT scores. The present invention does not output atomic-level predictions in the TP module because a large amount of message passing at the atomic level is computationally intensive and ineffective.

[0087] In some optional implementations of the embodiments of the present application, the Atom-ProteinQA model includes a geometry perception module, a topology perception module, and a cross-model fusion module, and the protein structure data is input into the Atom-ProteinQA model for quality assessment operation to obtain the quality assessment result, which specifically includes the following steps:

[0088] Acquire atomic feature data of protein structure data according to the geometry perception module;

[0089] Obtaining an initial topological map of protein structure data according to a topological perception module;

[0090] Perform feature fusion operation on atomic features and topological maps according to the cross-model fusion module to obtain an updated topological map;

[0091] A quality prediction operation is performed on the updated topology map according to the topology perception module and the message passing mechanism to obtain a third atomic-level prediction result.

[0092] In the embodiment of the present application, the cross model fusion (CMF) model can realize the feature interaction between the GP and TP modules. It aims to utilize the skip connection from the GP module to the TP module at certain layer levels. In this way, the advantages of geometric representation can be integrated to improve the residual prediction of the TP module through an end-to-end training protocol. At the same time, the atomic-level predictions of the GP module can also be mutually enhanced through the back-propagation of the TP module. Specifically, for the features of the lth sparse convolutional layer of the GP module (expressed as ), the present invention converts it into the same representation level as the lth layer of the TP module. Here, the feature The first layer of the TP module is converted to the atomic level, and the residue level of the latter layer is converted to the atomic level through residual maximum pooling. and TP module They are element-wise added and a nonlinear transformation is performed to fuse them.

[0093]

[0094] The representation represents element-wise addition, is a learnable transformation. Finally, the updated feature Will be used in the l+1 layer of the TP module.

[0095] For training, we optimize Atom-ProteinQA using the mean squared error (MSE) loss function, which predicts the residue scores for protein and RNA structure scoring as:

[0096]

[0097] in represents the true value score of the nth atom, Represents atomic features, Indicates j th The true value score of the residues is J represents the bait protein th The set of atomic features of the residue. Represents the function of the model of this application, and θ represents a learnable parameter.

[0098] In practical applications, Table 1 shows the performance results of the present invention on CATH-2084 and Decoy-8000, among which Atom-ProteinQA performs best in all indicators. For non-learning-based methods, Ornate achieves the best performance. That is, without a data-driven approach, it cannot effectively capture the available information in the protein and RNA structure scoring structure. For learning-based methods, Vorocnn cannot obtain satisfactory results by utilizing 3D-CNN alone. Since it roughly represents the atomic features of each residue in a single voxel, it cannot capture the geometry of the local area of ​​the protein and RNA structure scoring. Although DeepAccNet uses additional supervision and labels for protein and RNA structure score refinement, they cannot predict more accurate prediction scores than the present application, especially at the local level, i.e., and. Accordingly, only graph-based message passing does not produce the best performance. Specifically, GNNRefine lacks the ability to extract fine-grained geometric features from fine-grained protein and RNA structure scoring structures. This application also retrained the DeepUMQA model using a training dataset from Decoy-8000, which is the same dataset used to train the model of this application.

[0099] In addition, this application conducted quality assessment (QA) tests on the Decoy-8000 test set and the entire CASP14 dataset. The results show that Deep-UMQA performs exceptionally well in terms of global scores. However, the model of this application surpasses DeepUMQA in terms of residual level scores. In contrast, Atom-ProteinQA has achieved great progress in all scores, especially in local indicators and, by combining geometric perception and chemical topology learning through GP and TP modules. It can be found that Atom-ProteinQA is 0.13 and 0.14 higher than ProteinGCN and DeepAccNet on CATH-2084, respectively. These significant gains mainly come from the enhanced features of joint learning, which are obtained through geometric capture of high-resolution sparse voxels and chemical relationship reference rings through atomic-level graphs. The visualization results of three real cases are shown in Figure 2. Figure 5 As shown, a qualitative analysis of ProteinQA for cases 2wcrA, 1d2zB, and 5hw3A in the Decoy-8000 test set is performed to compare the results of DeepAccNet, ProteinGCN, and Atom-ProteinQA. The gray line represents the true LDDT score, and the colored lines are the results of different methods.

[0100]

[0101] Table 1: Comparison of quantitative performance of models (CATH-2084 and Decoy-8000 datasets)

[0102] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0103] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0104] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0105] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0106] Further references Figure 6 , as a response to the above Figure 2 The present application provides an embodiment of a protein model quality assessment device, and the device embodiment is Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0107] like Figure 6 As shown, the protein model quality assessment device 200 of the embodiment of the present application includes:

[0108] The file acquisition module 210 is used to receive a PDB file input by a user terminal, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring;

[0109] A model building module 220, for building an Atom-ProteinQA model according to the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm;

[0110] A request acquisition module 230, configured to receive a quality assessment request sent by a user terminal, wherein the quality assessment request at least includes protein structure data to be assessed;

[0111] The quality assessment module 240 is used to input the protein structure data into the Atom-ProteinQA model to perform a quality assessment operation to obtain a quality assessment result.

[0112] In an embodiment of the present application, a protein model quality assessment device 200 is provided, including: a file acquisition module 210, for receiving a PDB file input by a user terminal, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring; a model construction module 220, for constructing an Atom-ProteinQA model according to the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm; a request acquisition module 230, for receiving a quality assessment request sent by a user terminal, wherein the quality assessment request at least includes the protein structure data to be evaluated; a quality assessment module 240, for inputting the protein structure data into the Atom-ProteinQA model for quality assessment operation to obtain a quality assessment result. Compared with the prior art, the present application uses the LDDT score at the atomic level of protein and RNA structure scoring as supervision, effectively performs accurate atomic-level IDDL score prediction, and evaluates the structural changes of protein and RNA structure scoring at an atomic level.

[0113] In some optional implementations of the embodiments of the present application, the Atom-ProteinQA model includes a geometric perception module, and the quality assessment module includes:

[0114] A feature extraction submodule is used to perform feature extraction operations on protein structure data to obtain voxelized Cartesian coordinates of each residue atom in the protein structure data;

[0115] The first prediction submodule is used to capture the geometric features of the voxelized Cartesian coordinates according to the geometric perception module and generate a first atomic-level prediction result.

[0116] In some optional implementations of the embodiments of the present application, the Atom-ProteinQA model includes a topology awareness module, and the quality assessment module includes:

[0117] A topology map construction submodule is used to construct an atomic-level topology map corresponding to the protein structure data according to the topology perception module;

[0118] The second prediction submodule is used to output a second atomic-level prediction result according to the atomic-level topology map and the message passing mechanism.

[0119] To solve the above technical problems, the present application also provides a computer device. Figure 7 , Figure 7 This is a basic structural block diagram of a computer device according to an embodiment of the present application.

[0120] The computer device 300 includes a memory 310, a processor 320, and a network interface 330 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 300 having components 310-330, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), digital signal processors (DSP), embedded devices, etc.

[0121] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0122] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 310 can be an internal storage unit of the computer device 300, such as a hard disk or memory of the computer device 300. In other embodiments, the memory 310 can also be an external storage device of the computer device 300, such as a plug-in hard disk equipped on the computer device 300, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card, etc. Of course, the memory 310 can also include both the internal storage unit of the computer device 300 and its external storage device. In the embodiment of the present application, the memory 310 is generally used to store the operating system and various application software installed on the computer device 300, such as computer-readable instructions of the protein model quality assessment method, etc. In addition, the memory 310 can also be used to temporarily store various data that have been output or are to be output.

[0123] The processor 320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 320 is generally used to control the overall operation of the computer device 300. In the embodiment of the present application, the processor 320 is used to run the computer-readable instructions stored in the memory 310 or process data, such as computer-readable instructions for running the protein model quality assessment method.

[0124] The network interface 330 may include a wireless network interface or a wired network interface. The network interface 330 is generally used to establish a communication connection between the computer device 300 and other electronic devices.

[0125] The computer device provided in this application uses the LDDT scores at the atomic level of protein and RNA structure scoring as supervision to effectively perform accurate atomic-level IDDL score predictions and evaluate structural changes in protein and RNA structure scoring at an atomic level.

[0126] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the protein model quality assessment method as described above.

[0127] The computer-readable storage medium provided in the present application utilizes the LDDT scores at the atomic level of protein and RNA structure scoring as supervision to effectively perform accurate atomic-level IDDL score predictions and evaluate structural changes in protein and RNA structure scoring at an atomic level.

[0128] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0129] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A protein model quality assessment method, characterized in that: The steps include: Receiving a PDB file input by a user terminal, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring; Constructing an Atom-ProteinQA model based on the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm; Receiving a quality assessment request sent by a user terminal, wherein the quality assessment request at least includes protein structure data to be assessed; The protein structure data is input into the Atom-ProteinQA model to perform a quality assessment operation to obtain a quality assessment result.

2. The protein model quality assessment method according to claim 1, characterized in that: The preset algorithm first searches for a radius of P n The atom P n [i,j]'s neighbors, then select P by the preset index m and P n The neighboring atoms in P are calculated separately. m [i,j] and P n The respective distances of [i,j], P m and P n The difference between the individual distances in will be passed through the mapping function converted to predefined scores, and finally, the LDDT scores at the atomic, residue, and global levels were obtained by normalizing by the total number of neighbors.

3. The protein model quality assessment method according to claim 1, characterized in that: The Atom-ProteinQA model includes a geometric perception module. The step of inputting the protein structure data into the Atom-ProteinQA model for quality assessment to obtain a quality assessment result specifically includes the following steps: Performing a feature extraction operation on the protein structure data to obtain voxelized Cartesian coordinates of each residue atom in the protein structure data; The geometric features of the voxelized Cartesian coordinates are captured by the geometric perception module to generate a first atomic-level prediction result.

4. The protein model quality assessment method according to claim 1, characterized in that: The Atom-ProteinQA model includes a topology perception module. The step of inputting the protein structure data into the Atom-ProteinQA model for quality assessment to obtain a quality assessment result specifically includes the following steps: constructing an atomic-level topological map corresponding to the protein structure data according to the topological perception module; A second atomic-level prediction result is output according to the atomic-level topology map and the message passing mechanism.

5. The protein model quality assessment method according to claim 1, characterized in that: The Atom-ProteinQA model includes a geometry perception module, a topology perception module and a cross-model fusion module. The step of inputting the protein structure data into the Atom-ProteinQA model for quality assessment to obtain a quality assessment result specifically includes the following steps: Acquire atomic feature data of the protein structure data according to the geometric perception module; Acquire an initial topological map of the protein structure data according to the topological perception module; Performing a feature fusion operation on the atomic features and the topological map according to the cross-model fusion module to obtain an updated topological map; A quality prediction operation is performed on the updated topology map according to the topology perception module and the message passing mechanism to obtain a third atomic-level prediction result.

6. A protein model quality assessment device, characterized in that: include: A file acquisition module, used to receive a PDB file input by a user terminal, wherein the PDB file contains the characteristics of each atom in the protein and RNA structure scoring; A model building module, used to build an Atom-ProteinQA model according to the characteristics of each atom in the protein and RNA structure scoring and a preset algorithm; A request acquisition module, configured to receive a quality assessment request sent by a user terminal, wherein the quality assessment request at least includes protein structure data to be assessed; The quality assessment module is used to input the protein structure data into the Atom-ProteinQA model to perform a quality assessment operation to obtain a quality assessment result.

7. The protein model quality assessment device according to claim 6, characterized in that: The Atom-ProteinQA model includes a geometric perception module, and the quality assessment module includes: A feature extraction submodule, used for performing a feature extraction operation on the protein structure data to obtain voxelized Cartesian coordinates of each residue atom in the protein structure data; The first prediction submodule is used to capture the geometric features of the voxelized Cartesian coordinates according to the geometric perception module to generate a first atomic-level prediction result.

8. The protein model quality assessment device according to claim 6, characterized in that: The Atom-ProteinQA model includes a topology perception module, and the quality assessment module includes: A topology map construction submodule, used to construct an atomic-level topology map corresponding to the protein structure data according to the topology perception module; The second prediction submodule is used to output a second atomic-level prediction result according to the atomic-level topology map and the message passing mechanism.

9. A computer device comprising a memory and a processor, characterized in that: The memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the protein model quality assessment method according to any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the protein model quality assessment method according to any one of claims 1 to 5 are implemented.

Images