Method and system for predicting dynamic structure change of protein A beta 42 based on deep learning method

Abstract

The invention provides a method set system for predicting dynamic structure change of protein A beta 42 based on a deep learning method. The method comprises the following steps: extracting PDB of each frame from a molecular dynamics track; the 2DIR of each PDB is calculated; splitting the PDB according to the functional characteristic region; an alpha-carbon distance matrix distance map of each PDB is calculated, and an alpha-carbon distance matrix distance map of each PDB is calculated; according to the trained deep learning model, predicting an alpha-carbon distance matrix distance map of each region; constructing a protein three-dimensional structure; merging fragments and comparing the fragments with a three-dimensional structure: aligning the optimized three PDB segments, and merging the three PDB segments into a complete PDB protein three-dimensional structure; and finally, solving the RMSD error between the predicted PDB and the real PDB. According to the method, a division and treatment strategy is adopted, three sections of different regions in a protein structure are predicted by utilizing 2DIR, and finally, a three-dimensional structure is successfully constructed. According to the method, a new method is provided for exploring the structure and dynamic change of the Abeta42, a new view angle for researching the Alzheimer's disease is opened up, and important information is also provided for drug design.

Description

Technical Field

[0001] The present invention relates to the technical field of structural research of the Alzheimer's disease-causing protein Aβ42, and specifically to a method and system for predicting the dynamic structural changes of the protein Aβ42 based on a deep learning method. Background Art

[0002] Alzheimer's disease (AD), also often referred to as senile dementia, is a progressive neurodegenerative disease that mainly affects the elderly. The disease initially manifests as mild memory loss, and as the disease progresses, it gradually affects language ability, cognitive function, judgment and daily living ability. Patients may completely lose the ability to take care of themselves in the late stage of the disease. According to statistics from the World Health Organization, about 55 million people in the world suffer from AD, and the number of patients is growing at a rate of nearly 10 million new cases per year. It is estimated that by 2050, this number may rise to 139 million. Moreover, AD is regarded as the "grave" of new drug research and development in the field. The clinical failure rate in this field is as high as 99.6%. Pharmaceutical giants such as Johnson & Johnson, Pfizer, Eli Lilly, and Roche have all failed here. So far, there are very limited AD drugs on the market.

[0003] The pathogenesis of Alzheimer's disease is closely related to the misfolding and aggregation of β-amyloid protein (Aβ). Among them, Aβ42 is a particularly important β-amyloid protein fragment, which is directly related to the development of AD. Aβ42 is a longer amyloid protein fragment obtained by enzymatic cleavage from a larger precursor protein (Amyloid precursor protein, APP). Compared with Aβ fragments of other lengths, such as Aβ40, Aβ42 is more likely to aggregate in the brain to form so-called "plaques". These plaques are a major pathological feature of AD and are also considered to be the main driving factor in the development of the disease. Therefore, it is very important to study the dynamic structural changes of Aβ42. Understanding the dynamic structural changes of Aβ42 is of great significance for the development of prevention and treatment strategies for Alzheimer's disease. This can not only help researchers identify potential intervention points, but also provide important information for drug design.

[0004] Although traditional structural biology techniques such as X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy, and cryo-electron microscopy (Cryo-EM) are very effective in studying fixed structures, they are often technically limited in analyzing the dynamic processes and incompletely folded states of proteins. For example, NMR has strict requirements on sample quantity and state, X-ray crystallography relies on difficult-to-obtain protein crystals, and Cryo-EM, although applicable to large complexes, has room for improvement in resolving detailed information on the dynamic processes of certain proteins. Therefore, new methods are needed to study the dynamic changes of Aβ42.

[0005] Two-dimensional infrared spectroscopy (2DIR spectroscopy) is an advanced spectroscopic technique similar to the more widely known two-dimensional nuclear magnetic resonance (2D NMR) technique. It involves applying a series of infrared pulses to a sample and generating a two-dimensional spectrum by measuring the interactions between the pulses and the sample's response to these pulses. This two-dimensional spectrum shows the coupling relationship between different vibrational modes and can reveal intra- and inter-molecular interactions, especially those that are difficult to distinguish in traditional IR spectroscopy. Therefore, when exploring dynamically changing and complex biomolecules such as Aβ42, two-dimensional infrared spectroscopy (2DIR) provides a powerful alternative method with its unique time resolution and sensitivity to intermolecular interactions. 2DIR can capture instant changes in protein structure at the atomic level, enabling this example to gain a deeper understanding of the aggregation process and pathological mechanisms of these biomolecules, thereby providing important structural information for the study of related diseases and drug development.

[0006] However, due to the high dimensionality and complexity of the data generated by two-dimensional infrared spectroscopy (2DIR), its interpretation usually requires extensive expertise and technical experience. Summary of the invention

[0007] The technical problem to be solved by the present invention is that the interpretation of two-dimensional infrared spectroscopy (2DIR) is currently highly dependent on expert experience, and the data processing efficiency and accuracy need to be improved.

[0008] The present invention solves the above technical problems through the following technical means:

[0009] The method for predicting the dynamic structural changes of protein Aβ42 based on deep learning method includes the following steps:

[0010] Step 1: Extract the PDB of each frame from the molecular dynamics trajectory;

[0011] Step 2, calculate the 2DIR of each PDB;

[0012] Step 3: Split the PDB according to functional feature areas;

[0013] Step 4, calculate the α-carbon distance matrix distance map of each PDB;

[0014] Step 5: Predict the α-carbon distance map of each region based on the trained deep learning model;

[0015] Step 6: construct the three-dimensional structure of the protein;

[0016] Step 7, fragment merging and comparison with three-dimensional structure: align the three optimized PDB segments and merge them into a complete PDB protein three-dimensional structure; finally, calculate the RMSD error between the predicted PDB and the true PDB.

[0017] Furthermore, the step 2 specifically includes the following steps:

[0018] Step 2.1 Configure and run molecular dynamics simulations to prepare dynamic data for proteins: First, configure computing resources by loading appropriate GROMACS modules and setting up Slurm job scripts; then, create separate directories for each PDB file and move the files to the corresponding directories; generate the protein topology using PDB2gmx, set the simulation box using editconf, add solvent using solvate, and add ions to neutralize the system charge using genion; after completing these steps, prepare the input files required for the simulation using grompp and submit them to the computing cluster to execute the simulation;

[0019] Step 2.2 Generate Hamiltonian and Dipoles files: Use the NISE software package to process the data obtained from the molecular dynamics simulation to generate Hamiltonian and dipole moment files for 2DIR spectral simulations; ensure that these files are suitable for spectral calculations by copying and adjusting them in the directory of each simulation; in order to improve the statistical accuracy and data quality of the 2DIR simulation, Hamiltonian and Dipoles files are copied and merged multiple times, which can increase the data volume and optimize the results of subsequent spectral analysis;

[0020] Step 2.3 Run 2DIR simulation and data conversion: After loading the compiler and MPI library, run the NISE software to perform 1DIR and 2DIR spectral simulations; this process includes reading the adjusted Hamiltonian and Dipoles files, performing spectral simulations, and converting the results from text format to binary format for subsequent processing;

[0021] Step 2.4 Image processing and data organization: The generated spectral images were processed using Python scripts, including image generation, cropping, and resizing to meet analysis requirements; finally, the processed images and data files were renamed and organized to ensure that they could be efficiently stored and accessed.

[0022] Furthermore, the step 3 specifically includes the following steps:

[0023] The PDB of Aβ42 was split into PDB of 1-17 residues, PDB of 16-28 residues, and PDB of 27-42 residues.

[0024] Furthermore, the step 4 specifically includes the following steps:

[0025] Step 4.1 Atom positioning: First, locate all α-carbon atoms from the PDB file;

[0026] Step 4.2: extracting coordinates: extracting the x, y, z spatial coordinates of the α-carbon atom;

[0027] Step 4.3 Calculate the distance: For each pair of α-carbon atoms in the coordinates, use the Euclidean distance formula to calculate the distance between the α-carbon atoms: The distance formula is

[0028]

[0029] Step 4.4 Build a matrix: Organize all distance values ​​into a matrix, where the rows and columns of the matrix correspond to the indices of the α-carbon atoms.

[0030] Furthermore, the step 6 specifically includes the following steps:

[0031] Step 6.1 Selection of initial points: At the beginning of the algorithm, a set of initial coordinates is randomly generated as the starting point of the optimization process; this random initialization method is often used to ensure that the optimization process can explore the solution space from a variety of possible configurations and reduce the risk of the solution being affected by the initial conditions;

[0032] Step 6.2 Optimization process: Through an iterative method, the positions of these points are continuously adjusted to minimize the objective function; the objective function measures the deviation between the α-carbon distance matrix calculated based on the current coordinates and the constraint condition, that is, the predicted α-carbon distance matrix distance map;

[0033] Interpretation of the results of step 6.3: After the optimization is completed, the obtained coordinates should be as close to the real three-dimensional structure as possible while satisfying the given distance constraints.

[0034] The present invention also provides a system for predicting the dynamic structural changes of protein Aβ42 based on a deep learning method, comprising:

[0035] PDB data extraction module, used to extract the PDB of each frame from the molecular dynamics trajectory;

[0036] 2DIR calculation module, used to calculate the 2DIR of each PDB;

[0037] The region splitting module is used to split the PDB according to functional characteristics.

[0038] α-carbon distance matrix calculation module, used to calculate the α-carbon distance matrix distance map of each PDB;

[0039] The prediction module is used to predict the α-carbon distance map of each region based on the trained deep learning model;

[0040] Three-dimensional structure building blocks, used to construct protein three-dimensional structures;

[0041] The error calculation module is used to merge the fragments and compare them with the three-dimensional structure: the three optimized PDB fragments are aligned and merged into a complete PDB protein three-dimensional structure; finally, the RMSD error between the predicted PDB and the real PDB is calculated.

[0042] Furthermore, the 2DIR calculation module specifically includes the following steps:

[0043] Step 2.1 Configure and run molecular dynamics simulations to prepare dynamic data for proteins: First, configure computing resources by loading appropriate GROMACS modules and setting up Slurm job scripts; then, create separate directories for each PDB file and move the files to the corresponding directories; generate the protein topology using PDB2gmx, set the simulation box using editconf, add solvent using solvate, and add ions to neutralize the system charge using genion; after completing these steps, prepare the input files required for the simulation using grompp and submit them to the computing cluster to execute the simulation;

[0044] Step 2.2 Generate Hamiltonian and Dipoles files: Use the NISE software package to process the data obtained from the molecular dynamics simulation to generate Hamiltonian and dipole moment files for 2DIR spectral simulations; ensure that these files are suitable for spectral calculations by copying and adjusting them in the directory of each simulation; in order to improve the statistical accuracy and data quality of the 2DIR simulation, Hamiltonian and Dipoles files are copied and merged multiple times, which can increase the data volume and optimize the results of subsequent spectral analysis;

[0045] Step 2.3 Run 2DIR simulation and data conversion: After loading the compiler and MPI library, run the NISE software to perform 1DIR and 2DIR spectral simulations; this process includes reading the adjusted Hamiltonian and Dipoles files, performing spectral simulations, and converting the results from text format to binary format for subsequent processing;

[0046] Step 2.4 Image processing and data organization: The generated spectral images were processed using Python scripts, including image generation, cropping, and resizing to meet analysis requirements; finally, the processed images and data files were renamed and organized to ensure that they could be efficiently stored and accessed.

[0047] Furthermore, the region splitting module specifically includes the following steps:

[0048] The PDB of Aβ42 was split into PDB of 1-17 residues, PDB of 16-28 residues, and PDB of 27-42 residues.

[0049] Furthermore, the α-carbon distance matrix calculation module specifically includes the following steps:

[0050] Step 4.1 Atom positioning: First, locate all α-carbon atoms from the PDB file;

[0051] Step 4.2: extracting coordinates: extracting the x, y, z spatial coordinates of the α-carbon atom;

[0052] Step 4.3 Calculate the distance: For each pair of α-carbon atoms in the coordinates, use the Euclidean distance formula to calculate the distance between the α-carbon atoms: The distance formula is

[0053]

[0054] Step 4.4 Build a matrix: Organize all distance values ​​into a matrix, where the rows and columns of the matrix correspond to the indices of the α-carbon atoms.

[0055] Furthermore, the three-dimensional structure building module specifically includes the following steps:

[0056] Step 6.1 Selection of initial points: At the beginning of the algorithm, a set of initial coordinates is randomly generated as the starting point of the optimization process; this random initialization method is often used to ensure that the optimization process can explore the solution space from a variety of possible configurations and reduce the risk of the solution being affected by the initial conditions;

[0057] Step 6.2 Optimization process: Through an iterative method, the positions of these points are continuously adjusted to minimize the objective function; the objective function measures the deviation between the α-carbon distance matrix calculated based on the current coordinates and the constraint condition, that is, the predicted α-carbon distance matrix distance map;

[0058] Interpretation of the results of step 6.3: After the optimization is completed, the obtained coordinates should be as close to the real three-dimensional structure as possible while satisfying the given distance constraints.

[0059] The advantages of the present invention are:

[0060] The present invention predicts the α-carbon distance matrix of proteins based on deep learning methods, adopts a divide-and-conquer strategy, uses 2DIR to predict three different regions in the protein structure, and finally successfully constructs a three-dimensional structure. This method provides a new method for exploring the structure and dynamic changes of Aβ42, opens up a new perspective for studying Alzheimer's disease, and also provides important information for drug design. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a flow chart of a method for predicting the three-dimensional structure of Aβ42 based on a deep learning method in an embodiment of the present invention;

[0062] Figure 2 A deep learning workflow diagram in a method for predicting the three-dimensional structure of Aβ42 based on a deep learning method in an embodiment of the present invention;

[0063] Figure 3 The results of the method in the embodiment of the present invention are shown in part (the RMSD of the three figures from top to bottom are );

[0064] Figure 4 A network structure diagram in an embodiment of the present invention;

[0065] Figure 5 FIG. 4 is a logic principle diagram of MAE in an embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] This embodiment predicts the three-dimensional structure of Aβ42 based on a deep learning method, such as Figure 1As shown, proceed as follows:

[0068] Step 1: Extract the PDB of each frame from the molecular dynamics trajectory.

[0069] To extract the PDB file for each frame from a molecular dynamics simulation trajectory, use a Python library such as MDAnalysis. First, you need to load the trajectory and topology files, then iterate through each frame in the trajectory, handle periodic boundary problems, and save the structure of each frame in PDB format. This process involves initializing the simulation system, selecting an appropriate force field, setting the time step, and saving the data for each frame in an appropriate output directory to facilitate further structural analysis and visualization. This approach not only automates the data extraction process, but also ensures that the data is organized and stored in a format that is easy to process later.

[0070] Step 2: Calculate the 2DIR for each PDB.

[0071] Use supercomputing resources to perform step-by-step calculations.

[0072] Step 2.1 Molecular dynamics simulation preparation and execution. In the first step, this embodiment configures and runs molecular dynamics simulation to prepare dynamic data of the protein. First, configure computing resources by loading the appropriate GROMACS module and setting up the Slurm job script. Then, create a separate directory for each PDB file and move the files to the corresponding directory. Generate the topological structure of the protein using PDB2gmx, set the simulation box type using editconf, add solvent using solvate, and add ions to neutralize the system charge through genion. After completing these steps, use grompp to prepare the input files required for the simulation and submit them to the computing cluster to perform the simulation.

[0073] Step 2. Optimize the results of subsequent spectral analysis.

[0074] Step 2.3 runs 2DIR simulation and data conversion. In the third step, after loading necessary compilers and MPI libraries, run NISE software to perform 1DIR and 2DIR spectral simulations. This process includes reading the adjusted Hamiltonian and Dipoles files, performing spectral simulations, and generating Hamiltonian and Dipoles files from document 2 with the result. In this step, the present embodiment uses NISE (Nonlinear Infrared Spectroscopic Simulations) software package to process the data obtained from molecular dynamics simulations to generate Hamiltonian (Hamiltonian) and dipole moment (Dipoles) files for two-dimensional infrared (2DIR) spectral simulations. By copying and adjusting these files in the directory of each simulation, the present embodiment ensures that they are suitable for spectral calculations. In order to improve the statistical accuracy and data quality of 2DIR simulations, these Hamiltonian and Dipoles files can be copied and merged many times, so that the amount of data can be increased and the format is converted to binary format to facilitate subsequent processing. This step is the key link in generating spectral data that can be used for further analysis.

[0075] Step 2.4 Image processing and data organization. The last step focuses on generating and processing the image data of the 2DIR spectrum. The generated spectral images are processed using Python scripts, including image generation, cropping, and resizing to meet analysis requirements. Finally, the processed images and data files are renamed and organized to ensure that they can be efficiently stored and accessed, providing researchers with accurate spectral analysis results.

[0076] Step 3: Split the PDB according to functional feature areas.

[0077] Using a divide-and-conquer strategy, the PDB of Aβ42 was split into PDB of residues 1-17, PDB of residues 16-28, and PDB of residues 27-42, because Aβ42 can be divided into three regions: residues 1-16, 17-28, and 29-42, each with unique structural and functional properties. The residue 1-16 region is the N-terminal with relatively high polarity, which usually plays a role in the interaction between peptides and cell membranes or water-soluble molecules and has a flexible structure; the residue 17-28 region contains a key central hydrophobic region, and the hydrophobicity of this region is the main driving force for Aβ42 aggregation and fiber formation, and tends to form a β-folded structure; residues 29-42 are the C-terminal region, and the addition of this region significantly enhances the aggregation tendency of Aβ42, making it more inclined to form a stable fiber structure than Aβ40. Overall, the synergistic effect of these three fragments leads to the high aggregation of Aβ42 and the neurotoxicity associated with Alzheimer's disease.

[0078] Step 4: Calculate the α-carbon distance map of each PDB.

[0079] The α-carbon distance matrix is ​​an N×N matrix, where N is the number of amino acid residues in a protein. Each element of the matrix, dij, represents the Euclidean distance between the α-carbon atoms of the i-th and j-th amino acid residues. The α-carbon atom is part of each amino acid residue in a protein and is located on the main chain of the amino acid. In protein structure analysis, calculating the distance matrix between α-carbon atoms is a common method used to evaluate the three-dimensional structure and relative spatial arrangement of proteins.

[0080] Step 4.1 Atom location. First, locate all α-carbon atoms from the PDB file. This typically involves parsing the file and searching for lines with specific atom names (such as "CA").

[0081] Step 4.2 Extract coordinates. Extract the x, y, z spatial coordinates of these α-carbon atoms.

[0082] Step 4.3 Calculate the distance. For each pair of α-carbon atoms in the coordinates, use the Euclidean distance formula to calculate the distance between them. The distance formula is

[0083]

[0084] Step 4.4: Build a matrix. Organize all distance values ​​into a matrix, where the rows and columns of the matrix correspond to the index of the α-carbon atom. This distance matrix is ​​very helpful for understanding the three-dimensional configuration of proteins, especially when performing structural comparison, folding pattern recognition, or dynamic simulation analysis.

[0085] Step 5: Predict the α-carbon distance map of each fragment based on the trained deep learning model.

[0086] According to the trained model, the 2DIR of the test set is input to predict the α-carbon distance matrix distance map of the corresponding three regions.

[0087] Step 6: Construct the three-dimensional structure of the protein.

[0088] The "predicted α-carbon distance map" can be considered as a constraint to guide the generation of a three-dimensional structure. Specifically, this constraint defines the distance between atoms or molecules, and the task of the optimization algorithm is to adjust the spatial coordinates of these atoms so that the distance matrix generated by these coordinates conforms to these constraints as much as possible.

[0089] Step 6.1 Selection of initial points. At the beginning of the algorithm, a set of initial coordinates is randomly generated, which are the starting point of the optimization process. This random initialization method is often used to ensure that the optimization process can explore the solution space from a variety of possible configurations and reduce the risk of the solution being affected by the initial conditions.

[0090] Step 6.2 Optimization process. Through an iterative method, the positions of these points are continuously adjusted to minimize the objective function. The objective function here measures the deviation between the α-carbon distance matrix calculated based on the current coordinates and the constraints (i.e., the predicted α-carbon distance matrix distance map). Optimization method: The L-BFGS-B algorithm used is an efficient local search algorithm that updates the coordinate position by approximating the Hessian matrix to seek a configuration that minimizes the objective function.

[0091] Step 6.3 Interpretation of the results. After the optimization is completed, the coordinates obtained should be as close to the real three-dimensional structure as possible while satisfying the given distance constraints. However, it should be noted that due to the existence of local minima and the limitations of the optimization method, the obtained structure may not be the global optimal solution, but an approximate solution.

[0092] Step 7: Merge the fragments and compare with the three-dimensional structure.

[0093] The three optimized PDBs are aligned and merged into a complete PDB (protein three-dimensional structure). Finally, the RMSD error between the predicted PDB and the real PDB is calculated. Figure 3 As shown, Figure 3 The three figures in the middle show the effects of different RMSD errors.

[0094] The working principle of this embodiment is as follows:

[0095] The Swin Transformer pre-trained model is used and fine-tuned on this task to establish the spectrum-structure relationship between the descriptor of the two-dimensional infrared spectrum (2DIR) and the α-carbon distance matrix distance map, and then the three-dimensional structure is constructed through the predicted α-carbon distance matrix distance map. The following is the workflow of deep learning.

[0096] Step 1: Data collection and analysis.

[0097] To build a powerful and transferable deep learning (DL) model, the first task is to create a comprehensive and high-quality dataset. The molecular dynamics trajectory data used in this invention are derived from published studies, see the following references for details: Heller GT, Aprile FA, Michaels TCT, et al. Small-molecule sequestration of amyloid-β as a drug discovery strategy for Alzheimer's disease [J]. Science advances, 2020, 6 (45): eabb5924.

[0098] From the above references, we know that an advanced integration method is used to combine nuclear magnetic resonance (NMR) chemical shift data with molecular dynamics simulations to accurately study the dynamic behavior of the Aβ42 monomer state. In this process, metadynamics meta-inference technology is used, which can not only compensate for the inaccuracy of the force field, but also incorporate errors in experimental data and significantly enhance the sampling efficiency of the system. GROMACS2018.3 software is used with the PLUMED library 2.6.0-dev version for simulation, and the CHARMM22* force field and TIP3P water model are selected to ensure the reliability and accuracy of the simulation.

[0099] The initial simulation steps involved the generation of a linear peptide conformation of Aβ42 using PyMOL software, followed by a 1 ns dynamics simulation in vacuum to collapse the extended conformation of the peptide chain. Afterwards, the structure was dissolved in a rhombic dodecahedron box of volume 362 nm3 containing 11,746 water molecules, and the system was energy minimized using the steepest descent method to achieve stability. Next, a preliminary simulation of 2 ns was performed in the NVT ensemble at 600 K, from which 48 initial conformations were extracted for subsequent analysis. Subsequent simulations included 500 ps of equilibration in the NVT and NPT ensembles at 278 K, using the Bussi-Donadio-Parrinello thermostat and Berendsen pressure coupling techniques, while positional constraints were imposed on heavy atoms to ensure the accuracy of the simulations.

[0100] During the production run, long simulations were performed in the NPT ensemble using the Parrinello-Rahman manometer with a fixed time step of 2 fs. LINCS constraints were applied to all bonds, while the van der Waals forces were cut off at 1.2 nm and the electron interactions were treated using the particle mesh Ewald method.

[0101] In addition, this example also uses CamShift software to back-calculate the chemical shift at each time step to verify the accuracy of the simulation data. Since the error of the CamShift predictor may be greater than the chemical shift perturbation after the addition of the compound, this example uses the same chemical shift to constrain the Aβ42 monomer simulation. In meta-inference, a Gaussian noise model is used and an error parameter is set for each nuclear type, and an uninformative Jeffreys prior is used to process the error parameter.

[0102] In order to enhance the conformational sampling of Aβ42, this example uses parallel biased metadynamics with moderately mild and multi-walker protocols, and sets multiple collective variables (CVs). Six CVs were used in the unbound simulation, while 14 CVs were included in the bound simulation to enhance the contact sampling between the compound and the peptide, and four CVs were included to enhance the sampling of soft dihedral angles in small molecules. These simulations were performed for 27.8 microseconds until the data converged, and a total of 252,816 frames of three-dimensional structure were saved in the form of PDB. Overall, this series of complex simulation and analysis steps provides this example with detailed dynamic information about the Aβ42 monomer, which is of great significance for understanding its biological function and possible disease mechanisms.

[0103] Finally, all PDBs were converted into 2DIRs and the corresponding α-carbon distance matrix distance map was calculated, resulting in a comprehensive and high-quality dataset.

[0104] Step 2: Data preprocessing.

[0105] The 2DIR data are preprocessed before the descriptors are fed into the deep neural network learning training process.

[0106] Step 2.1 uses Python's Pillow library to automatically crop transparent boundaries in PNG images. It first loads the image and converts it to RGBA mode, then assumes the pixel color in the upper left corner of the image is the boundary color. It traverses each pixel to determine the actual boundaries of the main content of the image, and finally crops and saves the image based on these boundaries. By cropping the transparent boundaries in the image, the interference of non-critical areas is reduced, allowing the model to focus more on learning the effective content in the image. This can improve the model's ability to recognize key features of the image and reduce the risk of noise and overfitting during training and inference.

[0107] Step 2.2 scales the large image to a smaller size. Use the resize method to resize the image to 224x224 pixels. This is the standard input size for many commonly used deep learning models, such as convolutional neural networks. Such a standardization operation ensures that all input data has a consistent shape and size, which is beneficial to the training efficiency and performance of the model.

[0108] Together, these processing steps optimize the quality and consistency of the training data, allowing deep learning models to be trained on cleaner, more standardized datasets, thereby improving the model's generalization and prediction accuracy. In short, these image preprocessing techniques help improve the performance of models in real-world applications, reduce computational costs, speed up training, and improve the results of the final model.

[0109] Step 3: Model selection and construction.

[0110] The Swin Transformer model can be found as pre-trained models in deep learning frameworks like PyTorch or TensorFlow, and can be used through libraries like timm (PyTorch Image Models).

[0111] In order to select a deep learning model suitable for this task, this embodiment compares a group of common deep learning pre-trained models, including ResNet, U-Net, VGG, DenseNet, Inception, etc. However, the effects of these pre-trained models are not good.

[0112] The Swin Transformer can effectively extract features of different scales through its multi-level structure, which is particularly important for spectral data such as 2DIR, which can show molecular vibration information at multiple scales. This hierarchical processing helps capture spectral features from local to global, each of which may correspond to different interactions and functional domains in the molecular structure. By limiting the scope of self-attention to local windows, the Swin Transformer improves computational efficiency, which enables the model to process large-scale 2DIR datasets without sacrificing performance. The shifted window mechanism of the Swin Transformer increases the model's ability to integrate information across windows, allowing it to better understand local features in a global context. In the conversion of 2DIR to α-carbon distance matrix, this means that the model can better associate spectral features with spatial structural features represented in the distance matrix. At the same time, the architecture of the Swin Transformer supports inputs of various sizes, making it flexibly applicable to 2DIR spectral data of different resolutions. In addition, the Swin Transformer can learn general features on a large amount of data through pre-training, and then fine-tune for specific tasks, which is particularly effective in processing biological macromolecular structure predictions with complex patterns.

[0113] Step 4: Model training and evaluation.

[0114] like Figure 4 As shown, the model first loads weights from the pre-trained Swin Transformer and uses this powerful visual Transformer model to extract primary features of images (or other forms of input data). Swin Transformer was selected because it can process large-scale visual data and is particularly suitable for complex feature extraction tasks.

[0115] Feature upsampling and convolution processing segmentation_head is the core part of the model, which is responsible for further transforming the output features of SwinTransformer to adapt to the final task goal - predicting the α-carbon distance matrix. This part gradually expands the size of the feature map through multi-level upsampling and convolution layers, and finally reaches the required dimension to match the size of the protein distance matrix. Segmentation_head specifically includes:

[0116] Upsampling and convolutional layers: The upsampling method using bilinear interpolation gradually increases the size of the feature map, and convolutional layers are used after each upsampling stage to further process the features.

[0117] Batch Normalization and Activation Function: Batch normalization and Mish activation function are used after some convolutional layers, which helps the training stability and nonlinear feature expression of the model.

[0118] Self-attention layer: Introducing the self-attention layer in the feature processing flow enhances the model's ability to capture global dependencies, which is particularly important for understanding complex protein structures.

[0119] Training related methods

[0120] Weight initialization: The init_weights method is used to initialize the network weights, especially the Kaiming initialization is used for the convolutional layer, which helps prevent the gradient from disappearing or exploding in the early stages of training.

[0121] Freezing and unfreezing of layers: The methods freeze_swin_layers and unfreeze_swin_layers are provided to allow specific layers of the Swin Transformer to be frozen or unfrozen dynamically during training, which helps to improve the flexibility of model training and optimize the effect.

[0122] During the training process, the data of features and target variables are divided into training sets and validation sets. These data are strictly screened and cleaned to ensure the quality and accuracy of the data. Cross-validation is used to train the model to ensure the generalization ability of the model to unknown data.

[0123] MAE can quantitatively determine the agreement between the predicted α-carbon distance matrix and the true α-carbon distance matrix. When it comes to predicted and true matrices, such as the contact map (distance matrix) in a predicted protein structure, MAE can be applied to each element in the matrix, such as Figure 5 shown.

[0124] Step 5: Deployment and use of the model.

[0125] Save the trained model and load it directly when using it. Use the trained model to input the 2DIR of the protein and predict the α-carbon distance matrix of the three segments of the protein.

[0126] The key to this embodiment is to propose a deep learning-based model for predicting the three-dimensional structure of Aβ42 protein. The model uses 2DIR (two-dimensional infrared spectroscopy) data as input, and achieves high-precision prediction of Aβ42 protein structure by learning structural information and spectral features from a large number of data sets.

[0127] The technology of the present invention is not limited to the structural prediction of Aβ42 protein, but can also be applied to the structural prediction of other similar proteins, including but not limited to the conformation prediction of other proteins, protein-small molecule interaction simulation and other fields.

[0128] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the dynamic structural changes of protein Aβ42 based on a deep learning method, characterized in that: The following steps are involved: Step 1: Extract the PDB of each frame from the molecular dynamics trajectory; Step 2, calculate the 2DIR of each PDB; Step 3: Split the PDB according to functional feature areas; Step 4, calculate the α-carbon distance matrix distance map of each PDB; Step 5: Predict the α-carbon distance map of each region based on the trained deep learning model; Step 6: construct the three-dimensional structure of the protein; Step 7, fragment merging and comparison with three-dimensional structure: align the three optimized PDB segments and merge them into a complete PDB protein three-dimensional structure; finally, calculate the RMSD error between the predicted PDB and the true PDB.

2. The method for predicting the dynamic structural changes of protein Aβ42 based on a deep learning method according to claim 1, characterized in that: The step 2 specifically includes the following steps: Step 2.1 Configure and run molecular dynamics simulations to prepare dynamic data for proteins: First, configure computing resources by loading appropriate GROMACS modules and setting up Slurm job scripts; then, create separate directories for each PDB file and move the files to the corresponding directories; generate the protein topology using PDB2gmx, set the simulation box using editconf, add solvent using solvate, and add ions to neutralize the system charge using genion; after completing these steps, prepare the input files required for the simulation using grompp and submit them to the computing cluster to execute the simulation; Step 2.2 Generate Hamiltonian and Dipoles files: Use the NISE software package to process the data obtained from the molecular dynamics simulation to generate Hamiltonian and dipole moment files for 2DIR spectral simulations; ensure that these files are suitable for spectral calculations by copying and adjusting them in the directory of each simulation; in order to improve the statistical accuracy and data quality of the 2DIR simulation, Hamiltonian and Dipoles files are copied and merged multiple times, which can increase the data volume and optimize the results of subsequent spectral analysis; Step 2.3 Run 2DIR simulation and data conversion: After loading the compiler and MPI library, run the NISE software to perform 1DIR and 2DIR spectral simulations; this process includes reading the adjusted Hamiltonian and Dipoles files, performing spectral simulations, and converting the results from text format to binary format for subsequent processing; Step 2.4 Image processing and data organization: The generated spectral images were processed using Python scripts, including image generation, cropping, and resizing to meet analysis requirements; finally, the processed images and data files were renamed and organized to ensure that they could be efficiently stored and accessed.

3. The method for predicting the dynamic structural changes of protein Aβ42 based on a deep learning method according to claim 1, characterized in that: The step 3 specifically includes the following steps: The PDB of Aβ42 was split into PDB of 1-17 residues, PDB of 16-28 residues, and PDB of 27-42 residues.

4. The method for predicting the dynamic structural changes of protein Aβ42 based on a deep learning method according to claim 1, characterized in that: The step 4 specifically comprises the following steps: Step 4.1 Atom positioning: First, locate all α-carbon atoms from the PDB file; Step 4.2: extracting coordinates: extracting the x, y, z spatial coordinates of the α-carbon atom; Step 4.3 Calculate the distance: For each pair of α-carbon atoms in the coordinates, use the Euclidean distance formula to calculate the distance between the α-carbon atoms: The distance formula is Step 4.4 Build a matrix: Organize all distance values ​​into a matrix, where the rows and columns of the matrix correspond to the indices of the α-carbon atoms.

5. The method for predicting the dynamic structural changes of protein Aβ42 based on a deep learning method according to any one of claims 1 to 4, characterized in that: The step 6 specifically comprises the following steps: Step 6.1 Selection of initial points: At the beginning of the algorithm, a set of initial coordinates is randomly generated as the starting point of the optimization process; this random initialization method is often used to ensure that the optimization process can explore the solution space from a variety of possible configurations and reduce the risk of the solution being affected by the initial conditions; Step 6.2 Optimization process: Through an iterative method, the positions of these points are continuously adjusted to minimize the objective function; the objective function measures the deviation between the α-carbon distance matrix calculated based on the current coordinates and the constraint condition, that is, the predicted α-carbon distance matrix distance map; Interpretation of the results of step 6.3: After the optimization is completed, the obtained coordinates should be as close to the real three-dimensional structure as possible while satisfying the given distance constraints.

6. A system for predicting dynamic structural changes of protein Aβ42 based on deep learning methods, characterized in that: include: PDB data extraction module, used to extract the PDB of each frame from the molecular dynamics trajectory; 2DIR calculation module, used to calculate the 2DIR of each PDB; The region splitting module is used to split the PDB according to functional characteristics. α-carbon distance matrix calculation module, used to calculate the α-carbon distance matrix distance map of each PDB; The prediction module is used to predict the α-carbon distance matrix distancemap of each region based on the trained deep learning model; Three-dimensional structure building blocks, used to construct protein three-dimensional structures; The error calculation module is used to merge the fragments and compare them with the three-dimensional structure: the three optimized PDB fragments are aligned and merged into a complete PDB protein three-dimensional structure; finally, the RMSD error between the predicted PDB and the real PDB is calculated.

7. The system for predicting the dynamic structural changes of protein Aβ42 based on deep learning method according to claim 6, characterized in that: The 2DIR calculation module specifically includes the following steps: Step 2.1 Configure and run molecular dynamics simulations to prepare dynamic data for proteins: First, configure computing resources by loading appropriate GROMACS modules and setting up Slurm job scripts; then, create separate directories for each PDB file and move the files to the corresponding directories; generate the protein topology using PDB2gmx, set the simulation box using editconf, add solvent using solvate, and add ions to neutralize the system charge using genion; after completing these steps, prepare the input files required for the simulation using grompp and submit them to the computing cluster to execute the simulation; Step 2.2 Generate Hamiltonian and Dipoles files: Use the NISE software package to process the data obtained from the molecular dynamics simulation to generate Hamiltonian and dipole moment files for 2DIR spectral simulations; ensure that these files are suitable for spectral calculations by copying and adjusting them in the directory of each simulation; in order to improve the statistical accuracy and data quality of the 2DIR simulation, Hamiltonian and Dipoles files are copied and merged multiple times, which can increase the data volume and optimize the results of subsequent spectral analysis; Step 2.3 Run 2DIR simulation and data conversion: After loading the compiler and MPI library, run the NISE software to perform 1DIR and 2DIR spectral simulations; this process includes reading the adjusted Hamiltonian and Dipoles files, performing spectral simulations, and converting the results from text format to binary format for subsequent processing; Step 2.4 Image processing and data organization: The generated spectral images were processed using Python scripts, including image generation, cropping, and resizing to meet analysis requirements; finally, the processed images and data files were renamed and organized to ensure that they could be efficiently stored and accessed.

8. The system for predicting the dynamic structural changes of protein Aβ42 based on deep learning method according to claim 1, characterized in that: The region splitting module specifically includes the following steps: The PDB of Aβ42 was split into PDB of 1-17 residues, PDB of 16-28 residues, and PDB of 27-42 residues.

9. The system for predicting the dynamic structural changes of protein Aβ42 based on deep learning method according to claim 6, characterized in that: The α-carbon distance matrix calculation module specifically includes the following steps: Step 4.1 Atom positioning: First, locate all α-carbon atoms from the PDB file; Step 4.2: extracting coordinates: extracting the x, y, z spatial coordinates of the α-carbon atom; Step 4.3 Calculate the distance: For each pair of α-carbon atoms in the coordinates, use the Euclidean distance formula to calculate the distance between the α-carbon atoms: The distance formula is Step 4.4 Build a matrix: Organize all distance values ​​into a matrix, where the rows and columns of the matrix correspond to the indices of the α-carbon atoms.

10. The system for predicting dynamic structural changes of protein Aβ42 based on deep learning method according to any one of claims 6 to 9, characterized in that: The three-dimensional structure building module specifically includes the following steps: Step 6.1 Selection of initial points: At the beginning of the algorithm, a set of initial coordinates is randomly generated as the starting point of the optimization process; this random initialization method is often used to ensure that the optimization process can explore the solution space from a variety of possible configurations and reduce the risk of the solution being affected by the initial conditions; Step 6.2 Optimization process: Through an iterative method, the positions of these points are continuously adjusted to minimize the objective function; the objective function measures the deviation between the α-carbon distance matrix calculated based on the current coordinates and the constraint condition, that is, the predicted α-carbon distance matrix distance map; Interpretation of the results of step 6.3: After the optimization is completed, the obtained coordinates should be as close to the real three-dimensional structure as possible while satisfying the given distance constraints.

Images