Signal peptide generation method based on quantum-assisted conditional generative adversarial neural network

Abstract

The present invention discloses a signal peptide generation method based on quantum-assisted conditional generative adversarial neural network, which relates to the technical field of data generation based on quantum computing. The present invention adopts a brand-new computing mode based on the basic principles of quantum mechanics, namely quantum computing, which can ensure the quality of the data characteristics of the generated signal peptides, can adopt a real quantum computer, and utilize the huge Hilbert space and complete randomness of the real quantum computer, and can generate data characteristics of signal peptides of different categories such as bacteria and viruses.

Description

Background Art

[0002] Signal peptides are a class of short-chain polypeptide molecules that play a role in signal transmission in organisms. Signal peptides are usually composed of 20-30 amino acids and diffuse or transmit signals between cells or within cells. Signal peptides play an important role in physiological regulation, cell communication, and the regulation of biological processes. The structure and sequence of signal peptides are very diverse, and different signal peptides can trigger different cellular responses and physiological processes. Signal peptides are key factors in guiding newly synthesized proteins to move from the cytoplasm to the endoplasmic reticulum for transcription-translation complexes. This process is necessary for the correct folding and modification of proteins. Secretory proteins, cell membrane proteins, etc. related to signal peptides are closely related to the research and development of most drug targets.

[0003] Signal peptides play an important role in maintaining physiological balance and regulating cell functions. Research on signal peptides can provide a theoretical basis and experimental foundation for the development of new therapeutic drugs, and has broad application prospects for the treatment of cardiovascular, digestive, neurological and other diseases. Research on signal peptides can provide important theoretical support and experimental basis for disease treatment and drug development. For example, it may be possible to control the synthesis, modification and folding of proteins by regulating signal peptides, thereby achieving the purpose of treating certain diseases. At the same time, signal peptides may also become new targets in drug research and development, helping to develop more targeted drugs.

[0004] At present, people mainly use classical machine learning methods for signal peptide generation tasks, while the signal peptide generation task using quantum-assisted conditional generative adversarial neural networks combined with quantum computing has not yet been involved.

[0005] Existing technologies use machine learning models on classical computers to learn and generate signal peptides. Using classical neural networks to learn and generate signal peptide structural information requires the design of more complex neural network models and hyperparameter adjustment techniques, and the amount of data required for training is large. Classical computers use classical bits for calculations, which are quite different from the quantum bits used by quantum computers. In the case of the same bit, quantum computers can process and generate more data due to their huge Hilbert space, and the randomness of classical calculations is not completely random, while quantum computers have completely random characteristics when measuring based on their entanglement and superposition characteristics. Summary of the invention

[0006] The technical problem to be solved by the present invention is to address the deficiencies of the prior art and specifically provide a method for generating a signal peptide based on quantum-assisted conditional generation of an adversarial neural network, as follows:

[0007] 1) In the first aspect, the present invention provides a method for generating a signal peptide based on a quantum-assisted conditional generation of a neural network, and the specific technical scheme is as follows:

[0008] S1. Obtain quantum random noise based on the signal peptide database;

[0009] S2, inputting the signal peptide category label information and quantum random noise into the classical generative neural network to generate data features of false signal peptides, wherein the dimensions of the data features of the false signal peptides are the same as the dimensions of the data features of the real signal peptides;

[0010] S3, inputting the data features of the false signal peptide and the category label information corresponding to the false signal peptide into the classical discriminant neural network to obtain a first prediction result, and inputting the data features of the real signal peptide and the category label information corresponding to the real signal peptide into the classical discriminant neural network to obtain a second prediction result;

[0011] S4, combining the first prediction result and the second prediction result into a first loss function for training a classical discriminant neural network, and updating parameters of the classical discriminant neural network according to the first loss function;

[0012] S5. When the number of times the parameters of the classical discriminant neural network are updated reaches a preset number threshold, the false signal peptide feature information and the category label information corresponding to the false signal peptide feature information are input into the current classical discriminant neural network to obtain a prediction result, and a second loss function for training the classical generative neural network is constructed according to the prediction result. The parameters of the classical generative neural network are updated according to the second loss function, so that the current classical discriminant neural network discriminates the false signal peptide feature information as true;

[0013] S6, returning to execute S1, until the latest obtained classical discriminant neural network cannot determine whether the input is the generated false signal peptide feature information;

[0014] S7. Use the latest classical generative neural network combined with quantum random noise to generate data features of signal peptides.

[0015] The beneficial effects of the method for generating a signal peptide against a neural network based on quantum-assisted conditional generation provided by the present invention are as follows:

[0016] The use of quantum computing, a new computing model based on the basic principles of quantum mechanics, can ensure the quality of the data features of the generated signal peptides. A real quantum computer can be used, and the huge Hilbert space and complete randomness of the real quantum computer can be utilized. Moreover, data features of signal peptides of different categories such as bacteria and viruses can be generated.

[0017] On the basis of the above scheme, the signal peptide generation method based on quantum-assisted conditional generation of adversarial neural networks of the present invention can also be improved as follows.

[0018] Furthermore, according to the signal peptide database, quantum random noise is obtained, including:

[0019] Preprocessing the signal peptide database to obtain characteristic information for characterizing real signal peptides;

[0020] Based on the characteristic information of the real signal peptide, a quantum circuit is constructed;

[0021] All quantum bits in the quantum circuit are measured to obtain measurement results, which are used as quantum random noise.

[0022] Furthermore, according to the characteristic information characterizing the real signal peptide, a quantum circuit is constructed, including: determining the number of quantum bits according to the size of the characteristic information characterizing the real signal peptide, and constructing the quantum circuit according to the number of quantum bits.

[0023] Furthermore, before updating the parameters of the classical generative neural network according to the second loss function, the method also includes: fixing the parameters of the current classical discriminative neural network.

[0024] 2) In the second aspect, the present invention also provides a signal peptide generation system based on quantum-assisted conditional generation of adversarial neural networks, and the specific technical scheme is as follows:

[0025] Including: quantum random noise acquisition module, training module and signal peptide data feature generation module;

[0026] The quantum random noise acquisition module is used to: obtain quantum random noise according to the signal peptide database;

[0027] The training module is used to: input the signal peptide category label information and quantum random noise into the classical generative neural network to generate data features of false signal peptides, the dimension of the data features of the false signal peptides is the same as the dimension of the data features of the real signal peptides; input the data features of the false signal peptides and the category label information corresponding to the false signal peptides into the classical discriminant neural network to obtain a first prediction result, input the data features of the real signal peptides and the category label information corresponding to the real signal peptides into the classical discriminant neural network to obtain a second prediction result; combine the first prediction result and the second prediction result into a first loss function for training the classical discriminant neural network, and update the parameters of the classical discriminant neural network according to the first loss function; when the number of times the parameters of the classical discriminant neural network are updated reaches a preset number threshold, input the false signal peptide feature information and the category label information corresponding to the false signal peptide feature information into the current classical discriminant neural network to obtain a prediction result, construct a second loss function for training the classical generative neural network according to the prediction result, update the parameters of the classical generative neural network according to the second loss function, so that the current classical discriminant neural network discriminates the false signal peptide feature information as true, and re-call the quantum random noise acquisition module until the latest classical discriminant neural network cannot determine whether the input is the generated false signal peptide feature information;

[0028] The signal peptide data feature generation module is used to generate data features of signal peptides using the latest classical generative neural network combined with quantum random noise.

[0029] On the basis of the above scheme, the signal peptide generation system based on quantum-assisted conditional generation of adversarial neural networks of the present invention can also be improved as follows.

[0030] Furthermore, the quantum random noise acquisition module is specifically used for:

[0031] Preprocessing the signal peptide database to obtain characteristic information for characterizing real signal peptides;

[0032] Based on the characteristic information of the real signal peptide, a quantum circuit is constructed;

[0033] All quantum bits in the quantum circuit are measured to obtain measurement results, which are used as quantum random noise.

[0034] Furthermore, the quantum random noise acquisition module is also specifically used to: determine the number of quantum bits according to the size of characteristic information characterizing the real signal peptide, and construct a quantum circuit according to the number of quantum bits.

[0035] Furthermore, the training module is also specifically used to: fix the parameters of the current classical discriminant neural network before updating the parameters of the classical generative neural network according to the second loss function.

[0036] 3) In a third aspect, the present invention also provides a computer device, comprising a processor, the processor being coupled to a memory, the memory storing at least one computer program, and the at least one computer program being loaded and executed by the processor, so that the computer device implements any of the above-mentioned signal peptide generation methods based on quantum-assisted conditional generation of adversarial neural networks.

[0037] 4) In a fourth aspect, the present invention further provides a computer-readable storage medium, in which at least one computer program is stored, and at least one computer program is loaded and executed by a processor so that a computer implements any of the above-mentioned signal peptide generation methods based on quantum-assisted conditional generation of adversarial neural networks.

[0038] It should be noted that the beneficial effects achieved by the technical solutions of the second to fourth aspects of the present invention and the corresponding possible implementation methods can be found in the above-mentioned technical effects of the first aspect and its corresponding possible implementation methods, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0040] Figure 1 A schematic flow chart of a method for generating a signal peptide based on quantum-assisted conditional generation of an adversarial neural network according to an embodiment of the present invention;

[0041] Figure 2 A quantum circuit for generating quantum random noise;

[0042] Figure 3 Generate network structures for quantum-assisted conditional adversarial neural networks;

[0043] Figure 4 Generate the network structure of a classic neural network;

[0044] Figure 5 It is the network structure of the classic discriminant neural network;

[0045] Figure 6 The figure is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0047] like Figure 1 As shown, a method for generating a signal peptide based on a quantum-assisted conditional generative adversarial neural network according to an embodiment of the present invention comprises the following steps:

[0048] S1. According to the signal peptide database, quantum random noise is obtained, specifically:

[0049] S10, preprocessing the signal peptide database to obtain characteristic information for characterizing the real signal peptide;

[0050] S11. Constructing a quantum circuit according to characteristic information characterizing a real signal peptide. Specifically, determining the number of quantum bits according to the size of the characteristic information characterizing a real signal peptide, and constructing a quantum circuit according to the number of quantum bits.

[0051] S12. Measure all quantum bits in the quantum circuit to obtain measurement results, and use the measurement results as quantum random noise.

[0052] Among them, the process of preprocessing the signal peptide database is as follows:

[0053] Each signal peptide is obtained from the signal peptide database, i.e., the Signal Peptide database. Based on one-hot encoding, each signal peptide is encoded into a feature vector to obtain feature information for characterizing the real signal peptide. For example, assuming that the signal peptide contains 30 amino acids, after one-hot encoding, a 30×20-dimensional feature vector representing the real signal peptide is obtained, i.e., the feature information characterizing the real signal peptide.

[0054] The process of constructing a quantum circuit is as follows:

[0055] Based on R y (θ) and R x (θ) Rotation quantum gate and CZ quantum gate are used to construct a quantum circuit for generating quantum random noise. Here, 10 quantum bits are used. The specific quantum circuit is as follows: Figure 2 shown.

[0056] S2, inputting the signal peptide category label information and quantum random noise into the classical generative neural network to generate data features of false signal peptides, wherein the dimensions of the data features of the false signal peptides are the same as the dimensions of the data features of the real signal peptides;

[0057] S3, inputting the data features of the false signal peptide and the category label information corresponding to the false signal peptide into the classical discriminant neural network to obtain a first prediction result, and inputting the data features of the real signal peptide and the category label information corresponding to the real signal peptide into the classical discriminant neural network to obtain a second prediction result;

[0058] The specific setting method of the category label information of the signal peptide is: the category label information of the bacterial signal peptide is marked as 1, and the category label information of the viral signal peptide is marked as 0. The category label information corresponding to the false signal peptide can be set according to the actual situation.

[0059] The quantum-assisted conditional generative adversarial neural network includes a classical generative neural network and a classical discriminative neural network. The classical generative neural network is also called a generator, and the classical discriminative neural network is also called a discriminator. The overall architecture is as follows: Figure 3 shown.

[0060] Figure 3 In the figure, DNN represents a classical discriminant neural network, GNN represents a classical generative neural network, quantum noise refers to quantum random noise, real data refers to the data features of real signal peptides, and labels refer to category label information. The signal peptide category label information and quantum random noise are input into the classical generative neural network to generate data features of fake signal peptides with a data dimension of 30×20. The data features of the fake signal peptides and the category label information corresponding to the fake signal peptides are then input into the classical discriminant neural network to obtain a first prediction result. The data features of the real signal peptides with a dimension of 30×20 obtained through the signal peptide database and the category label information corresponding to the real signal peptides are input into the classical discriminant neural network to obtain a second prediction result.

[0061] Among them, the classic generative neural network includes multi-layer classic fully connected neural network, such as Figure 4 As shown, the number of neural network nodes in each layer of the classical fully connected neural network is [11, 200, 400, 600], the dimension of the input layer is 11, which is used to input an 11-dimensional feature vector composed of quantum random noise and category label information corresponding to false signal peptides, and the dimension of the output layer is 30×20, which is used to generate data features of false signal peptides.

[0062] After obtaining the data features of the false signal peptide, the data features of the false signal peptide and the category label information corresponding to the false signal peptide are input into the classical discriminant neural network to obtain the first prediction result, and the data features of the real signal peptide and the category label information corresponding to the real signal peptide are input into the classical discriminant neural network to obtain the second prediction result. The network structure of the classical discriminant neural network is as follows: Figure 5As shown, the classical discriminant neural network includes a multi-layer classical fully connected neural network, and the number of neural network nodes of each layer of the classical fully connected neural network is [601, 400, 200, 10, 1]. The dimension of the input layer is 601, which is used to input 601-dimensional data features of real signal peptides and fake signal peptides and their corresponding category label information. The dimension of the last hidden layer is 10, which is used to correspond to the dimension of the quantum random noise generated by the quantum circuit. The dimension of the output layer is 1. The output layer performs a sigmoid function on the data features of the fake signal peptides and the data features of the real signal peptides input by the input layer to obtain a prediction result of the input fake signal peptide data features, i.e., a first prediction result, and a prediction result of the input real signal peptide data features, i.e., a second prediction result. The prediction results can be used to determine whether the input signal peptide data features are fake signal peptide data features or real signal peptide data features. The prediction result of 1 indicates that the discriminator determines that the input is the real signal peptide data feature, and the result of 0 indicates that the discriminator determines that the input is the fake signal peptide data feature.

[0063] S4, combining the first prediction result and the second prediction result into a first loss function for training a classical discriminant neural network, and updating parameters of the classical discriminant neural network according to the first loss function;

[0064] Among them, the first loss function LD is: LD = -E x [log(D(x))]-E z [log(1-D(G(z)))],E x [log(D(x))] represents: the average of the second prediction results D(x) of the classic discriminant neural network, i.e., the discriminator predicts the feature information x composed of the data features and category label information of each real signal peptide in the training data set batch, E z [log(1-D(G(z)))] represents: the average of the first prediction results D(G(z)) of the classical discriminant neural network, i.e., the discriminator, for the feature information z composed of the data features and the category label information of each false signal peptide in the training data set batch, and the batch size is 100, which can also be set according to the actual situation. Then, the parameters of the classical discriminant neural network are updated based on the first loss function.

[0065] S5. When the number of times the parameters of the classical discriminant neural network are updated reaches a preset number threshold, the false signal peptide feature information and the category label information corresponding to the false signal peptide feature information are input into the current classical discriminant neural network to obtain a prediction result, and a second loss function for training the classical generative neural network is constructed according to the prediction result. The parameters of the classical generative neural network are updated according to the second loss function, so that the current classical discriminant neural network discriminates the false signal peptide feature information as true;

[0066] S6, returning to execute S1, until the latest obtained classical discriminant neural network cannot determine whether the input is the generated false signal peptide feature information;

[0067] Among them, the preset number threshold can be set according to actual conditions.

[0068] Before updating the parameters of the classical generative neural network according to the second loss function, the method further includes: fixing the parameters of the current classical discriminative neural network.

[0069] Among them, the second loss function is: LG = -E z [log(D(G(z)))], update the parameters of the classical generative neural network according to the second loss function, so that the current classical discriminant neural network can identify the false signal peptide feature information as true. The generator, i.e., the classical generative neural network, and the discriminator, i.e., the classical discriminant neural network, are in an adversarial relationship. The purpose of the classical discriminant neural network is to distinguish whether the input is the data feature of a real signal peptide or the data feature of a false signal peptide, and the purpose of the classical generative neural network is to allow the classical discriminant neural network to recognize the data feature of the false signal peptide generated by it as true. In other words, the newly obtained classical discriminant neural network cannot distinguish the data feature of the false signal peptide generated by the newly obtained classical generative neural network as false.

[0070] The distribution of quantum random noise generated by the quantum circuit is represented by q θ Indicates that the distribution of the data output by the last hidden layer of the classic discriminant neural network is p, and q is obtained θ The cross entropy loss function of p and p is used to update the parameters of the quantum circuit based on the cross entropy loss function. Then, the back propagation algorithm is used to update the parameters of the fully connected neural network in the classical generative neural network and the classical discriminative neural network. Specifically:

[0071] The quantum circuit can generate 10-dimensional quantum random noise data, and its corresponding distribution is q θ Similarly, the last hidden layer of the discriminator can also output 10-dimensional feature data, whose corresponding distribution is represented by p, and then q can be obtained θ The cross entropy loss function of p is L(θ) = -E x[p log max(q θ ,ε)], where E x [] represents the cross entropy loss function and the average of the training data sample x, ε represents a small value, here 1e-14, and then the SPSA (Simultaneous perturbation stochastic approximation) approximation algorithm is used to update the constructed cross entropy loss function Figure 2 Parameters of the quantum circuit used to generate quantum random noise are shown.

[0072] Then, the back propagation algorithm is used to update the parameters of the fully connected neural network in the generator and the discriminator. In other words, the back propagation algorithm is used to update the parameters of the fully connected neural network in the classic generative neural network and the classic discriminative neural network. Finally, the discriminator and the generator are trained against each other for multiple times and stopped after convergence.

[0073] S7. Use the latest classical generative neural network combined with quantum random noise to generate data features of signal peptides.

[0074] The present invention adopts a self-designed quantum-assisted conditional generative adversarial neural network. First, the downloaded signal peptide database is encoded into a feature vector based on one-hot as feature information characterizing the real signal peptide. Then, based on the size of this feature information, the number of quantum bits used is selected. Then, a quantum circuit is constructed and all quantum bits are measured. The measurement result is used as quantum random noise. Then, the signal peptide category label information and quantum random noise are input into a classical generative neural network to generate false signal peptide data features with the same dimension as the real signal peptide data. Then, the signal peptide feature information generated by the classical generative neural network and the real signal peptide feature information after one-hot encoding and the corresponding category label information are respectively input into the constructed classical discriminant neural network. Then, the classical discriminant neural network is output to obtain the prediction results of the real data and the data generated by the classical generative neural network, and the discrimination results of the real data and the generated data are combined into the loss function of training the classical discriminant neural network. According to the obtained loss function corresponding to the classical discriminant neural network, the parameters of the classical discriminant neural network are updated so that the discriminator can well identify whether the input is a real signal peptide or a signal peptide generated by the generator. After the parameters of the classical discriminant neural network are updated for a set number of times, the signal peptide feature data generated by the classical generative neural network and its corresponding signal peptide category label information are input into the classical discriminant neural network, and the loss function for training the classical generative neural network is constructed according to the prediction results of the generated data output by the classical discriminant neural network. Then, the parameters of the discriminator are fixed according to this loss function, and the parameters of the generator are updated so that the discriminator can identify the generated signal peptide as true. Through the continuous confrontation between the generator and the discriminator, the discriminator will not be able to distinguish whether the input is a generated signal peptide or a real signal peptide. Finally, the trained generator can be used to generate a new signal peptide that is different from but similar to the real signal peptide and corresponds to the signal peptide category label of the input generator through the quantum random noise generated by the input quantum circuit.

[0075] The beneficial effects of the present invention are as follows:

[0076] 1) Construct an efficient quantum-assisted conditional generative adversarial neural network model to generate signal peptides with different structures, which can greatly accelerate the drug development process.

[0077] 2) The quantum-assisted conditional generative adversarial neural network proposed in the present invention adopts a brand-new computing model based on the basic principles of quantum mechanics, namely quantum computing. It is expected to surpass the classical machine learning algorithm in terms of the effect of signal peptide generation. In addition, the present invention adopts a classical neural network to perform the discriminator and generator tasks, and only generates quantum random noise through a quantum system. Therefore, the present invention can use a real quantum computer to utilize its huge Hilbert space and complete randomness for the signal peptide generation task. In addition, the present invention introduces conditional information corresponding to the signal peptide in the generative adversarial neural network, so that signal peptides corresponding to the characteristics of different categories (bacteria and viruses) can be generated.

[0078] 3) Currently, there has been no specific research on the conditional generation of signal peptides in combination with quantum computing, and the stability of quantum-assisted conditional generation adversarial neural network training is expected to exceed that of pure classical machine learning algorithms.

[0079] In the above embodiments, although the steps are numbered S1, S2, etc., these are only specific embodiments given by the present invention. Those skilled in the art may adjust the execution order of S1, S2, etc. according to actual conditions, which is also within the protection scope of the present invention. It can be understood that in some embodiments, some or all of the above embodiments may be included.

[0080] A signal peptide generation system based on a quantum-assisted conditional generative adversarial neural network according to an embodiment of the present invention comprises: a quantum random noise acquisition module, a training module and a signal peptide data feature generation module;

[0081] The quantum random noise acquisition module is used to: obtain quantum random noise according to the signal peptide database;

[0082] The training module is used to: input the signal peptide category label information and quantum random noise into the classical generative neural network to generate data features of false signal peptides, the dimension of the data features of the false signal peptides is the same as the dimension of the data features of the real signal peptides; input the data features of the false signal peptides and the category label information corresponding to the false signal peptides into the classical discriminant neural network to obtain a first prediction result, input the data features of the real signal peptides and the category label information corresponding to the real signal peptides into the classical discriminant neural network to obtain a second prediction result; combine the first prediction result and the second prediction result into a first loss function for training the classical discriminant neural network, and update the parameters of the classical discriminant neural network according to the first loss function; when the number of times the parameters of the classical discriminant neural network are updated reaches a preset number threshold, input the false signal peptide feature information and the category label information corresponding to the false signal peptide feature information into the current classical discriminant neural network to obtain a prediction result, construct a second loss function for training the classical generative neural network according to the prediction result, update the parameters of the classical generative neural network according to the second loss function, so that the current classical discriminant neural network discriminates the false signal peptide feature information as true, and re-call the quantum random noise acquisition module until the latest classical discriminant neural network cannot determine whether the input is the generated false signal peptide feature information;

[0083] The signal peptide data feature generation module is used to generate data features of signal peptides using the latest classical generative neural network combined with quantum random noise.

[0084] Optionally, in the above technical solution, the quantum random noise acquisition module is specifically used for:

[0085] Preprocessing the signal peptide database to obtain characteristic information for characterizing real signal peptides;

[0086] Based on the characteristic information of the real signal peptide, a quantum circuit is constructed;

[0087] All quantum bits in the quantum circuit are measured to obtain measurement results, which are used as quantum random noise.

[0088] Optionally, in the above technical solution, the quantum random noise acquisition module is further specifically used to: determine the number of quantum bits according to the size of characteristic information characterizing the real signal peptide, and construct a quantum circuit according to the number of quantum bits.

[0089] Optionally, in the above technical solution, the training module is further specifically used to: fix the parameters of the current classical discriminant neural network before updating the parameters of the classical generative neural network according to the second loss function.

[0090] It should be noted that the beneficial effects of the signal peptide generation system based on quantum-assisted conditional generation of antagonistic neural networks provided in the above embodiment are the same as the beneficial effects of the signal peptide generation method based on quantum-assisted conditional generation of antagonistic neural networks, which will not be repeated here. In addition, when the system provided in the above embodiment realizes its functions, it only takes the division of the above-mentioned functional modules as an example. In practical applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the system can be divided into different functional modules according to actual conditions to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0091] In another embodiment, it includes a real signal peptide database data preprocessing and random noise quantum circuit building module, a generator and discriminator building module, a discriminator training module, a generator training module, and a model parameter updating module.

[0092] The real signal peptide data preprocessing and random noise quantum circuit construction module is used to: encode each signal peptide in the downloaded signal peptide database into a feature vector based on one-hot encoding. According to the size of the signal peptide feature dimension after one-hot encoding, select the appropriate number of quantum bits, construct a quantum circuit to generate quantum random noise, so as to generate a fake signal peptide through the generator designed later.

[0093] The generator and discriminator construction modules are used to: construct a generator and a discriminator according to the characteristics of the real data set, so as to generate false signal peptide features and distinguish between real signal peptides and generated signal peptides.

[0094] The training discriminator module is used to: combine the discriminator's discrimination results of the real signal peptide data and the generated signal peptide data input into a loss function, and then update the discriminator's parameters based on the loss function so that the discriminator can distinguish whether the input is a real signal peptide or a signal peptide generated by the generator.

[0095] The training generator module is used to: input the signal peptide feature data generated by the generator and the corresponding category label information of the signal peptide into the classical discriminant neural network to obtain the prediction result, construct the loss function based on this result, and then fix the parameters of the discriminator. Based on this loss function, the parameters of the generator are updated so that the discriminator can identify the signal peptide generated by the generator as true.

[0096] The model parameter updating module is used to: update the parameters of the quantum circuit that generates quantum random noise based on the SPSA approximation algorithm, the parameters of the generator and the discriminator can be updated through the back propagation algorithm of the machine learning framework, and finally, the discriminator and the generator are trained adversarially for multiple times, so that the signal peptide generation method based on the quantum-assisted conditional generative adversarial neural network proposed in the present invention stops after convergence.

[0097] This is illustrated by the following examples.

[0098] Real signal peptide data preprocessing and random noise quantum circuit building modules:

[0099] Each signal peptide downloaded from the Signal Peptide database is encoded as a feature vector based on one-hot encoding. Here, it is assumed that the signal peptide contains 30 amino acids. After one-hot encoding, a 30×20-dimensional feature vector representing the signal peptide will be obtained. y (θ) and R x (θ) Rotation quantum gate and CZ quantum gate are used to construct a quantum circuit for generating quantum random noise. Here, 10 quantum bits are used. The specific quantum circuit is as follows: Figure 2 As shown, for the signal peptide tag, the bacterial signal peptide tag is marked as 1 and the viral signal peptide tag is marked as 0.

[0100] Generator and discriminator building blocks:

[0101] The overall architecture of the quantum-assisted conditional generative adversarial neural network designed by the present invention is as follows: Figure 3 As shown, DNN represents a classical discriminative neural network, GNN represents a classical generative neural network, quantum random noise and signal peptide category label information are input into the generator to generate a 30×20-dimensional fake signal peptide feature vector, and a 30×20-dimensional real signal peptide feature vector is obtained based on the one-hot encoding of the real signal peptide database in the previous module. Then, the generated signal peptide feature vector and the real signal peptide feature vector are combined with their respective category labels and input into the discriminator to obtain the discriminator's prediction result for the input data.

[0102] The generator is implemented using a fully connected neural network. Figure 4 As shown in the figure, it is composed of a multi-layer classic fully connected neural network. The number of neural network nodes in each layer is [11, 200, 400, 600]. The dimension of the input layer is 11, which is used to input 11-dimensional data from Figure 1 The quantum random noise measured by the quantum circuit shown is combined with the characteristic information of the signal peptide category label, and the dimension of the output layer is 30×20 to generate false signal peptide data.

[0103] After obtaining the false signal peptide data generated by the classical generative neural network, the generated false signal peptide and true signal peptide feature data can be combined with their corresponding signal peptide category labels and input into the discriminator. The structure of the discriminator is as follows: Figure 5 As shown in the figure, it is composed of a multi-layer classical fully connected neural network, and the number of neural network nodes in each layer is [601, 400, 200, 10, 1]. The dimension of the input layer is 601, which is used to input 601-dimensional real and generated signal peptide feature data and their corresponding category labels. The dimension of the last hidden layer is 10, which corresponds to the dimension of the quantum random noise generated by the quantum circuit. The dimension of the output layer is 1. By executing a sigmoid function on the output as the prediction result of the input signal peptide feature data, it is judged whether the input signal peptide is a real signal peptide or a false signal peptide. The prediction result of 1 indicates that the input is a real signal peptide feature, and the result of 0 indicates that the input is a false signal peptide feature.

[0104] Train the discriminator module:

[0105] The real signal peptide data features of 600 dimensions after one-hot encoding and the signal peptide feature data generated by the generator are combined with their corresponding category label information and input as follows Figure 5 In the discriminator shown, the prediction results of the discriminator for the real signal peptide input sample and the generated signal peptide feature data sample are obtained respectively, and then the two prediction results are combined into a loss function, that is, the first loss function is shown as follows:

[0106] LD=-E x [log(D(x))]-E z [log(1-D(G(z)))]

[0107] Among them, the symbol E x / z [ ] represents the average of the training data set batch of the discriminator's prediction results for the real or generated signal peptide feature data, where the batch size is 500. Where x represents the combination of the real signal peptide data feature and its corresponding label information, z represents the combination of the quantum random noise and label information input to the generator, and G(z) represents the signal peptide feature data generated by the generator after operating the quantum random noise and label information input therein. D(x) represents the discriminator's prediction result for the real signal peptide sample, and D(G(z)) represents the discriminator's prediction result for the generated signal peptide sample. Then, based on the above-constructed loss function for the discriminator, the parameters of the discriminator are updated so that the discriminator can distinguish whether the input is a real signal peptide or a signal peptide generated by the generator.

[0108] Training Generator Module:

[0109] The combined information of 11-dimensional quantum random noise and signal peptide category labels is input into the generator, and then the generated signal peptide feature data and signal peptide category label information are input into the discriminator together to obtain the prediction result of the discriminator to construct the loss function of the generator, i.e., the first loss function, as shown in the following formula:

[0110] LG=-E z [log(D(G(z)))]

[0111] Then, the parameters of the fixed discriminator are updated based on this loss function to update the parameters of the generator, so that the discriminator can identify the signal peptide generated by the generator as true. The generator and the discriminator are in an adversarial relationship. The purpose of the discriminator is to distinguish whether the input is true signal peptide data or false signal peptide data, and the purpose of the generator is to allow the discriminator to identify the signal peptide it generates as true.

[0112] Update model parameter module:

[0113] pass Figure 2 The quantum circuit can generate 10-dimensional quantum random noise data, and its corresponding distribution is q θ Indicates that, similarly, the last hidden layer of the discriminator can also output 10-dimensional feature data, and its corresponding distribution is represented by p, and then q can be obtained θ The cross entropy loss function of p is L(θ) = -E x [p log max(q θ ,ε)], where E x [] represents the cross entropy loss function and the average obtained based on the training data sample x. Then the SPSA approximation algorithm is used to update the constructed cross entropy loss function Figure 2 Parameters of the quantum circuit for generating quantum random noise are shown.

[0114] Then, the back propagation algorithm can be used to update the parameters of the fully connected neural network in the generator and the discriminator. Finally, the discriminator and the generator are trained against each other for multiple times, so that the signal peptide generation method based on the quantum-assisted conditional generation adversarial neural network proposed in the present invention stops after convergence, so that the trained generator of the present invention can be used to generate new signal peptide data corresponding to bacteria and viruses by inputting signal peptide category labels and quantum random noise generated by quantum circuits.

[0115] like Figure 6As shown, a computer device 300 according to an embodiment of the present invention includes a processor 320, the processor 320 is coupled to a memory 310, and the memory 310 stores at least one computer program 330, and the at least one computer program 330 is loaded and executed by the processor 320, so that the computer device 300 implements any of the above-mentioned signal peptide generation methods based on quantum-assisted conditional generation of adversarial neural networks, specifically:

[0116] The computer device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors 320 (Central Processing Units, CPU) and one or more memories 310, wherein at least one computer program 330 is stored in the one or more memories 310, and the at least one computer program 330 is loaded and executed by the one or more processors 320, so that the computer device 300 implements any of the signal peptide generation methods based on quantum-assisted conditional generation of adversarial neural networks provided in the above embodiments. Of course, the computer device 300 may also have components such as a wired or wireless network interface, a keyboard, and an input and output interface for input and output, and the computer device 300 may also include other components for implementing device functions, which will not be repeated here.

[0117] A computer-readable storage medium according to an embodiment of the present invention stores at least one computer program, and the at least one computer program is loaded and executed by a processor so that a computer implements any of the above-mentioned signal peptide generation methods based on quantum-assisted conditional generation of adversarial neural networks.

[0118] Optionally, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0119] In an exemplary embodiment, a computer program product or a computer program is also provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes any of the above-mentioned signal peptide generation methods based on quantum-assisted conditional generation of adversarial neural networks.

[0120] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects and represent the definition of a specific order or sequence. The order of use of similar objects can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than the order shown or described.

[0121] Those skilled in the art know that the present invention can be implemented as a system, method or computer program product. Therefore, the present disclosure can be specifically implemented in the following forms, namely: it can be complete hardware, it can be complete software (including firmware, resident software, microcode, etc.), or it can be a combination of hardware and software, which is generally referred to as "circuit", "module" or "system" herein. In addition, in some embodiments, the present invention can also be implemented in the form of a computer program product in one or more computer-readable media, and the computer-readable medium contains computer-readable program code.

[0122] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0123] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A method for generating signal peptides based on quantum-assisted conditional generative adversarial neural networks, characterized in that: include: S1. Obtain quantum random noise based on the signal peptide database; S2, inputting the signal peptide category label information and quantum random noise into a classical generative neural network to generate data features of a false signal peptide, wherein the dimension of the data features of the false signal peptide is the same as the dimension of the data features of the real signal peptide; S3, inputting the data features of the false signal peptide and the category label information corresponding to the false signal peptide into the classical discriminant neural network to obtain a first prediction result, and inputting the data features of the real signal peptide and the category label information corresponding to the real signal peptide into the classical discriminant neural network to obtain a second prediction result; S4, combining the first prediction result and the second prediction result into a first loss function for training the classical discriminant neural network, and updating the parameters of the classical discriminant neural network according to the first loss function; S5. When the number of times the parameters of the classical discriminant neural network are updated reaches a preset number threshold, the false signal peptide feature information and the category label information corresponding to the false signal peptide feature information are input into the current classical discriminant neural network to obtain a prediction result, and a second loss function for training the classical generative neural network is constructed according to the prediction result. The parameters of the classical generative neural network are updated according to the second loss function, so that the current classical discriminant neural network discriminates the false signal peptide feature information as true; S6, returning to execute S1, until the latest obtained classical discriminant neural network cannot determine whether the input is the generated false signal peptide feature information; S7, using the latest classical generative neural network and combining the quantum random noise to generate data features of the signal peptide; According to the signal peptide database, quantum random noise is obtained, including: Preprocessing the signal peptide database to obtain characteristic information for characterizing real signal peptides; Obtain each signal peptide from the signal peptide database, encode each signal peptide into a feature vector based on one-hot encoding, and obtain feature information for characterizing the real signal peptide; Based on the characteristic information of the real signal peptide, a quantum circuit is constructed; All quantum bits in the quantum circuit are measured to obtain measurement results, which are used as the quantum random noise.

2. The method for generating signal peptides based on quantum-assisted conditional generative adversarial neural networks according to claim 1, characterized in that: According to the characteristic information characterizing the real signal peptide, a quantum circuit is constructed, including: determining the number of quantum bits according to the size of the characteristic information characterizing the real signal peptide, and constructing the quantum circuit according to the number of quantum bits.

3. The method for generating signal peptides based on quantum-assisted conditional generative adversarial neural networks according to claim 2, characterized in that: Before updating the parameters of the classical generative neural network according to the second loss function, the method also includes: fixing the parameters of the current classical discriminative neural network.

4. A signal peptide generation system based on quantum-assisted conditional generation of adversarial neural networks, characterized in that: include: Quantum random noise acquisition module, training module and signal peptide data feature generation module; The quantum random noise acquisition module is used to: obtain quantum random noise according to a signal peptide database; The training module is used to: input the signal peptide category label information and quantum random noise into the classical generative neural network to generate fake signal peptide data features, the dimension of the fake signal peptide data features is the same as the dimension of the real signal peptide data features; input the fake signal peptide data features and the category label information corresponding to the fake signal peptide into the classical discriminant neural network to obtain a first prediction result, input the real signal peptide data features and the category label information corresponding to the real signal peptide into the classical discriminant neural network to obtain a second prediction result; combine the first prediction result and the second prediction result into a first loss function for training the classical discriminant neural network, and update the first loss function according to the first loss function. The parameters of the classical discriminant neural network are updated; when the number of times the parameters of the classical discriminant neural network are updated reaches a preset number threshold, the false signal peptide characteristic information and the category label information corresponding to the false signal peptide characteristic information are input into the current classical discriminant neural network to obtain a prediction result, a second loss function for training the classical generative neural network is constructed according to the prediction result, the parameters of the classical generative neural network are updated according to the second loss function, so that the current classical discriminant neural network discriminates the false signal peptide characteristic information as true, and the quantum random noise acquisition module is re-called until the latest classical discriminant neural network cannot determine whether the input is the generated false signal peptide characteristic information; The signal peptide data feature generation module is used to: generate data features of the signal peptide using the latest classical generative neural network combined with the quantum random noise; The quantum random noise acquisition module is specifically used for: Preprocessing the signal peptide database to obtain characteristic information for characterizing real signal peptides; Obtain each signal peptide from the signal peptide database, encode each signal peptide into a feature vector based on one-hot encoding, and obtain feature information for characterizing the real signal peptide; Based on the characteristic information of the real signal peptide, a quantum circuit is constructed; All quantum bits in the quantum circuit are measured to obtain measurement results, which are used as the quantum random noise.

5. A signal peptide generation system based on quantum-assisted conditional generative adversarial neural network according to claim 4, characterized in that: The quantum random noise acquisition module is also specifically used to: determine the number of quantum bits according to the size of the characteristic information characterizing the real signal peptide, and construct a quantum circuit according to the number of quantum bits.

6. A signal peptide generation system based on quantum-assisted conditional generative adversarial neural network according to claim 4, characterized in that: The training module is also specifically used to: fix the parameters of the current classical discriminant neural network before updating the parameters of the classical generative neural network according to the second loss function.

7. A computer device, characterized in that: The computer device includes a processor, the processor is coupled to a memory, and at least one computer program is stored in the memory. The at least one computer program is loaded and executed by the processor so that the computer device implements a signal peptide generation method based on quantum-assisted conditional generation of adversarial neural networks as described in any one of claims 1 to 3.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one computer program, and the at least one computer program is loaded and executed by the processor so that the computer implements a signal peptide generation method based on quantum-assisted conditional generation of adversarial neural networks as described in any one of claims 1 to 3.

Images