Method and system for removing complex noise based on single particle cryo-electron microscopy images

Abstract

The present application belongs to the field of cryo-EM image processing, and particularly relates to a complex noise removal method and system based on single-particle cryo-EM images. The present application aims to solve the problem of low resolution and accuracy and insufficient generalization of the three-dimensional structure obtained by the denoising method of the prior art. The present application comprises: constructing a standard cryo-EM multi-particle image dataset; pre-training a background noise removal network and a structure noise removal network based on the standard cryo-EM multi-particle image dataset, and combining the two to obtain a complex noise removal network; performing three-dimensional reconstruction on a single particle set to obtain a three-dimensional reconstructed particle set, calculating the correlation coefficient of the three-dimensional reconstructed particle and the three-dimensional electron density map, constructing a complex noise removal objective function based on the correlation coefficient, and optimizing the complex noise removal network based on the complex noise removal objective function; and removing complex noise by using the optimized complex noise removal network. The present application improves the denoising performance and generalization ability of the model.

Description

Technical Field

[0001] This invention belongs to the field of cryo-electron microscopy image processing, and specifically relates to a method, system, electronic device and storage medium for complex noise removal based on single-particle cryo-electron microscopy images. Background Technology

[0002] Cryo-electron microscopy (cryo-EM) is a crucial research tool in structural biology. It constructs three-dimensional (3D) structures of proteins and macromolecular complexes from numerous two-dimensional (2D) micrographs, achieving near-atomic resolution of biomolecules and allowing researchers to glimpse how they perform their functions. However, cryo-EM images typically exhibit extremely low signal-to-noise ratios (SNRs), and the noise characteristics differ significantly from natural images. These noises include not only randomly distributed background noise but also structural noise of specific shapes within individual particles. Current denoising methods primarily focus on particle picking in 2D micrographs and traditional denoising and enhancement of single particles, failing to address the specific types of particle noise. This results in inherent noise remaining in the denoised images, ultimately reducing the resolution and accuracy of the 3D structures. Furthermore, current cryo-EM denoising methods are limited to single particles, raising questions about their generalizability.

[0003] Therefore, existing denoising methods still have some inherent noise, and the resolution and accuracy of the 3D structure are not high enough, and the generalization is insufficient. Summary of the Invention

[0004] To address the aforementioned problems in existing technologies, namely that current denoising methods still suffer from inherent noise, insufficient resolution and accuracy of 3D structures, and inadequate generalization, this invention provides a complex noise removal method based on single-particle cryo-electron microscopy images. The method includes:

[0005] A standard cryo-electron microscopy multi-particle image dataset was constructed based on the three-dimensional electron density maps of protein particles;

[0006] The background noise removal network and the structural noise removal network are pre-trained based on the standard cryo-electron microscopy multi-particle image dataset; the pre-trained background noise removal network and the pre-trained structural noise removal network are then concatenated and combined to obtain a complex noise removal network;

[0007] The three-dimensional electron density map of the protein particles in the standard cryo-electron microscopy multi-particle image dataset, which has undergone centralized projection and noise addition, is input into the complex noise removal network, which outputs a single particle set.

[0008] The individual particle set is reconstructed in three dimensions to obtain a three-dimensional reconstructed particle set, and the correlation coefficient between the three-dimensional reconstructed particles and the three-dimensional electron density maps of individual protein particles is calculated; based on the correlation coefficient, a complex noise removal objective function is constructed, and the complex noise removal network is optimized.

[0009] Complex noise in the three-dimensional electron density map of the protein particles is removed using a finely tuned complex noise removal network.

[0010] In a preferred embodiment, a method for constructing a standard cryo-electron microscopy multiparticle image dataset includes:

[0011] The three-dimensional electron density map of the protein particles is subjected to projection noise processing to obtain a particle projection map that simulates real complex noise.

[0012] The three-dimensional electron density map of the protein particles is denoised to obtain a three-dimensional electron density map with background noise removed. Then, the three-dimensional electron density map with background noise removed is projected at a fixed angle to obtain a particle projection map with background noise removed.

[0013] Structural noise removal is performed on the three-dimensional electron density map with background noise removed to obtain a three-dimensional electron density map with background noise and structural noise removed. A clean particle projection map with background noise and structural noise removed is obtained by projecting the three-dimensional electron density map with background noise and structural noise removed at a fixed angle.

[0014] The standard cryo-electron microscopy multi-particle image dataset is constructed based on the particle projection map corresponding to each particle that simulates real complex noise, the particle projection map corresponding to each particle with background noise removed, and the clean particle projection map corresponding to each particle with background noise and structural noise removed.

[0015] In a preferred embodiment, the training image pairs for the background noise removal network, the structural noise removal network, and the complex noise removal network during training are as follows:

[0016] The training image pairs for the background noise removal network are:

[0017] The training image pairs for the structural noise removal network are:

[0018] The training image pairs for the complex noise removal network:

[0019] Where, x i This represents the network input, where y0 represents the true label, and x represents the input. m x represents a particle projection diagram simulating realistic complex noise. bThe particle projection map after removing background noise, x bs This represents a clean particle projection image after removing background and structural noise.

[0020] In a preferred embodiment, the background noise removal network, the structural noise removal network, and the complex noise removal network are all trained using a diffusion model learning method under set noise conditions. When the network is a background noise removal network, the set noise conditions are noise conditions simulating real complex noise. When the network is a structural noise removal network, the set noise conditions are noise conditions under background noise removal. When the network is a complex noise removal network, the set noise conditions are noise conditions under both background noise and structural noise removal.

[0021] The background noise removal network, the structural noise removal network, and the complex noise removal network are learned using a diffusion model learning method under specific noise conditions. The specific process of back-inference of the diffusion model is as follows:

[0022]

[0023] Among them, y T The input image is a Gaussian noise image that follows a Gaussian distribution with a mean of 0 and a variance equal to the identity matrix, i.e.: y T ~N(0, I), y 0：T =[y0,y1…,y T ], y t-1 y represents the noise distribution output by the background noise removal network at step t-1. t This represents the noise distribution output by the diffusion model at step t, where T represents the total number of steps set during backpropagation, t represents the variable between 0 and T, y0 is the target image, P(*) represents the distribution function of the current data, and x i It is a particle projection map of specific noise, μ θ (x i ,y t ,γ t Let be the parameterized mean of the current distribution, with parameter θ, and x ∈ [value]. i ,y t ,γ t For the input network, γ t The variance of adding noise at step t, This represents the variance of the current distribution.

[0024] In a preferred embodiment, the objective function for removing complex noise is:

[0025]

[0026] Among them, L jLet E be the objective function for the j-th particle in the large-scale cryo-electron microscopy multi-particle image dataset. (x,y) γ represents the expected value with parameters x and y, where x and y are training image pairs extracted from the dataset, and γ represents the noise variance. m The particle projection diagram representing the simulated real complex noise of the j-th particle, x bs f represents the clean particle projection map of the j-th particle after removing background noise and structure noise. θ This represents a complex noise removal network, where 's' represents the 3D reconstruction particle Δ. θ The correlation coefficient with the three-dimensional electron density map Δ of protein particles, where λ represents the balance weight, ε is a constant, ε ~ N(0, I), and N is the number of particles in the standard cryo-electron microscopy multi-particle image dataset.

[0027] A third aspect of the present invention provides an electronic device comprising:

[0028] At least one processor; and

[0029] A memory communicatively connected to at least one of the processors; wherein,

[0030] The memory stores instructions that can be executed by the processor to implement the aforementioned complex noise removal method based on single-particle cryo-electron microscopy images.

[0031] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer instructions for execution by the computer to implement the above-described method for complex noise removal based on single-particle cryo-electron microscopy images.

[0032] The beneficial effects of this invention are:

[0033] (1) This application performs denoising tasks specifically according to the type of particle noise, thereby improving the resolution and accuracy of the three-dimensional structure;

[0034] (2) This invention proposes a method for specific denoising based on noise category, which improves the denoising performance of the model by correctly identifying noise in cryo-electron microscopy images and truly realizes signal recovery based on the real attributes of noise.

[0035] (3) This application can realize a more standard denoising method applicable to large-scale multi-particle denoising, and improve the generalization ability of the denoising method. Attached Figure Description

[0036] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0037] Figure 1 This is a method for removing complex noise based on single-particle cryo-electron microscopy images in an embodiment of the present invention;

[0038] Figure 2 This is a schematic diagram of the structure of a computer system used to implement the methods, systems, and devices of this application. Detailed Implementation

[0039] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0040] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0041] This invention provides a method for removing complex noise from single-particle cryo-electron microscopy images, the method comprising:

[0042] A standard cryo-electron microscopy multi-particle image dataset was constructed based on the three-dimensional electron density maps of protein particles;

[0043] The background noise removal network and the structural noise removal network are pre-trained based on the standard cryo-electron microscopy multi-particle image dataset; the pre-trained background noise removal network and the pre-trained structural noise removal network are then concatenated and combined to obtain a complex noise removal network;

[0044] The three-dimensional electron density map of the protein particles in the standard cryo-electron microscopy multi-particle image dataset, which has undergone centralized projection and noise addition, is input into the complex noise removal network, which outputs a single particle set.

[0045] The individual particle set is reconstructed in three dimensions to obtain a three-dimensional reconstructed particle set, and the correlation coefficient between the three-dimensional reconstructed particles and the three-dimensional electron density maps of individual protein particles is calculated; based on the correlation coefficient, a complex noise removal objective function is constructed, and the complex noise removal network is optimized.

[0046] Complex noise in the three-dimensional electron density map of the protein particles is removed using a finely tuned complex noise removal network.

[0047] To more clearly illustrate the complex noise removal method based on single-particle cryo-electron microscopy images of this invention, the following will be combined with... Figure 1 The steps in the embodiments of the present invention will be described in detail below.

[0048] The key idea of ​​this invention is to first construct a standard large-scale cryo-electron microscopy multi-particle image dataset based on publicly available three-dimensional electron density maps of protein particles. Then, a background noise removal network and a structural noise removal network are pre-trained separately to construct a complex noise removal network. Next, the denoised two-dimensional image and the three-dimensional reconstructed particles are used to construct a complex noise removal objective function, thereby optimizing the complex noise removal network and ultimately completing the complex noise removal process.

[0049] The method for removing complex noise based on single-particle cryo-electron microscopy images according to the first embodiment of the present invention is described in detail below:

[0050] A standard cryo-electron microscopy multi-particle image dataset was constructed based on the three-dimensional electron density maps of protein particles;

[0051] In this embodiment, the protein particle data can be obtained from publicly available cryo-electron microscopy databases, such as EMPIAR, EMD, and PDB. The 3D projection tool can be any tool currently used for 3D projection of protein particles, such as Aspire. The method for removing background noise and structural noise from the 3D electron density map can be any network model currently used for noise removal from the 3D electron density map of protein particles, such as DeepTracer.

[0052] In this embodiment, the standard cryo-electron microscopy multi-particle image dataset includes a particle projection map simulating real complex noise for each particle, a particle projection map with background noise removed for each particle, and a clean particle projection map with background noise and structural noise removed for each particle.

[0053] In this embodiment, the method for constructing a standard cryo-electron microscopy multi-particle image dataset includes: performing projection noise processing on the three-dimensional electron density map of the protein particles to obtain a particle projection map simulating real complex noise; performing background noise removal on the three-dimensional electron density map of the protein particles to obtain a three-dimensional electron density map without background noise, and then performing fixed-angle projection on the three-dimensional electron density map without background noise to obtain a particle projection map without background noise; performing structural noise removal on the three-dimensional electron density map without background noise and structural noise to obtain a three-dimensional electron density map without background noise and structural noise, and performing fixed-angle projection on the three-dimensional electron density map without background noise and structural noise to obtain a clean particle projection map without background noise and structural noise; and constructing the standard cryo-electron microscopy multi-particle image dataset based on the particle projection map simulating real complex noise corresponding to each particle, the particle projection map without background noise corresponding to each particle, and the clean particle projection map without background noise and structural noise corresponding to each particle.

[0054] The background noise removal network and the structural noise removal network were pre-trained using the cryo-electron microscopy multi-particle image dataset; the pre-trained background noise removal network and the pre-trained structural noise removal network were then concatenated and combined to obtain a complex noise removal network;

[0055] The training image pairs for the background noise removal network, the structure noise removal network, and the complex noise removal network during training are as follows:

[0056] The training image pairs for the background noise removal network are:

[0057] The training image pairs for the structural noise removal network are:

[0058] The training image pairs for the complex noise removal network:

[0059] Where, x i This represents the network input, where y0 represents the true label, and x represents the input. m x represents a particle projection diagram simulating realistic complex noise. b The particle projection map after removing background noise, x bs This represents a clean particle projection image after removing background and structural noise.

[0060] In this embodiment, the background noise removal network, the structural noise removal network, and the complex noise removal network are all trained using a diffusion model learning method under defined noise conditions. When the network is a background noise removal network, the noise conditions are set to simulate real complex noise conditions; when the network is a structural noise removal network, the noise conditions are set to remove background noise; and when the network is a complex noise removal network, the noise conditions are set to remove both background and structural noise. The baseline model for these networks can be any network that is effective in denoising.

[0061] In this embodiment, the background noise removal network, the structural noise removal network, and the complex noise removal network are learned using a diffusion model learning method under specific noise conditions. The specific process of the diffusion model back-inference is as follows:

[0062]

[0063] Among them, yT The input image is a Gaussian noise image that follows a Gaussian distribution with a mean of 0 and a variance equal to the identity matrix, i.e.: y T ~N(0, I), y 0：T =[y0,y1…,y T ], y t-1 y represents the noise distribution output by the background noise removal network at step t-1. t This represents the noise distribution output by the diffusion model at step t, where T represents the total number of steps set during backpropagation, t represents the variable between 0 and T, y0 is the target image, P(*) represents the distribution function of the current data, and x i It is a particle projection map of specific noise. Depending on the corresponding conditions, it can be a particle projection map simulating real complex noise, a particle projection map with background noise removed, or a clean particle projection map with both background and structural noise removed. μ θ (x i ,y t ,γ t Let be the parameterized mean of the current distribution, with parameter θ, and x ∈ [value]. i ,y t ,γ t For the input network, γ t The variance of adding noise at step t, This represents the variance of the current distribution.

[0064] The three-dimensional electron density map of the protein particles in the standard cryo-electron microscopy multi-particle image dataset, after centralized projection and noise addition, is input into a complex noise removal network, which outputs a single particle set.

[0065] The individual particle set is reconstructed in three dimensions to obtain a three-dimensional reconstructed particle set, and the correlation coefficient between the three-dimensional reconstructed particles and the three-dimensional electron density map of the individual protein particles is calculated. Based on the correlation coefficient, a complex noise removal objective function is constructed, and the complex noise removal network is optimized.

[0066] In this embodiment, the method for calculating the correlation coefficient between the three-dimensional reconstructed particle and the three-dimensional electron density map of a single protein particle can be any well-known method in the industry for calculating correlation coefficients, such as the Pearson correlation coefficient.

[0067] In this embodiment, the objective function for complex noise removal is:

[0068]

[0069] Among them, L j Let E be the objective function for the j-th particle in the large-scale cryo-electron microscopy multi-particle image dataset. (x,y) This represents the expected value with parameters x and y, where x and y are training image pairs extracted from the dataset.m The particle projection diagram representing the simulated real complex noise of the j-th particle, x bs f represents the clean particle projection map of the j-th particle after removing background noise and structure noise. θ This represents a complex noise removal network, where 's' represents the 3D reconstruction particle Δ. θ The correlation coefficient with the three-dimensional electron density map Δ of protein particles, where λ represents the balance weight, ε is a constant, ε ~ N(0, I), and N is the number of particles in the standard cryo-electron microscopy multi-particle image dataset.

[0070] Complex noise in the three-dimensional electron density map of the protein particles is removed using a finely tuned complex noise removal network.

[0071] Although the steps in the above embodiments are described in the above order, those skilled in the art will understand that in order to achieve the effect of this embodiment, different steps do not need to be executed in such an order. They can be executed simultaneously (in parallel) or in a reverse order. These simple variations are all within the protection scope of this invention.

[0072] A complex noise removal system based on single-particle cryo-electron microscopy images according to a second embodiment of the present invention, the system comprising:

[0073] Dataset building unit for constructing standard cryo-electron microscopy multi-particle image datasets based on three-dimensional electron density maps of protein particles;

[0074] The network pre-training unit is used to pre-train the background noise removal network and the structure noise removal network based on the standard cryo-electron microscopy multi-particle image dataset; the pre-trained background noise removal network and the pre-trained structure noise removal network are spliced ​​and combined to obtain a complex noise removal network;

[0075] The particle set output unit is used to input the three-dimensional electron density map of the standard cryo-electron microscopy multi-particle image dataset after centralized projection and noise addition processing into the complex noise removal network, and output a single particle set.

[0076] The network optimization unit is used to perform three-dimensional reconstruction on the single particle set to obtain a three-dimensional reconstructed particle set, calculate the correlation coefficient between the three-dimensional reconstructed particles and the three-dimensional electron density map of the single protein particle, construct a complex noise removal objective function based on the correlation coefficient, and optimize the complex noise removal network.

[0077] A complex noise removal unit is used to remove complex noise from the three-dimensional electron density map of the protein particles using a tuned complex noise removal network.

[0078] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the system described above can be found in the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0079] It should be noted that the complex noise removal system based on single-particle cryo-electron microscopy images provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the modules or steps in the embodiments of the present invention can be further decomposed or combined. For example, the modules in the above embodiments can be merged into one module, or further divided into multiple sub-modules to complete all or part of the functions described above. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing the various modules or steps and are not considered as an improper limitation of the present invention.

[0080] An electronic device according to a third embodiment of the present invention includes: at least one processor; and a memory communicatively connected to at least one of the processors; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to implement the above-described method for complex noise removal based on single-particle cryo-electron microscopy images.

[0081] A computer-readable storage medium according to a fourth embodiment of the present invention stores computer instructions, which are executed by the computer to implement the above-described method for complex noise removal based on single-particle cryo-electron microscopy images.

[0082] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the electronic devices and storage media described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0083] Those skilled in the art will recognize that the modules and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. The programs corresponding to the software modules and method steps can be placed in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the art. To clearly illustrate the interchangeability of electronic hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the invention.

[0084] The following is for reference. Figure 2 It shows a schematic diagram of the structure of a computer system for implementing the methods, systems, and devices of this application. Figure 2 The server shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0085] like Figure 2 As shown, the computer system includes a Central Processing Unit (CPU) 601, which can perform various appropriate actions and processes based on programs stored in Read Only Memory (ROM) 602 or programs loaded from storage section 608 into Random Access Memory (RAM) 603. The RAM 603 also stores various programs and data required for system operation. The CPU 601, ROM 602, and RAM 603 are interconnected via a bus 604. An Input / Output (I / O) interface 605 is also connected to the bus 604.

[0086] The following components are connected to I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 610 as needed so that computer programs read from it can be installed into storage section 608 as needed.

[0087] Specifically, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 609, and / or installed from removable medium 611. When the computer program is executed by central processing unit (CPU) 601, it performs the functions defined in the methods of this application. It should be noted that the computer-readable medium described above in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0088] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0089] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0090] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus / device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent in such process, method, article, or apparatus / device.

[0091] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for removing complex noise based on single-particle cryo-electron microscopy images, characterized in that, The method includes: A standard cryo-electron microscopy multi-particle image dataset was constructed based on the three-dimensional electron density maps of protein particles; The background noise removal network and the structural noise removal network are pre-trained based on the standard cryo-electron microscopy multi-particle image dataset; the pre-trained background noise removal network and the pre-trained structural noise removal network are then concatenated and combined to obtain a complex noise removal network; The three-dimensional electron density map of the protein particles in the standard cryo-electron microscopy multi-particle image dataset, which has undergone centralized projection and noise addition, is input into the complex noise removal network, which outputs a single particle set. The individual particle set is reconstructed in three dimensions to obtain a three-dimensional reconstructed particle set, and the correlation coefficient between the three-dimensional reconstructed particles and the three-dimensional electron density maps of individual protein particles is calculated; based on the correlation coefficient, a complex noise removal objective function is constructed, and the complex noise removal network is optimized. The complex noise in the three-dimensional electron density map of the protein particles is removed using a finely tuned complex noise removal network. Methods for constructing standard cryo-electron microscopy multi-particle image datasets include: The three-dimensional electron density map of the protein particles is subjected to projection noise processing to obtain a particle projection map that simulates real complex noise. The three-dimensional electron density map of the protein particles is denoised to obtain a three-dimensional electron density map with background noise removed. Then, the three-dimensional electron density map with background noise removed is projected at a fixed angle to obtain a particle projection map with background noise removed. Structural noise removal is performed on the three-dimensional electron density map with background noise removed to obtain a three-dimensional electron density map with background noise and structural noise removed. A clean particle projection map with background noise and structural noise removed is obtained by projecting the three-dimensional electron density map with background noise and structural noise removed at a fixed angle. The standard cryo-electron microscopy multi-particle image dataset is constructed based on the particle projection map corresponding to each particle that simulates real complex noise, the particle projection map corresponding to each particle with background noise removed, and the clean particle projection map corresponding to each particle with background noise and structural noise removed.

2. The method for removing complex noise based on single-particle cryo-electron microscopy images according to claim 1, characterized in that, The training image pairs for the background noise removal network, the structure noise removal network, and the complex noise removal network during training are as follows: The training image pairs for the background noise removal network are: The training image pairs for the structural noise removal network are: The training image pairs for the complex noise removal network: Where, x i This represents the network input, where y0 represents the true label, and x represents the input. m x represents a particle projection diagram simulating realistic complex noise. b The particle projection map after removing background noise, x bs This represents a clean particle projection image after removing background and structural noise.

3. The method for removing complex noise based on single-particle cryo-electron microscopy images according to claim 2, characterized in that, During training, the background noise removal network, the structural noise removal network, and the complex noise removal network all employ a diffusion model learning method under set noise conditions. When the network is a background noise removal network, the set noise conditions are those simulating real complex noise conditions; when the network is a structural noise removal network, the set noise conditions are those under background noise removal conditions. When the network is a complex noise removal network, the noise conditions set are the noise conditions under the conditions of removing background noise and structural noise.

4. The method for removing complex noise based on single-particle cryo-electron microscopy images according to claim 3, characterized in that, The background noise removal network, the structural noise removal network, and the complex noise removal network are learned using a diffusion model learning method under specified noise conditions. The specific process of back-inference for the diffusion model is as follows: Among them, y T The input image is a Gaussian noise image that follows a Gaussian distribution with a mean of 0 and a variance equal to the identity matrix, i.e.: y T ~N(0, I), y 0：T =[y0, y1, ..., y T ], y t-1 y represents the noise distribution output by the background noise removal network at step t-1. t This represents the noise distribution output by the diffusion model at step t, where T represents the total number of steps set during backpropagation, t represents the variable between 0 and T, y0 is the target image, P(*) represents the distribution function of the current data, and x i It is a particle projection map under set noise conditions, including the x m The x b The x bs μ θ (x i ,y t ,γ t Let be the parameterized mean of the current distribution, with parameter θ, and x ∈ [value]. i ,y t ,γ t For the input network, γ t The variance of adding noise at step t, This represents the variance of the current distribution.

5. The method for removing complex noise based on single-particle cryo-electron microscopy images according to claim 4, characterized in that, The objective function for complex noise removal is: Among them, L j Let E be the objective function for the j-th particle in the standard cryo-electron microscopy multi-particle image dataset. (x，y) E represents the expected value with parameters x and y, where x and y are training image pairs extracted from the dataset. (ε，γ) This represents the expected value with parameters ε and γ, where γ represents the noise variance, and x... m The particle projection diagram representing the simulated real complex noise of the j-th particle, x bs f represents the clean particle projection map of the j-th particle after removing background noise and structure noise. θ This represents a complex noise removal network, where 's' represents the 3D reconstruction particle Δ. θ The correlation coefficient with the three-dimensional electron density map Δ of protein particles, where λ represents the balance weight, ε is a constant, ε ~ N(0, I), and N is the number of particles in the standard cryo-electron microscopy multi-particle image dataset.

6. A complex noise removal system based on single-particle cryo-electron microscopy images, characterized in that, The system includes: Dataset building unit for constructing standard cryo-electron microscopy multi-particle image datasets based on three-dimensional electron density maps of protein particles; The network pre-training unit is used to pre-train the background noise removal network and the structure noise removal network based on the standard cryo-electron microscopy multi-particle image dataset; the pre-trained background noise removal network and the pre-trained structure noise removal network are spliced ​​and combined to obtain a complex noise removal network; The particle set output unit is used to input the three-dimensional electron density map of the standard cryo-electron microscopy multi-particle image dataset after centralized projection and noise addition processing into the complex noise removal network, and output a single particle set. The network optimization unit is used to perform three-dimensional reconstruction on the single particle set to obtain a three-dimensional reconstructed particle set, calculate the correlation coefficient between the three-dimensional reconstructed particles and the three-dimensional electron density map of the single protein particle, construct a complex noise removal objective function based on the correlation coefficient, and optimize the complex noise removal network. A complex noise removal unit is used to remove complex noise from the three-dimensional electron density map of the protein particles using a finely tuned complex noise removal network. Methods for constructing standard cryo-electron microscopy multi-particle image datasets include: The three-dimensional electron density map of the protein particles is subjected to projection noise processing to obtain a particle projection map that simulates real complex noise. The three-dimensional electron density map of the protein particles is denoised to obtain a three-dimensional electron density map with background noise removed. Then, the three-dimensional electron density map with background noise removed is projected at a fixed angle to obtain a particle projection map with background noise removed. Structural noise removal is performed on the three-dimensional electron density map with background noise removed to obtain a three-dimensional electron density map with background noise and structural noise removed. A clean particle projection map with background noise and structural noise removed is obtained by projecting the three-dimensional electron density map with background noise and structural noise removed at a fixed angle. The standard cryo-electron microscopy multi-particle image dataset is constructed based on the particle projection map corresponding to each particle that simulates real complex noise, the particle projection map corresponding to each particle with background noise removed, and the clean particle projection map corresponding to each particle with background noise and structural noise removed.

7. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by the processor to implement the complex noise removal method based on single-particle cryo-electron microscopy images as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are executed by the computer to implement the complex noise removal method based on single-particle cryo-electron microscopy images as described in any one of claims 1-5.

Images