A protein coding method based on position sequence matrix

Abstract

The application provides a protein coding method based on a position sequence matrix, which comprises the following steps: classifying amino acid sequences according to the dipole and volume of the side chain of the amino acid; constructing a sequence matrix and a position matrix; each element in the sequence matrix is used for indicating the frequency of two-by-two combination of all amino acids in the protein sequence; the position matrix is used for indicating the position information of any two groups of amino acids in the protein sequence; the diagonal line and the value above the diagonal line of the sequence matrix are selected together to code the amino acid sequence data, and the reciprocal of the one-dimensional sequence length is added as a component of the sequence matrix coding to distinguish the amino acid sequence length, so that the amino acid sequence data is coded into a feature vector. The position information and the sequence information of the protein sequence data are combined to code the protein sequence, so that the protein feature information can be fully obtained, the accuracy of protein interaction recognition is improved, and the robustness of the protein interaction prediction algorithm is enhanced.

Description

Technical Field

[0001] This invention relates to the field of protein coding technology, and more specifically, to a protein coding method based on positional sequence matrices. Background Technology

[0002] Protein coding methods transform heterogeneous amino acid sequences into homogeneous vector features, which are then input into machine learning models for training. Therefore, protein sequence coding methods are also a key factor affecting model performance.

[0003] Currently, commonly used protein coding methods include CT, AC, LD, and MOS, but each method has its own limitations. CT considers the sequence relationship of three adjacent amino acids, AC considers the sequence relationship of 30 amino acids, LD only considers the adjacency effect of two types of adjacent amino acids, and MOS considers the frequency information of the entire protein sequence. None of these methods consider the positional relationship of the entire protein sequence. Therefore, how to extract the positional information of the entire protein sequence is a key technical problem to improve the performance of protein interaction prediction. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies in predicting protein interactions, such as low performance and poor robustness, by providing a protein encoding method based on position sequence matrices to solve the above problems.

[0005] To achieve the above objectives, a first aspect of the present invention provides a protein encoding method based on a position sequence matrix, the method comprising:

[0006] S1. The amino acid sequences are classified according to the dipole and volume of their side chains.

[0007] S2, construct the sequence matrix and the position matrix; where each element in the sequence matrix is ​​used to indicate the frequency of all pairwise combinations of amino acids in the protein sequence; the position matrix is ​​used to indicate the position information of any two groups of amino acids in the protein sequence;

[0008] S3 simultaneously selects the values ​​on the diagonal of the sequence matrix and those above the diagonal to encode the amino acid sequence data, and adds the reciprocal of the one-dimensional sequence length as a component of the sequence matrix encoding to distinguish the amino acid sequence length, thereby encoding the amino acid sequence data into a feature vector.

[0009] Furthermore, the construction of the sequence matrix and the position matrix, wherein constructing the sequence matrix MOS includes:

[0010] Define a nonempty finite set: Ω = {w1, ..., w} N}, where N is the number of categories of the sequence;

[0011] Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L; the position matrix of sequence S is represented as:

[0012] MOS = [m ij ] N × N (1)

[0013] In the above formula, m ij :

[0014]

[0015] Based on the definition of a sequence matrix, the sum of all elements in the sequence matrix equals...

[0016] Therefore, for any two sequences, when the sequence lengths are different or the sequence lengths are the same but the number of at least one element is different, their corresponding sequence matrices are different.

[0017] Furthermore, the construction of the sequence matrix and the position matrix, wherein constructing the position matrix includes:

[0018] Define a nonempty finite set: Ω = {w1, ..., w} N}, where N is the number of categories of the sequence;

[0019] Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L; the position matrix of sequence S is represented as:

[0020] MOP = [x ij ] N*N (3)

[0021] In the above formula, x ij :

[0022]

[0023] The position matrix is ​​a symmetric matrix, and the elements on the diagonal and above the diagonal contain the position information of all elements in the amino acid sequence.

[0024] Furthermore, step S3 includes encoding the amino acid sequence data into a vector as follows:

[0025] MSP = [MOS] 11 ,...,MOS 1N ,...MOSNN ,1 / L]+[MOP 11 ,...,MOP 1N ...MOP NN ,1 / L].

[0026] Furthermore, a second aspect of the present invention provides a method for predicting protein interactions based on the protein-coding method described above using a positional sequence matrix, the method comprising:

[0027] The encoded protein sequence is used as input to the first deep neural network model, and the protein interaction classification result is output based on the first deep neural network model.

[0028] The first deep neural network model includes the following training steps:

[0029] In the forward propagation phase, samples are taken from the human protein dataset and used as input to the deep neural network. After successive transformations, the samples are transmitted to the output layer to calculate the actual output of layer l. The calculation formula is as follows:

[0030]

[0031] With dropout applied, the output of layer l is:

[0032]

[0033] Where l = 1, ..., N, a (l-1) For the input data of layer l, w l b is the connection weight matrix between layer (l-1) and layer l. l δ represents the bias of the l-th layer, and δ represents the activation of the l-th layer.

[0034] During the backpropagation phase, the actual output O is calculated. i and ideal output Y i The difference is then calculated, and the weight matrix is ​​adjusted by backpropagation using the method of minimizing the error. The error calculation formula for the output layer is as follows:

[0035]

[0036]

[0037] Among them, E i O ik T ik These are the error of the i-th sample, the output of the k-th neuron in the output layer of the i-th sample, and the expected output of the k-th neuron in the output layer of the i-th sample, respectively.

[0038] Furthermore, a third aspect of the present invention provides a protein-coding device based on a positional sequence matrix, the device comprising:

[0039] The classification module classifies the amino acid sequences based on the dipole and volume of the side chains of the amino acids.

[0040] The component module constructs a sequence matrix and a position matrix; where each element in the sequence matrix indicates the frequency of all pairwise combinations of amino acids in the protein sequence; the position matrix indicates the positional information of any two pairs of amino acids in the protein sequence.

[0041] The encoding module simultaneously selects the values ​​on the diagonal of the sequence matrix and those above the diagonal to encode the amino acid sequence data. It also adds the reciprocal of the one-dimensional sequence length as a component of the sequence matrix encoding to distinguish the length of the amino acid sequence, thereby encoding the amino acid sequence data into a feature vector.

[0042] Furthermore, a fourth aspect of the present invention provides an electronic device comprising: one or more processors, and a memory for storing one or more computer programs; the computer programs being configured to be executed by the one or more processors, the programs including methods for performing a protein coding method based on a position sequence matrix as described above, or methods for performing protein interaction prediction as described above.

[0043] Furthermore, a fifth aspect of the present invention provides a storage medium storing a computer program; the program is loaded and executed by a processor to implement the protein encoding method based on position sequence matrix as described above, or the protein interaction prediction method as described above.

[0044] In this invention, amino acid sequences are classified based on the dipole and volume of their side chains; a sequence matrix and a position matrix are constructed; each element in the sequence matrix indicates the frequency of pairwise combinations of all amino acids in the protein sequence; the position matrix indicates the positional information of any two groups of amino acids in the protein sequence; simultaneously, the values ​​on the diagonal of the sequence matrix and those above the diagonal are selected to encode the amino acid sequence data, and the reciprocal of the one-dimensional sequence length is added as a component of the sequence matrix encoding to distinguish the length of the amino acid sequence, thereby encoding the amino acid sequence data into a feature vector. By combining the positional and sequence information of the protein sequence data to encode the protein sequence, sufficient protein feature information can be obtained, improving the accuracy of protein interaction identification and enhancing the robustness of the protein interaction prediction algorithm. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 This is a schematic flowchart of the protein encoding method based on position sequence matrix disclosed in an embodiment of the present invention;

[0047] Figure 2 This is a flowchart of the position matrix calculation algorithm disclosed in an embodiment of the present invention;

[0048] Figure 3 This is a flowchart illustrating the overall execution of the position sequence matrix encoding method disclosed in this embodiment of the invention.

[0049] Figure 4 This is a schematic diagram of the structure of a protein coding device based on a position sequence matrix disclosed in an embodiment of the present invention;

[0050] Figure 5 This is a schematic diagram of the electronic device structure disclosed in an embodiment of the present invention. Detailed Implementation

[0051] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this application more comprehensive and complete, and to fully convey the concept of the exemplary embodiments to those skilled in the art.

[0052] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this application. However, those skilled in the art will recognize that the technical solutions of this application can be practiced without one or more of the specific details, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this application.

[0053] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0054] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0055] It should be noted that "multiple" as mentioned in this article refers to two or more.

[0056] The implementation details of the technical solutions in the embodiments of this application are described in detail below:

[0057] In this embodiment, before protein encoding, dataset collection and preprocessing are generally required. Data collection typically involves gathering positive and negative datasets from the Human Protein Reference Database (HPRD, 2007 version) and the Swiss-Prot (http: / / www.expasy.org / sprot / , version 57.3) protein reference database. Protein sequences with fewer than 50 amino acid residues are removed from the collected datasets, as are protein sequence pairs containing uncommon amino acids such as B, J, O, U, X, and Z. After processing, 36,591 pairs of positively correlated datasets and 36,324 pairs of negatively correlated datasets are obtained. Each time, 30,000 pairs of positively correlated datasets and 30,000 pairs of negatively correlated datasets are randomly selected to form the training dataset. 15% of the training set is randomly selected as the validation set, and the remainder is used as the test set to validate the model, thus implementing the method of this embodiment.

[0058] Please see Figure 1 , Figure 1 This is a schematic flowchart of a protein encoding method based on a position sequence matrix disclosed in an embodiment of the present invention. Figure 1 As shown, an embodiment of the present invention provides a protein encoding method based on a position sequence matrix, comprising:

[0059] S1. The amino acid sequences are classified according to the dipole and volume of their side chains.

[0060] Specifically, in this embodiment, 20 common amino acids are divided into 7 different groups based on the dipole and volume of their side chains, as shown in Table 1 below. Referring to the classification in the table below, the amino acid sequences are replaced with the categories corresponding to the amino acids in the amino acid classification table, and the dimension of the sequence matrix is ​​reduced from 20×20 to 7×7.

[0061] Table 1. Amino acid classification based on side chain dipolarity and volume.

[0062]

[0063]

[0064] S2, construct the sequence matrix and the position matrix; where each element in the sequence matrix is ​​used to indicate the frequency of all pairwise combinations of amino acids in the protein sequence; the position matrix is ​​used to indicate the position information of any two groups of amino acids in the protein sequence;

[0065] Furthermore, the construction of the sequence matrix and the position matrix, wherein constructing the sequence matrix MOS includes:

[0066] Suppose a nonempty finite set: Ω = {w1, ..., w} N}, where N is the number of categories of the sequence, which can be 7 here.

[0067] Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L; the position matrix of sequence S is represented as:

[0068] MOS = [m ij ] N×N (1)

[0069] In the above formula, m ij :

[0070]

[0071] Based on the definition of a sequence matrix, the sum of all elements in the sequence matrix equals...

[0072] Therefore, for any two sequences, when the sequence lengths are different or the sequence lengths are the same but the number of at least one element is different, their corresponding sequence matrices are different.

[0073] Suppose a nonempty finite set: Ω = {w1, ..., w7}, where w i This is an amino acid classification exercise. Given the sequence: S = S1, S2, ..., S... L Let L represent the length of sequence S, Si ∈ Ω, 1 ≤ i ≤ L. The sequence vector of sequence S can be represented as: in This represents the number of occurrences of wi in sequence S. Each element in the sequence matrix can be viewed as the frequency of all pairwise combinations of amino acids in the protein sequence, and can thus be solved iteratively using the sequence vectors of the right (left) terminal sequences of the protein sequence. The sequence matrix algorithm is calculated as follows:

[0074] The sequence matrix algorithm is calculated as follows:

[0075] Input sequence: S = S1, S2, ..., S L ;

[0076] Output sequence matrix: MOS = [m ij ] N×N .

[0077] Step 1. Initial value setting: i←L, VOS←VOS0=0, MOS←MOS0=0;

[0078] Step 2. VOS[s i ]←VOS[s i +1;

[0079] Step 3. MOS[s] i ]←MOS[s i ]+VOS;

[0080] Step 4. i ← i-1;

[0081] Step 5. If i≥1, go to step 2; otherwise, end.

[0082] Furthermore, the position matrix represents the positional information of any two groups of amino acids in a protein sequence. Therefore, each row (column) of the position matrix represents the positional distribution of each group of amino acids relative to the corresponding amino acid sequence in that row (column). This positional information includes two main categories: the order of amino acids within the same group and the positional information between amino acids from different groups.

[0083] Constructing the position matrix includes:

[0084] Suppose a nonempty finite set: Ω = {w1, ..., w} N}, where N is the number of categories of the sequence;

[0085] Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L; the position matrix of sequence S is represented as:

[0086] MOP = [x ij ] N*N (3)

[0087] In the above formula, x ij :

[0088]

[0089] Based on the definition of a position matrix, it is known that a position matrix is ​​a symmetric matrix, and the elements on the diagonal and above the diagonal contain the positional information of all elements in the amino acid sequence. For example... Figure 2 The diagram shown is a flowchart of the position matrix calculation algorithm in this embodiment.

[0090] Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L. The position matrix of sequence S can be represented as: MOP=[x ij ] N*N The algorithm flow is as follows:

[0091] Input sequence: Sequence S = S1, S2, ..., S L ;

[0092] Output sequence matrix: position matrix MOP = [x ij ] N*N .

[0093] Step 1 For i=1; i<=L; i++;

[0094] Step 2: For j = i + 1; j <= L; i++;

[0095] Step 3: MOP[xij]←i+j;

[0096] Step 4: If i ≠ j;

[0097] Step 5: MOP[xji]←i+j

[0098] Step 5. If i≥1, go to step 2; otherwise, end.

[0099] S3 simultaneously selects the values ​​on the diagonal of the sequence matrix and those above the diagonal to encode the amino acid sequence data, and adds the reciprocal of the one-dimensional sequence length as a component of the sequence matrix encoding to distinguish the amino acid sequence length, thereby encoding the amino acid sequence data into a feature vector.

[0100] Furthermore, step S3 includes encoding the amino acid sequence data into a vector as follows:

[0101] MSP = [MOS] 11 ,...,MOS 1N ,...MOS NN ,1 / L]+[MOP 11 ,...,MOP 1N ...MOP NN ,1 / L].

[0102] Specifically, in this embodiment, after calculating the position matrix and sequence matrix, since the frequencies of the elements on the diagonal and off-diagonal of the sequence matrix correspond one-to-one, and the product of two elements on the diagonal of the sequence matrix is ​​equal to the sum of the symmetrical elements above and below the diagonal, we simultaneously select the values ​​on both the diagonal and above the diagonal of the sequence matrix to encode the amino acid sequence data. Adding the reciprocal of the one-dimensional sequence length as a component of the sequence matrix encoding distinguishes the length of the amino acid sequence. Finally, the amino acid sequence data is encoded into a 58-dimensional vector, the specific form of which is shown in the following formula:

[0103] MSP = [MOS] 11 ,...,MOS 17 ,...MOS 77 ,1 / L]+[MOP 11 ,...,MOP 17 ...MOP 77 ,1 / L].

[0104] To improve the convergence speed and accuracy of the model, we adjust m using equations 6 and 7. ij Normalization is performed, and the normalized p ij As the final matrix components.

[0105]

[0106]

[0107] Where L is the length of the protein sequence.

[0108] Furthermore, such as Figure 3 The diagram shows the overall execution flowchart of the positional sequence matrix encoding method in this embodiment. After starting, amino acids are classified, and sequence matrices and position matrices are constructed for each. Encoding is performed based on the positional sequence matrices, and after normalization as described above, the structure is output and the process ends. Amino acids are divided into 7 categories according to their side chain dipole and volume. The amino acid sequences are replaced with the corresponding categories in the amino acid classification table. The classified protein sequences are used as input to the positional sequence matrix encoding method, and the protein-protein interaction feature vector is output. Elements in the sequence are read from back to front, the sequence vector is calculated, and a positional sequence matrix is ​​constructed until the leftmost sequence element is read. Finally, the sequence matrix is ​​normalized.

[0109] Table 2 shows the encoding performance of five encoding methods on the human protein dataset. The number of mis-encoded entries reflects the mis-encoding performance of the five encoding methods on the human protein dataset. Encoding time reflects the average encoding time of the four encoding methods on the human protein dataset. In terms of the number of mis-encoded entries, the CT and LD methods have 40 mis-encoded entries, while AC, MOS, and MSP have no mis-encoded entries. This is mainly because the CT and LD methods contain fewer adjacent amino acid effects. In terms of encoding time, AC has the longest encoding time, and MOS has the shortest. A comprehensive comparison of the above encoding methods shows that the MSP encoding method has a slightly longer encoding time than MOS, but compared with CT, AC, and LD, MSP significantly reduces the vector dimension while reducing encoding time, making it suitable for applications with high encoding time requirements, such as real-time encoding.

[0110] Table 2 shows the coding results on the human dataset.

[0111]

[0112] Furthermore, a second aspect of the present invention provides a method for predicting protein interactions based on the protein-coding method described above using a positional sequence matrix, the method comprising:

[0113] The encoded protein sequence is used as input to the first deep neural network model, and the protein interaction classification result is output based on the first deep neural network model.

[0114] The first deep neural network model includes the following training steps:

[0115] In the forward propagation phase, samples are taken from the human protein dataset and used as input to the deep neural network. After successive transformations, the samples are transmitted to the output layer to calculate the actual output of layer l. The calculation formula is as follows:

[0116]

[0117] With dropout applied, the output of layer l is:

[0118]

[0119] Where l = 1, ..., N, a (l-1) For the input data of layer l, w l b is the connection weight matrix between layer (l-1) and layer l. l δ represents the bias of the l-th layer, and δ represents the activation of the l-th layer.

[0120] During the backpropagation phase, the actual output O is calculated. i and ideal output Y iThe difference is then calculated, and the weight matrix is ​​adjusted by backpropagation using the method of minimizing the error. The error calculation formula for the output layer is as follows:

[0121]

[0122]

[0123] Among them, E i O ik T ik These are the error of the i-th sample, the output of the k-th neuron in the output layer of the i-th sample, and the expected output of the k-th neuron in the output layer of the i-th sample, respectively.

[0124] To study the performance of the protein sequence encoding method MSP, five encoding methods (CT, AC, LD, MOS, and MSP) were combined with DNN to construct five different models: DNN-CT, DNN-AC, DNN-LD, DNN-MOS, and DNN-MSP. All datasets used were human datasets. The activation function for all five models is ReLU, the optimization algorithm is Adam, the batch size is 128, the number of iterations is 30, and each iteration has 10,000 steps. Among them, the learning rate of DNN-MSP is 0.05, the network width is 1024, and the network depth is 1024-1024-1024-1024; the learning rates of DNN-CT, DNN-AC, DNN-LD and DNN-MOSDNN-CT, DNN-AC and DNN-LD are 0.001, 0.001, 0.0001 and 0.01 respectively, the network width is 256, and the network depth is 256-256-256, 256-256-256, 256-128-64-32 and 512-512-512 respectively.

[0125] Table 3 shows a comparison of the prediction performance of different prediction models on the human dataset in this embodiment. As can be seen from Table 3, the accuracy of DNN-MSP is higher than that of DNN-CT (95.12%), DNN-AC (95.17%), DNN-LD (95.60%), and DNN-MOS (94.34%). This indicates that the DNN-MSP protein interaction prediction model has better predictive performance and can improve the prediction performance of protein-protein interactions. The superior predictive performance of the DNN-MSP model is likely because MSP calculates not only relative frequencies but also positional features, i.e., the positional information between amino acids in the same group and the sequence of amino acids in different groups, retaining more sequence information. Therefore, obtaining better predictions is reasonable.

[0126] Table 3 Predictive performance of different prediction models on human datasets

[0127]

[0128] In addition, such as Figure 4 As shown, a third aspect of the present invention provides a protein-coding device based on a positional sequence matrix, the device comprising:

[0129] Classification module 10 classifies the amino acid sequences according to the dipole and volume of the side chains of the amino acids;

[0130] Component module 20 constructs a sequence matrix and a position matrix; wherein, each element in the sequence matrix is ​​used to indicate the frequency of all pairwise combinations of amino acids in the protein sequence; the position matrix is ​​used to indicate the position information of any two groups of amino acids in the protein sequence;

[0131] The encoding module 30 simultaneously selects the values ​​on the diagonal of the sequence matrix and those on the diagonal to encode the amino acid sequence data, and adds the reciprocal of the one-dimensional sequence length as a component of the sequence matrix encoding to distinguish the amino acid sequence length, thereby encoding the amino acid sequence data into a feature vector.

[0132] In addition, such as Figure 5 As shown, a fourth aspect of the present invention provides an electronic device comprising: one or more processors, and a memory for storing one or more computer programs; the computer programs being configured to be executed by the one or more processors, the programs including methods for performing a protein encoding method based on a position sequence matrix as described above, or for performing a method for predicting protein interactions as described above.

[0133] Furthermore, a fifth aspect of the present invention provides a storage medium storing a computer program; the program is loaded and executed by a processor to implement the protein encoding method based on position sequence matrix as described above, or the protein interaction prediction method as described above.

[0134] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0135] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or units, or may be electrical, mechanical or other forms of connection.

[0136] The units described as separate components may or may not be physically separate. As will be appreciated by those skilled in the art, the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0137] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0138] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or grid device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0139] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A protein coding method based on positional sequence matrices, characterized in that, The method includes: S1. The amino acid sequences are classified according to the dipole and volume of their side chains. S2, construct the sequence matrix and the position matrix; where each element in the sequence matrix is ​​used to indicate the frequency of all pairwise combinations of amino acids in the protein sequence; the position matrix is ​​used to indicate the position information of any two groups of amino acids in the protein sequence; S3 simultaneously selects the values ​​on the diagonal of the sequence matrix and those above the diagonal to encode the amino acid sequence data, and adds the reciprocal of the one-dimensional sequence length as a component of the sequence matrix encoding to distinguish the amino acid sequence length, thereby encoding the amino acid sequence data into a feature vector. The construction of the sequence matrix and position matrix, wherein the construction of the sequence matrix MOS includes: Define a nonempty finite set: Ω = {w1, ..., w} N }, where N is the number of categories of the sequence; Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L; the sequence matrix of sequence S is represented as: MOS=[m ij ] N×N (1) In the above formula, m ij : Based on the definition of a sequence matrix, the sum of all elements in the sequence matrix equals... When i≠j, Therefore, for any two sequences, when the sequence lengths are different or the sequence lengths are the same but the number of at least one element is different, their corresponding sequence matrices are different. The construction of the position matrix includes: Define a nonempty finite set: Ω = {w1, ..., w} N }, where N is the number of categories of the sequence; Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L; the position matrix of sequence S is represented as: MOP=[x ij ] N*N (3) In the above formula, x ij : The position matrix is ​​a symmetric matrix, and the elements on the diagonal and above the diagonal contain the position information of all elements in the amino acid sequence. The amino acid sequence data is encoded into the following vector: MSP=[MOS 11 ,...,MOS 1N ,...MOS NN ,1 / L]+[MOP 11 ,...,MOP 1N ,...MOP NN ,1 / L]。 2. A protein-coding device based on a positional sequence matrix, characterized in that, The device includes: The classification module classifies the amino acid sequences based on the dipole and volume of the side chains of the amino acids. The component module constructs a sequence matrix and a position matrix; where each element in the sequence matrix indicates the frequency of all pairwise combinations of amino acids in the protein sequence; the position matrix indicates the positional information of any two pairs of amino acids in the protein sequence. The encoding module simultaneously selects the values ​​on the diagonal of the sequence matrix and those above the diagonal to encode the amino acid sequence data, and adds the reciprocal of the one-dimensional sequence length as a component of the sequence matrix encoding to distinguish the amino acid sequence length, thereby encoding the amino acid sequence data into a feature vector. Constructing the sequence matrix and position matrix, wherein constructing the sequence matrix MOS includes: Define a nonempty finite set: Ω = {w1, ..., w} N }, where N is the number of categories of the sequence; Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L; the sequence matrix of sequence S is represented as: MOS=[m ij ] N×N (1) In the above formula, m ij : Based on the definition of a sequence matrix, the sum of all elements in the sequence matrix equals... When i≠j Therefore, for any two sequences, when the sequence lengths are different or the sequence lengths are the same but the number of at least one element is different, their corresponding sequence matrices are different. The construction of the position matrix includes: Define a nonempty finite set: Ω = {w1, ..., w} N }, where N is the number of categories of the sequence; Given a sequence: S = S1, S2, ..., S L Where L represents the length of sequence S, S i ∈Ω, 1≤i≤L; the position matrix of sequence S is represented as: MOP=[x ij ] N*N (3) In the above formula, x ij : The position matrix is ​​a symmetric matrix, and the elements on the diagonal and above the diagonal contain the position information of all elements in the amino acid sequence. The amino acid sequence data is encoded into the following vector: MSP=[MOS 11 ,...,MOS 1N ,...MOS NN ,1 / L]+[MOP 11 ,...,MOP 1N ,...MOP NN ,1 / L]。 3. An electronic device, the electronic device comprising: One or more processors and a memory, the memory being used to store one or more computer programs; characterized in that the computer programs are configured to be executed by the one or more processors, the programs including methods for performing the protein coding method based on position sequence matrices as described in claim 1.

4. A storage medium storing a computer program; the program being loaded and executed by a processor to implement the protein encoding method based on a position sequence matrix as described in claim 1.

Images