Three-dimensional structure reasoning method, device and equipment of amino acid sequence and medium

Abstract

The invention discloses a three-dimensional structure reasoning method, device and equipment of an amino acid sequence and a medium, relates to the technical field of deep learning, is applied to reasoning equipment, and comprises the following steps: splitting a reasoning model into a first sub-model and a second sub-model; performing operation on the to-be-reasoned amino acid sequence by using the first sub-model through the first processor; splitting the second sub-model, the weight data of the second sub-model and the operation result by using a splitting unit determined based on the dimension information through a second processor, compiling the split sub-model into an executable file, and performing operation on the split operation result by using the split weight data and the executable file to obtain a plurality of operation results; and splicing the plurality of operation results through a second processor, and determining a three-dimensional structure of the amino acid sequence to be reasoned by utilizing a splicing result. According to the method, the problems that the memory of a high-performance computing unit is limited and inference operation with relatively high dynamic performance cannot be well processed are solved.

Description

Technical Field

[0001] The present invention relates to the field of deep learning technology, and in particular to a three-dimensional structure reasoning method, device, equipment and medium for amino acid sequences. Background Art

[0002] The Alphafold model uses deep learning methods to infer the three-dimensional structure of proteins based solely on their amino acid sequences, which has potential value for both biological and medical research.

[0003] However, the alphafold model requires a lot of computation and is slow to reason on the central processing unit (CPU), making it difficult to be widely used in practice. To this end, researchers have tried to perform reasoning on high-performance computing units such as GPGPU (General Purpose Computing on GPU) and artificial intelligence accelerators. However, high-performance computing units often have memory limitations, and the three-dimensional structure of proteins is highly dynamic during reasoning, and high-performance computing units cannot handle highly dynamic reasoning operations well.

[0004] Therefore, the above technical problems need to be solved by those skilled in the art urgently. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a three-dimensional structure reasoning method, device, equipment and medium for amino acid sequences, which can solve the problem that the high-performance computing unit has limited memory and cannot handle the reasoning operation with strong dynamics well. The specific scheme is as follows:

[0006] In a first aspect, the present application discloses a three-dimensional structure reasoning method for an amino acid sequence, which is applied to a reasoning device, comprising:

[0007] Determine target dimension information according to the length information of the amino acid sequence to be inferred;

[0008] Splitting the target inference model into a first sub-model adapted to the operation characteristics of the local first processor and a second sub-model adapted to the operation characteristics of the local second processor; the second processor is a processor of a different type from the first processor and meets a preset high performance condition, and a memory unit in the second processor accesses the device memory of the inference device through a preset connection channel;

[0009] Allocating the first sub-model and the second sub-model to the first processor and the second processor respectively;

[0010] Using the first processor and the first sub-model to operate the amino acid sequence to be inferred, to obtain a first operation result;

[0011] The second processor uses the splitting unit determined based on the target dimension information to split the second sub-model, the weight data of the second sub-model, and the first operation result, and then compiles the split sub-model into an executable file, and uses the split weight data and the executable file to calculate the split operation result to obtain multiple second operation results;

[0012] The second processor splices a plurality of the second operation results, and uses the processing result obtained after processing the splicing result to determine the three-dimensional structure of the amino acid sequence to be inferred.

[0013] Optionally, before the second processor splits the second sub-model, the weight data of the second sub-model, and the first operation result by using the splitting unit determined based on the target dimension information, the method further includes:

[0014] The weight data of the second sub-model stored in the device memory is copied to the memory unit through the preset connection channel, so as to obtain the weight data of the second sub-model from the memory unit.

[0015] Optionally, the step of performing a calculation on the amino acid sequence to be inferred by the first processor using the first sub-model to obtain a first calculation result includes:

[0016] Using the first sub-model, the first processor compares the amino acid sequence to be inferred with the amino acid sequences in a preset gene database to obtain a first multiple sequence alignment matrix;

[0017] Using the first sub-model, the first processor compares the amino acid sequence to be inferred with the structural data in a preset structural database to obtain a first structural comparison matrix;

[0018] Correspondingly, splitting the first operation result by the second processor includes:

[0019] Splitting the first multiple sequence alignment matrix and the first structural alignment matrix by the second processor to obtain the post-splitting operation result including the split first multiple sequence alignment matrix and the split first structural alignment matrix;

[0020] Furthermore, the executable file includes a plurality of neural network components that share the weight data after the splitting; accordingly, the weight data after the splitting is used together with the executable file to perform operations on the operation results after the splitting to obtain a plurality of second operation results, including:

[0021] The first multiple sequence alignment matrix after splitting and the first structure alignment matrix after splitting are operated according to the post-splitting weight data and the neural network component to respectively obtain a plurality of the second operation results corresponding to the first multiple sequence alignment matrix after splitting and a plurality of the second operation results corresponding to the first structure alignment matrix after splitting.

[0022] Optionally, after the operation is performed on the first multiple sequence alignment matrix after splitting and the first structure alignment matrix after splitting according to the weight data after splitting and the neural network component, the method further includes:

[0023] The weight data of the second sub-model in the memory unit is released.

[0024] Optionally, the splicing of the plurality of second operation results by the second processor includes:

[0025] splicing, by the second processor, a plurality of the second operation results corresponding to the split first multiple sequence alignment matrix to obtain a second multiple sequence alignment matrix;

[0026] The second processor concatenates a plurality of the second operation results corresponding to the split first structure comparison matrix to obtain a second structure comparison matrix.

[0027] Optionally, the process of determining the three-dimensional structure of the amino acid sequence to be inferred using the processing result obtained after processing the splicing result includes:

[0028] The second multiple sequence alignment matrix and the second structure alignment matrix are processed based on a preset attention mechanism to obtain a target weight matrix; wherein the target weight matrix contains weight values ​​of multiple target amino acid sequences that match the amino acid sequence to be inferred.

[0029] Optionally, determining the three-dimensional structure of the amino acid sequence to be inferred using the processing result obtained after processing the splicing result includes:

[0030] An outer product operation is performed on the target weight matrix and the first multiple sequence alignment matrix, and the characteristics of the amino acid sequence to be inferred are determined based on the average value of the outer product operation results, so as to obtain the three-dimensional structure of the amino acid sequence to be inferred based on the characteristics of the amino acid sequence to be inferred.

[0031] In a second aspect, the present application discloses a three-dimensional structure reasoning device for an amino acid sequence, which is applied to a reasoning device, comprising:

[0032] A dimension determination module, used to determine target dimension information according to the length information of the amino acid sequence to be inferred;

[0033] A first splitting module is used to split the target inference model into a first sub-model adapted to the operation characteristics of the local first processor and a second sub-model adapted to the operation characteristics of the local second processor; the second processor is a processor of a different type from the first processor and meets a preset high performance condition, and a memory unit in the second processor accesses the device memory of the inference device through a preset connection channel;

[0034] A model allocation module, configured to allocate the first sub-model and the second sub-model to the first processor and the second processor respectively;

[0035] A first operation module, configured to operate the amino acid sequence to be inferred by using the first sub-model through the first processor to obtain a first operation result;

[0036] A second splitting module, configured to split the second sub-model, the weight data of the second sub-model, and the first operation result by using the second processor and a splitting unit determined based on the target dimension information;

[0037] A second operation module is used to compile the split sub-model into an executable file, and use the split weight data and the executable file to operate the split operation results to obtain multiple second operation results;

[0038] The three-dimensional structure determination module is used to splice multiple second operation results through the second processor, and determine the three-dimensional structure of the amino acid sequence to be inferred using the processing result obtained after processing the splicing result.

[0039] In a third aspect, the present application discloses an electronic device, including:

[0040] Memory, used to store computer programs;

[0041] A processor is used to execute the computer program to implement the three-dimensional structure reasoning method of the amino acid sequence disclosed above.

[0042] In a fourth aspect, the present application discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the three-dimensional structure reasoning method of the amino acid sequence disclosed above.

[0043] It can be seen that the present application proposes a three-dimensional structure reasoning method for an amino acid sequence, which is applied to an inference device, including: determining target dimension information according to the length information of the amino acid sequence to be inferred; splitting the target reasoning model into a first sub-model adapted to the operation characteristics of the local first processor and a second sub-model adapted to the operation characteristics of the local second processor; the second processor is a processor of a different type from the first processor and meets the preset high performance conditions, and the memory unit in the second processor accesses the device memory of the inference device through a preset connection channel; the first sub-model and the second sub-model are respectively assigned to the first processor and the second processor; the first processor uses the first sub-model to operate on the amino acid sequence to be inferred to obtain a first operation result; the second processor uses the splitting unit determined based on the target dimension information to split the second sub-model, the weight data of the second sub-model and the first operation result, and then compiles the split sub-model into an executable file, and uses the split weight data and the executable file to operate on the split operation result to obtain multiple second operation results; the second processor splices the multiple second operation results, and determines the three-dimensional structure of the amino acid sequence to be inferred using the processing result obtained after processing the splicing result.

[0044] Beneficial effects: For the reasoning architecture in the present application, the memory and the unit memory of the computing unit are interconnected, so that the computing unit can realize the reasoning operation with the help of the memory interconnected with the unit memory, solving the problem of the unit memory limitation of the high-performance computing unit during reasoning. Further, the present application splits the target reasoning model into a first sub-model and a second sub-model, and assigns the first sub-model to the central processing unit and the second sub-model to the computing unit, so that the central processing unit and the high-performance computing unit respectively complete the reasoning operation adapted to each other, reducing the workload of the high-performance computing unit, and further improving the problem of the unit memory limitation of the high-performance computing unit during reasoning. In addition, the present application completes the operation of the second sub-model involving a large amount of calculation by the high-performance computing unit, while solving the problem of memory limitation. In addition, the present application proposes to compile the split sub-model into an executable file, so that the dynamic reasoning operation is converted into a static calculation graph, solving the problem that the high-performance computing unit cannot handle the reasoning operation with strong dynamics well. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0046] Figure 1 A flowchart of a three-dimensional structure reasoning method for an amino acid sequence disclosed in the present application;

[0047] Figure 2 A schematic diagram of an inference device disclosed in the present application;

[0048] Figure 3 A schematic diagram of the structure of a three-dimensional structure inference device for an amino acid sequence disclosed in the present application;

[0049] Figure 4 This is a structural diagram of an electronic device disclosed in this application. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0051] High-performance computing units often have limited memory, and the three-dimensional structure of proteins is highly dynamic during reasoning, and high-performance computing units cannot handle highly dynamic reasoning operations well. To this end, the present application embodiment proposes a three-dimensional structure reasoning solution for amino acid sequences, which can solve the problem that high-performance computing units have limited memory and cannot handle highly dynamic reasoning operations well.

[0052] To this end, the embodiment of the present application proposes a three-dimensional structure reasoning solution for amino acid sequences, which solves the problem that the high-performance computing unit has limited memory and cannot handle highly dynamic reasoning operations well.

[0053] The present application discloses a method for inferring the three-dimensional structure of an amino acid sequence, see Figure 1 As shown, the method includes:

[0054] Step S11: Determine the target dimension information according to the length information of the amino acid sequence to be inferred.

[0055] In this embodiment, the input dimensional information includes target dimensional information, and the target dimensional information varies with the length information of the amino acid sequence. Therefore, in this embodiment, the length information of the amino acid sequence to be inferred should be first determined, and the target dimensional information should be determined according to the length information of the amino acid sequence to be inferred, so as to determine the splitting unit according to the target dimensional information. For example, the size of the splitting unit is 4. It can be understood that the size of the splitting unit in this application depends on the actual situation and is not limited here.

[0056] Step S12: split the target reasoning model into a first sub-model adapted to the computing characteristics of the local first processor and a second sub-model adapted to the computing characteristics of the local second processor; the second processor is a processor of a type different from that of the first processor and meets preset high-performance conditions, and the memory unit in the second processor accesses the device memory of the reasoning device through a preset connection channel.

[0057] Figure 2 A schematic diagram of an inference device is provided, in which a first processor is connected to a device memory, and a memory unit in a second processor accesses the device memory of the inference device through a preset connection channel. Since the second processor can directly interact with the memory unit and the first processor can directly interact with the device memory, the present application connects the device memory with the unit memory, solving the problem that the second processor cannot directly interact with the device memory and the first processor cannot interact with the second processor. For example, the first processor includes a central processing unit, and the second processor includes a high-performance computing unit.

[0058] In this embodiment, the target reasoning model is split into a first sub-model adapted to the computing characteristics of the local first processor and a second sub-model adapted to the computing characteristics of the local second processor. It should be pointed out that the computing characteristics mean that different processors are suitable for different computing processes. For example, the central processing unit is suitable for processing conditions, branch statements, etc., and the high-performance computing unit is suitable for processing matrix multiplication, convolution, etc.

[0059] Step S13: assigning the first sub-model and the second sub-model to the first processor and the second processor respectively.

[0060] In this embodiment, the first sub-model and the second sub-model are respectively assigned to the first processor and the second processor, so that the first processor can perform corresponding operations based on the first sub-model, and the second processor can perform corresponding operations based on the second sub-model.

[0061] Step S14: using the first processor to operate the amino acid sequence to be inferred using the first sub-model to obtain a first operation result.

[0062] In this embodiment, the first processor uses the first sub-model to compare the amino acid sequence to be inferred with the amino acid sequences in a preset gene database to obtain a first multiple sequence alignment matrix, and the first processor uses the first sub-model to compare the amino acid sequence to be inferred with the structural data in a preset structure database to obtain a first structure alignment matrix.

[0063] This embodiment completes the search of the gene database and the search of the structure database through the first sub-model, wherein the amino acid sequences in the gene database are, for example, the amino acid sequences of fish, rabbits, and humans, and the structure database contains the three-dimensional structure information of the corresponding amino acid sequences. Since this part mainly involves database query operations, it is placed on the central processing unit for calculation. In this way, the present application processes part of the calculations through the central processing unit, which reduces the workload of the high-performance computing unit and makes the use of computing resources more balanced.

[0064] Step S15: The second processor uses the splitting unit determined based on the target dimension information to split the second sub-model, the weight data of the second sub-model and the first operation result, and then compiles the split sub-model into an executable file, and uses the split weight data and the executable file to calculate the split operation result to obtain multiple second operation results.

[0065] Considering the limitation of the high-performance computing unit, this embodiment needs to split the second sub-model, the weight data of the second sub-model and the first operation result by the second processor using the splitting unit determined based on the target dimension information. Specifically, the first operation result is split by the second processor, including: splitting the first multiple sequence alignment matrix and the first structure alignment matrix by the second processor to obtain the split operation result including the split first multiple sequence alignment matrix and the split first structure alignment matrix. Considering the problem that the high-performance computing unit cannot handle the reasoning operation with strong dynamicity well, this embodiment compiles the split sub-model into an executable file to realize the conversion of the dynamic reasoning operation into a static calculation graph, and uses the split weight data and the executable file to calculate the split operation result to obtain multiple second operation results. Specifically, the split first multiple sequence alignment matrix and the split first structure alignment matrix are calculated according to the split weight data and the neural network component to obtain multiple second operation results corresponding to the split first multiple sequence alignment matrix and multiple second operation results corresponding to the split first structure alignment matrix. It should be pointed out that in the process of using neural network components for calculations, the results of the calculation process need to be processed through Dropout (random deactivation) to prevent overfitting.

[0066] In this embodiment, due to the limitation of the high-performance computing unit, the weight data of the second sub-model is stored in the device memory, and is copied from the device memory to the memory unit when needed. Specifically, the weight data of the second sub-model stored in the device memory is copied to the memory unit through the preset connection channel, so as to obtain the weight data of the second sub-model from the memory unit.

[0067] That is, when a certain weight data is needed, the weight data is copied from the device memory to the memory unit, and most of the rest of the time, the weight data is stored in the device memory, which greatly saves the unit memory of the high-performance computing unit and avoids the problem of limited unit memory of the high-performance computing unit. Furthermore, after the calculation is completed, the weight data of the second sub-model in the memory unit is released to further save the unit memory of the computing unit and avoid the problem of limited unit memory of the computing unit.

[0068] Step S16: The second processor splices a plurality of the second operation results, and uses the processing result obtained after processing the splicing results to determine the three-dimensional structure of the amino acid sequence to be inferred.

[0069] In this embodiment, the second processor splices multiple second operation results corresponding to the first multiple sequence alignment matrix after the splitting to obtain a second multiple sequence alignment matrix, and the second processor splices multiple second operation results corresponding to the first structure alignment matrix after the splitting to obtain a second structure alignment matrix. Further, the second multiple sequence alignment matrix and the second structure alignment matrix are processed based on a preset attention mechanism to obtain a target weight matrix; wherein the target weight matrix contains weight values ​​of multiple target amino acid sequences that match the amino acid sequence to be inferred. And the target weight matrix and the first multiple sequence alignment matrix are subjected to an outer product operation, and the characteristics of the amino acid sequence to be inferred are determined according to the average value of the outer product operation results, so as to obtain the three-dimensional structure of the amino acid sequence to be inferred according to the characteristics of the amino acid sequence to be inferred.

[0070] In this embodiment, in the process of processing the second multiple sequence alignment matrix and the second structure alignment matrix based on the preset attention mechanism to obtain the target weight matrix, the column vectors of the second multiple sequence alignment matrix and the second structure alignment matrix are processed based on the attention mechanism to obtain a specific amino acid sequence, and the row vectors of the second multiple sequence alignment matrix and the second structure alignment matrix are processed based on the attention mechanism to obtain the species information of the amino acid sequence, that is, the target weight may include 3 amino acid sequences, which are the amino acid sequences of humans, fish and birds, respectively, and the probabilities of the three amino acid sequences are 0.8, 0.1 and 0.1, respectively. It should be pointed out that since this submodule involves a lot of vector calculations, this module is placed on a high-performance computing unit for calculation to obtain better performance.

[0071] In summary, the present application determines the dimensional information of the tensor in the inference model according to the length of the amino acid sequence to be inferred, and divides the inference so that the sub-model obtained by the division can be executed on a high-performance computing unit or on a central processing unit. Further, for the sub-model that can be executed on the board, it is compiled into a file executable by the high-performance computing unit according to the static calculation graph. In the present application, the dimensional size of the target dimension of each sub-model is consistent with the split unit, so the sub-model after splitting is a sub-model with repeated structure. In this way, each executable file can be obtained through a compilation operation, which greatly reduces the time of model compilation and saves computing resources. Furthermore, in the present application, not all sub-models are executed by high-performance computing units. For some sub-models suitable for processing on the central processing unit, they can be calculated on the central processing unit, so that the overall performance is improved and the problem of limited memory of the high-performance computing unit is solved.

[0072] For example, a company's board T100 is a high-performance computing unit, and the early version of the board can only support the calculation of static graphs. Using the method described in this application, the task of T100 and the central processing unit to perform alphafold model reasoning is completed, which has higher performance than the pure central processing unit reasoning solution.

[0073] In addition, based on the method described in the present application, the present application can be further improved. For example, for the GFPGAN model that realizes face reconstruction, the GFPGAN model is split into multiple sub-models, and the split sub-models are assigned to the central processing unit and the high-performance computing unit. The central processing unit then completes the corresponding reconstruction operation. Furthermore, the operation results of the central processing unit are sent to the sub-model on the high-performance computing unit to complete the final operation through the sub-model on the high-performance computing unit. In this way, the promotion and application of the GFPGAN model in the face fuzzy reconstruction scenario is realized.

[0074] Correspondingly, the present application also discloses a three-dimensional structure reasoning device for an amino acid sequence, see Figure 3 As shown, the device comprises:

[0075] A dimension determination module 11 is used to determine target dimension information according to the length information of the amino acid sequence to be inferred;

[0076] A first splitting module 12 is used to split the target inference model into a first sub-model adapted to the operation characteristics of the local first processor and a second sub-model adapted to the operation characteristics of the local second processor; the second processor is a processor of a different type from the first processor and meets a preset high performance condition, and a memory unit in the second processor accesses the device memory of the inference device through a preset connection channel;

[0077] A model allocation module 13, configured to allocate the first sub-model and the second sub-model to the first processor and the second processor respectively;

[0078] A first operation module 14 is used to operate the amino acid sequence to be inferred by using the first sub-model through the first processor to obtain a first operation result;

[0079] A second splitting module 15 is used to split the second sub-model, the weight data of the second sub-model and the first operation result by using the splitting unit determined based on the target dimension information through the second processor;

[0080] A second operation module 16 is used to compile the split sub-model into an executable file, and use the split weight data and the executable file to operate the split operation results to obtain multiple second operation results;

[0081] The three-dimensional structure determination module 17 is used to splice multiple second operation results through the second processor, and determine the three-dimensional structure of the amino acid sequence to be inferred using the processing result obtained after processing the splicing results.

[0082] Among them, for more specific working processes of the above-mentioned modules, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0083] It can be seen that the present application proposes a three-dimensional structure reasoning method for an amino acid sequence, which is applied to an inference device, including: determining target dimension information according to the length information of the amino acid sequence to be inferred; splitting the target reasoning model into a first sub-model adapted to the operation characteristics of the local first processor and a second sub-model adapted to the operation characteristics of the local second processor; the second processor is a processor of a different type from the first processor and meets the preset high performance conditions, and the memory unit in the second processor accesses the device memory of the inference device through a preset connection channel; the first sub-model and the second sub-model are respectively assigned to the first processor and the second processor; the first processor uses the first sub-model to operate on the amino acid sequence to be inferred to obtain a first operation result; the second processor uses the splitting unit determined based on the target dimension information to split the second sub-model, the weight data of the second sub-model and the first operation result, and then compiles the split sub-model into an executable file, and uses the split weight data and the executable file to operate on the split operation result to obtain multiple second operation results; the second processor splices the multiple second operation results, and determines the three-dimensional structure of the amino acid sequence to be inferred using the processing result obtained after processing the splicing result.

[0084] Beneficial effects: For the reasoning architecture in the present application, the memory and the unit memory of the computing unit are interconnected, so that the computing unit can realize the reasoning operation with the help of the memory interconnected with the unit memory, solving the problem of the unit memory limitation of the high-performance computing unit during reasoning. Further, the present application splits the target reasoning model into a first sub-model and a second sub-model, and assigns the first sub-model to the central processing unit and the second sub-model to the computing unit, so that the central processing unit and the high-performance computing unit respectively complete the reasoning operation adapted to each other, reducing the workload of the high-performance computing unit, and further improving the problem of the unit memory limitation of the high-performance computing unit during reasoning. In addition, the present application completes the operation of the second sub-model involving a large amount of calculation by the high-performance computing unit, while solving the problem of memory limitation. In addition, the present application proposes to compile the split sub-model into an executable file, so that the dynamic reasoning operation is converted into a static calculation graph, solving the problem that the high-performance computing unit cannot handle the reasoning operation with strong dynamics well.

[0085] Furthermore, an embodiment of the present application also provides an electronic device. Figure 4 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content in the diagram cannot be considered as any limitation on the scope of use of the present application.

[0086] Figure 4A schematic diagram of the structure of an electronic device 20 provided in an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a display screen 23, an input and output interface 24, a communication interface 25, a power supply 26 and a communication bus 27. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the three-dimensional structure reasoning method of the amino acid sequence disclosed in any of the aforementioned embodiments. In addition, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0087] In this embodiment, the power supply 26 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 25 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input and output interface 24 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0088] In addition, the memory 22, as a carrier for storing resources, may be a read-only memory, a random access memory, a disk or an optical disk, etc., and the resources stored thereon may include a computer program 221, and the storage method may be temporary storage or permanent storage. Among them, the computer program 221 may further include a computer program that can be used to complete the three-dimensional structure reasoning method of the amino acid sequence performed by the electronic device 20 disclosed in any of the aforementioned embodiments, and can be used to complete other specific tasks.

[0089] Furthermore, an embodiment of the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the three-dimensional structure reasoning method of the amino acid sequence disclosed above is implemented.

[0090] For the specific steps of the method, reference may be made to the corresponding contents disclosed in the aforementioned embodiments, which will not be described in detail here.

[0091] Furthermore, an embodiment of the present application also discloses a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the three-dimensional structure reasoning method of amino acid sequences disclosed above.

[0092] The various embodiments in this application are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0093] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0094] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0095] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0096] The above is a detailed introduction to the three-dimensional structure reasoning method, device, equipment, and storage medium for an amino acid sequence provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for a person skilled in the art, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A three-dimensional structure reasoning method for an amino acid sequence, characterized in that: Applied to inference devices, including: Determine target dimension information according to the length information of the amino acid sequence to be inferred; Splitting the target inference model into a first sub-model adapted to the operation characteristics of the local first processor and a second sub-model adapted to the operation characteristics of the local second processor; the second processor is a processor of a different type from the first processor and meets a preset high performance condition, and a memory unit in the second processor accesses the device memory of the inference device through a preset connection channel; Allocating the first sub-model and the second sub-model to the first processor and the second processor respectively; Using the first processor and the first sub-model to operate the amino acid sequence to be inferred, to obtain a first operation result; The second processor uses the splitting unit determined based on the target dimension information to split the second sub-model, the weight data of the second sub-model, and the first operation result, and then compiles the split sub-model into an executable file, and uses the split weight data and the executable file to calculate the split operation result to obtain multiple second operation results; The second processor splices a plurality of the second operation results, and uses the processing result obtained after processing the splicing result to determine the three-dimensional structure of the amino acid sequence to be inferred.

2. The three-dimensional structure inference method of an amino acid sequence according to claim 1, characterized in that: Before the second processor splits the second sub-model, the weight data of the second sub-model, and the first operation result using the splitting unit determined based on the target dimension information, the method further includes: The weight data of the second sub-model stored in the device memory is copied to the memory unit through the preset connection channel, so as to obtain the weight data of the second sub-model from the memory unit.

3. The three-dimensional structure inference method of amino acid sequence according to claim 2, characterized in that: The step of performing a calculation on the amino acid sequence to be inferred by the first processor using the first sub-model to obtain a first calculation result includes: Using the first sub-model, the first processor compares the amino acid sequence to be inferred with the amino acid sequences in a preset gene database to obtain a first multiple sequence alignment matrix; Using the first sub-model, the first processor compares the amino acid sequence to be inferred with the structural data in a preset structural database to obtain a first structural comparison matrix; Correspondingly, splitting the first operation result by the second processor includes: Splitting the first multiple sequence alignment matrix and the first structural alignment matrix by the second processor to obtain the post-splitting operation result including the split first multiple sequence alignment matrix and the split first structural alignment matrix; Furthermore, the executable file includes a plurality of neural network components that share the weight data after the splitting; accordingly, the weight data after the splitting is used together with the executable file to perform operations on the operation results after the splitting to obtain a plurality of second operation results, including: The first multiple sequence alignment matrix after splitting and the first structure alignment matrix after splitting are operated according to the post-splitting weight data and the neural network component to respectively obtain a plurality of the second operation results corresponding to the first multiple sequence alignment matrix after splitting and a plurality of the second operation results corresponding to the first structure alignment matrix after splitting.

4. The three-dimensional structure inference method of amino acid sequence according to claim 3, characterized in that: After the operation is performed on the first multiple sequence alignment matrix after splitting and the first structure alignment matrix after splitting according to the weight data after splitting and the neural network component, the method further includes: The weight data of the second sub-model in the memory unit is released.

5. The three-dimensional structure inference method of amino acid sequence according to claim 3, characterized in that: The step of combining the plurality of the second operation results by the second processor includes: splicing, by the second processor, a plurality of the second operation results corresponding to the split first multiple sequence alignment matrix to obtain a second multiple sequence alignment matrix; The second processor concatenates a plurality of the second operation results corresponding to the split first structure comparison matrix to obtain a second structure comparison matrix.

6. The three-dimensional structure inference method of amino acid sequence according to claim 5, characterized in that: The process of determining the three-dimensional structure of the amino acid sequence to be inferred by using the processing result obtained after processing the splicing result includes: The second multiple sequence alignment matrix and the second structure alignment matrix are processed based on a preset attention mechanism to obtain a target weight matrix; wherein the target weight matrix contains weight values ​​of multiple target amino acid sequences that match the amino acid sequence to be inferred.

7. The three-dimensional structure inference method of amino acid sequence according to claim 6, characterized in that: The method of determining the three-dimensional structure of the amino acid sequence to be inferred by using the processing result obtained after processing the splicing result includes: An outer product operation is performed on the target weight matrix and the first multiple sequence alignment matrix, and the characteristics of the amino acid sequence to be inferred are determined based on the average value of the outer product operation results, so as to obtain the three-dimensional structure of the amino acid sequence to be inferred based on the characteristics of the amino acid sequence to be inferred.

8. A three-dimensional structure inference device for an amino acid sequence, characterized in that: Applied to inference devices, including: A dimension determination module, used to determine target dimension information according to the length information of the amino acid sequence to be inferred; A first splitting module is used to split the target inference model into a first sub-model adapted to the operation characteristics of the local first processor and a second sub-model adapted to the operation characteristics of the local second processor; the second processor is a processor of a different type from the first processor and meets a preset high performance condition, and a memory unit in the second processor accesses the device memory of the inference device through a preset connection channel; A model allocation module, configured to allocate the first sub-model and the second sub-model to the first processor and the second processor respectively; A first operation module, configured to operate the amino acid sequence to be inferred by using the first sub-model through the first processor to obtain a first operation result; A second splitting module, configured to split the second sub-model, the weight data of the second sub-model, and the first operation result by using the second processor and a splitting unit determined based on the target dimension information; A second operation module is used to compile the split sub-model into an executable file, and use the split weight data and the executable file to operate the split operation results to obtain multiple second operation results; The three-dimensional structure determination module is used to splice multiple second operation results through the second processor, and determine the three-dimensional structure of the amino acid sequence to be inferred using the processing result obtained after processing the splicing result.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the three-dimensional structure reasoning method of an amino acid sequence as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the three-dimensional structure reasoning method of an amino acid sequence as described in any one of claims 1 to 7.

Images