Protein function identification method and device, computer equipment and storage medium

Abstract

The invention relates to a protein function identification method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring a structural feature sequence of a to-be-detected protein; determining structure arrangement information corresponding to each substructure in the structure feature sequence based on a relative arrangement relationship of each substructure in the structure feature sequence, and generating a structure feature vector corresponding to each substructure based on the structure arrangement information; inputting to-be-detected input data determined based on the structure feature vector of each substructure into a protein function recognition model, and outputting a protein function feature vector; and according to the arrangement information of each element in the protein function feature vector, determining a function attribute corresponding to the to-be-detected protein. By adopting the method, the protein function recognition efficiency can be effectively improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a protein function identification method, apparatus, computer equipment and storage medium. Background Art

[0002] In recent years, the academic field's demand for understanding the earth's life and the market's demand for promoting the development of the biomedical industry have become increasingly high. Therefore, it is extremely important to have an in-depth understanding and accurate understanding of the characteristics of proteins. Whether the functional identification of proteins can be achieved efficiently and accurately is of great value to promoting the development of the biomedical industry.

[0003] In traditional technology, the functional identification of proteins is completed through biological experimental verification, which is inefficient. Summary of the invention

[0004] Based on this, it is necessary to provide a protein function identification method, device, computer equipment and storage medium to address the above technical problems, which can effectively improve the efficiency of protein function identification.

[0005] In a first aspect, the present application provides a method for identifying protein function, comprising:

[0006] Obtaining the structural characteristic sequence of the protein to be tested;

[0007] Determine the structural arrangement information corresponding to each substructure in the structural feature sequence based on the relative arrangement relationship of each substructure in the structural feature sequence, and generate the structural feature vector corresponding to each substructure based on the structural arrangement information;

[0008] Inputting the input data to be tested determined based on the structural feature vectors of each substructure into the protein function recognition model, and outputting the protein function feature vector;

[0009] According to the arrangement information of each element in the protein function feature vector, the functional attributes corresponding to the protein to be tested are determined.

[0010] In one embodiment, obtaining the structural characteristic sequence of the protein to be tested includes:

[0011] Obtaining the protein amino acid sequence, protein secondary structure sequence and fusion weight of the protein to be tested;

[0012] The protein amino acid sequence and the protein secondary structure sequence are fused based on the fusion weight to obtain fusion input data;

[0013] The fusion input data is taken as a sequence of structural features.

[0014] In one embodiment, the structural feature sequence is a secondary structure sequence of a protein to be tested, and the structural arrangement information corresponding to each substructure in the structural feature sequence is determined based on the relative arrangement relationship of each substructure in the structural feature sequence, and a corresponding structural feature vector is generated based on the structural arrangement information, including:

[0015] Classifying each substructure in the secondary structure sequence to obtain at least three classification sets corresponding to preset structure categories;

[0016] Obtain at least three corresponding target substructures based on the classification set;

[0017] The structural arrangement information corresponding to each substructure in the secondary structure sequence is determined based on the relative arrangement relationship of each target substructure, and the corresponding structural feature vector is generated based on the structural arrangement information.

[0018] In a second aspect, the present application also provides a method for constructing a protein function identification model, comprising:

[0019] Obtain protein secondary structure sequences and corresponding protein function sets;

[0020] Determine the structural arrangement information corresponding to each substructure in the protein secondary structure sequence based on the relative arrangement relationship of each substructure in the protein secondary structure sequence, and generate a corresponding structural feature vector based on the structural arrangement information;

[0021] Based on the relative arrangement relationship of each sub-function in the protein function set, a functional feature vector corresponding to the protein function set is generated;

[0022] Based on the training sample data determined by the structural feature vectors and functional feature vectors of each substructure, the initial protein function recognition model is trained to obtain a target protein function recognition model, which is used to recognize the corresponding protein function according to the input protein data.

[0023] In one embodiment, obtaining a protein secondary structure sequence and a corresponding protein function set includes:

[0024] Obtaining a first correspondence, wherein the first correspondence is used to characterize the correspondence between the protein amino acid sequence and the corresponding protein function set;

[0025] Obtaining a second corresponding relationship, wherein the second corresponding relationship is used to characterize the corresponding relationship between the protein amino acid sequence and the corresponding protein secondary structure sequence;

[0026] Based on the first correspondence and the second correspondence, the protein secondary structure sequence and the corresponding protein function set are determined.

[0027] In one embodiment, based on the relative arrangement relationship of each sub-function in the protein function set, a function feature vector corresponding to the protein function set is generated, including:

[0028] When the number of each sub-function in the protein function set is greater than a preset threshold, determining a target sub-function from each sub-function according to the preset threshold;

[0029] Based on the relative arrangement relationship of the target sub-functions, a functional feature vector corresponding to the protein function set is generated.

[0030] In a third aspect, the present application also provides a protein function identification device, comprising:

[0031] An acquisition module is used to obtain the secondary structure sequence corresponding to the protein to be tested;

[0032] A data processing module, used to determine the structural arrangement information corresponding to each substructure in the secondary structure sequence based on the relative arrangement relationship of each substructure in the secondary structure sequence, and generate a structural feature vector corresponding to each substructure based on the structural arrangement information;

[0033] The recognition module is used to input the input data to be tested determined based on the structural feature vectors of each substructure into the protein function recognition model and output the protein function feature vector; according to the arrangement information of each element in the protein function feature vector, the functional attributes corresponding to the protein to be tested are determined.

[0034] In a fourth aspect, the present application also provides a protein function recognition model construction device, comprising:

[0035] An acquisition module is used to obtain a protein secondary structure sequence and a corresponding protein function set;

[0036] A conversion module is used to determine the structural arrangement information corresponding to each substructure in the protein secondary structure sequence based on the relative arrangement relationship of each substructure in the protein secondary structure sequence, and generate a corresponding structural feature vector based on the structural arrangement information; based on the relative arrangement relationship corresponding to each subfunction in the protein function set, generate a functional feature vector corresponding to the protein function set;

[0037] The training module is used to train the initial protein function recognition model based on the training sample data determined by the structural feature vectors and functional feature vectors of each substructure to obtain a target protein function recognition model, which is used to identify the corresponding protein function according to the input protein data.

[0038] In a fifth aspect, the present application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the above-mentioned protein function identification method or protein function identification model construction method.

[0039] In a sixth aspect, the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned protein function identification method or protein function identification model construction method.

[0040] The above-mentioned protein function identification method, device, computer equipment and storage medium obtain the structural feature sequence of the protein to be tested, and determine the structural arrangement information corresponding to each substructure based on the arrangement relationship of each substructure in the structural feature sequence, and generate the structural feature vector corresponding to each substructure according to the structural arrangement information, and then generate the input data to be tested based on the structural feature vector of each substructure, and input it into the protein function identification model for identification, and then determine the functional attributes corresponding to the protein to be tested. In this way, the spatial structural feature information of the protein to be tested can be effectively characterized by the feature vector constructed based on the arrangement information of each substructure in the structural feature sequence of the protein to be tested, and the protein function can be identified based on its spatial structural feature information, thereby realizing rapid and effective identification of protein functions, and effectively improving the efficiency of protein function identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A diagram showing an application environment of a protein function identification method in one embodiment;

[0042] Figure 2 A schematic diagram of a method for identifying protein function in one embodiment;

[0043] Figure 3 A schematic diagram of a process for constructing fused input data in one embodiment;

[0044] Figure 4 A schematic diagram of a process for generating a structural feature vector in one embodiment;

[0045] Figure 5 A schematic diagram of a method for constructing a protein function identification model in one embodiment;

[0046] Figure 6 A schematic diagram of a process for determining a protein secondary structure sequence and a corresponding protein function set in one embodiment;

[0047] Figure 7 A schematic diagram of a process for generating a functional feature vector in one embodiment;

[0048] Figure 8A schematic diagram of predicting protein function based on protein structure in one embodiment;

[0049] Fig. 9 A curve diagram of protein function prediction results in one embodiment;

[0050] Fig.10 A curve diagram of the effect of eukaryotic protein function recognition in one embodiment;

[0051] Fig.11 is a structural block diagram of a protein function identification device in one embodiment;

[0052] Fig.12 A structural block diagram of a device for constructing a protein function recognition model in one embodiment;

[0053] Fig.13 is an internal structure diagram of a computer device in one embodiment;

[0054] Fig.14 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0056] The protein function identification method provided in the embodiments of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The server 104 obtains the structural feature sequence of the protein to be tested sent by the terminal 102; the server 104 determines the structural arrangement information corresponding to each substructure in the structural feature sequence based on the relative arrangement relationship of each substructure in the structural feature sequence, and generates a structural feature vector corresponding to each substructure based on the structural arrangement information; the input data to be tested determined based on the structural feature vector of each substructure is input into the protein function recognition model, and the protein function feature vector is output; according to the arrangement information of each element in the protein function feature vector, the functional attribute corresponding to the protein to be tested is determined. Among them, the terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, Internet of Things devices and portable wearable devices, and the Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented with an independent server or a server cluster consisting of multiple servers.

[0057] In one embodiment, Figure 2 As shown, a protein function identification method is provided, and the method is applied to Figure 1 The computer device 104 in the example is used as an example to illustrate, and the following steps are included:

[0058] Step S202, obtaining the structural characteristic sequence of the protein to be tested.

[0059] Among them, the structural characteristic sequence of the protein to be tested includes but is not limited to one or more of the primary structure information, secondary structure information, tertiary structure and quaternary structure of the protein to be tested. The structural characteristic sequence is used to characterize the spatial structural characteristics of the protein, wherein the primary structure information of the protein is the amino acid sequence information, the secondary structure information is the α-helix-β-folding structure information, the tertiary structure information is the information that the protein polypeptide chains are mutually configured on the basis of the secondary structure to form a specific structure, and the quaternary structure information refers to the information of the polymer structure formed by several polypeptide chains with independent tertiary structures connected to each other by non-covalent bonds.

[0060] Specifically, the computer device obtains the structural feature sequence of the protein to be tested from a database or other computer devices.

[0061] Step S204, determining the structural arrangement information corresponding to each substructure in the structural feature sequence based on the relative arrangement relationship of each substructure in the structural feature sequence, and generating a structural feature vector corresponding to each substructure based on the structural arrangement information.

[0062] Among them, the structural feature sequence contains at least two substructure sequences, and each substructure is distributed in the form of a one-dimensional sequence. Taking the secondary structure of protein as an example, the secondary structure information can be represented by the sequence CCBECHHHHH, where the secondary structure of protein can be divided into 8 different substructure types, usually represented by 8 letters ['C', 'B', 'E', 'G', 'I', 'H', 'S', 'T']; the structural arrangement information is used to characterize the relative position relationship between each substructure in the structural feature sequence. Since the structural arrangement information is in the form of a one-dimensional sequence distribution, the structural arrangement information is information that characterizes the front-to-back arrangement relationship between each substructure. For example, in the secondary structure sequence CCBECHHHHH, the third substructure type is a "B" structure, and its structural arrangement information is the position of position number 3. Therefore, a structural feature vector corresponding to the substructure "B" is generated. The structural feature vector can characterize the corresponding position information of the substructure "B" structure in the entire secondary structure sequence and the type of the substructure.

[0063] Specifically, the computer device traverses each substructure in the structural feature sequence obtained in the above step, obtains the substructure type corresponding to each substructure and the position arrangement information (structural arrangement information) corresponding to the substructure in the structural feature sequence through the traversal, and then generates the structural feature vector corresponding to each substructure according to the position arrangement information.

[0064] Step S206, inputting the input data to be tested determined based on the structural feature vectors of each substructure into the protein function recognition model, and outputting the protein function feature vector.

[0065] Specifically, the computer device fuses the structural feature vectors of each substructure obtained in the above steps to obtain the input data to be tested. The fusion method includes but is not limited to sorting and combining the structural feature vectors of each substructure according to the arrangement order of each substructure in the structural feature sequence, and then obtaining a structural feature matrix. The number of columns of the structural feature matrix is ​​determined according to the type of substructure category, and the number of rows of the structural feature matrix is ​​determined according to the number of substructures in the structural feature sequence; the structural feature sequence is then used as the input data to be tested, input into the protein function recognition model, and the protein function feature vector is output.

[0066] Step S208, determining the functional attribute corresponding to the protein to be tested according to the arrangement information of each element in the protein functional characteristic vector.

[0067] Among them, the protein function feature vector is used to characterize the functional attribute information of the corresponding protein.

[0068] Specifically, the computer device obtains the protein function feature vector according to the above steps, and then traverses each vector element in the protein function feature vector, and obtains the functional attribute information corresponding to the protein function feature vector according to the traversal result. For example, when the protein function feature vector is [1,1,0,1,0], it means that the protein functional attributes represented by the first flag bit, the second flag bit and the fourth flag bit of the protein function feature vector are the functional attributes of the corresponding protein, wherein the functional attribute information includes but is not limited to GO (Gene Ontology) functional annotations, molecular functions of corresponding protein molecules, biological processes and cell classification information.

[0069] In this embodiment, the structural characteristic sequence of the protein to be tested is obtained, and based on the arrangement relationship of each substructure in the structural characteristic sequence, the structural arrangement information corresponding to each substructure is determined, and the structural characteristic vector corresponding to each substructure is generated according to the structural arrangement information, and then the input data to be tested is generated based on the structural characteristic vector of each substructure, and input into the protein function recognition model for recognition, thereby determining the functional attributes corresponding to the protein to be tested. In this way, the spatial structural characteristic information of the protein to be tested can be effectively characterized by the characteristic vector constructed based on the arrangement information of each substructure in the structural characteristic sequence of the protein to be tested, and the protein function is recognized based on its spatial structural characteristic information, thereby realizing rapid and effective recognition of protein function, and effectively improving the efficiency of protein function recognition.

[0070] In one embodiment, Figure 3 As shown, the structural characteristic sequence of the protein to be tested is obtained, including:

[0071] Step S302, obtaining the protein amino acid sequence, protein secondary structure sequence and fusion weight of the protein to be tested.

[0072] Among them, the protein amino acid sequence is also called the primary structure of the protein. The protein amino acid sequence is used to characterize the types and arrangement information of amino acids in the polypeptide chain that makes up the protein. The protein secondary structure sequence includes the spatial structural characteristics of the protein polypeptide chain, such as α-helix-β-folding and other spatial structural information.

[0073] Specifically, the computer device reads the protein amino acid sequence, protein secondary structure sequence and fusion weight of the protein to be tested from a database or receives them from other computer devices.

[0074] Step S304: fuse the protein amino acid sequence and the protein secondary structure sequence based on the fusion weight to obtain fusion input data.

[0075] Among them, the fusion weight is used to characterize the proportion of each of the two data in the fused input data after the fusion of the protein amino acid sequence and the protein secondary structure sequence. The method for determining the fusion weight includes but is not limited to determining it according to the type and complexity of the protein, and it can also be obtained by iterative optimization through optimization algorithms such as particle swarm algorithm and genetic algorithm, or it can be selected based on the experience of technicians, and there is no specific limitation here.

[0076] Specifically, the computer device combines the protein amino acid sequence and the protein secondary structure sequence according to the data fusion ratio represented by the fusion weight to obtain fusion input data.

[0077] Step S306: taking the fused input data as a structural feature sequence.

[0078] In this embodiment, the protein amino acid sequence and the protein secondary structure sequence of the protein to be tested are fused according to a certain ratio to obtain fused input data, and the fused input data is then used as structural feature data, thereby realizing the construction of the fused input data corresponding to the protein function identification, so that the constructed fused input data can contain more spatial structural feature information of the protein to be tested, and effectively improve the accuracy of the structural feature sequence in effectively representing the characteristics of the protein to be tested.

[0079] In one embodiment, Figure 4 As shown, the structural feature sequence is the secondary structure sequence of the protein to be tested, and the structural arrangement information corresponding to each substructure in the structural feature sequence is determined based on the relative arrangement relationship of each substructure in the structural feature sequence, and the corresponding structural feature vector is generated based on the structural arrangement information, including:

[0080] Step S402: classify each substructure in the secondary structure sequence to obtain at least three classification sets corresponding to preset structure categories.

[0081] Among them, the substructures in the secondary structure sequence are divided into 8 categories, namely ['C', 'B', 'E', 'G', 'I', 'H', 'S', 'T'], where each different category of substructure is identified by a different letter.

[0082] Specifically, the computer device divides 8 different substructure types into 3 types, and the specific division methods include but are not limited to: the substructures of class 'C', class 'S' and class 'T' are uniformly divided into class 'C'; the substructures of class 'E' and class 'B' are uniformly divided into class 'E'; class 'H', class 'G' and class 'I' are uniformly divided into class 'H', that is, three classification sets are obtained, namely C = {C, S, T}, E = {E, B}, H = {H, G, I}.

[0083] Step S404: obtaining corresponding at least three target substructures based on the classification set.

[0084] Specifically, the computer device uses the substructure types of the three classification sets determined in the above manner as target substructures, namely, Class C, Class E, and Class H.

[0085] Step S406, determining the structural arrangement information corresponding to each substructure in the secondary structure sequence based on the relative arrangement relationship of each target substructure, and generating a corresponding structural feature vector based on the structural arrangement information.

[0086] In this embodiment, each substructure in the secondary structure sequence is classified to obtain at least three classification sets corresponding to preset structure categories, and at least three corresponding target substructures are obtained based on the classification sets. The structural arrangement information corresponding to each substructure in the secondary structure sequence is determined based on the relative arrangement relationship of each target substructure, and a corresponding structural feature vector is generated based on the structural arrangement information. In this way, multiple substructure categories are streamlined into a preset number of target substructure categories, which effectively reduces redundant data, simplifies the data complexity of the protein secondary structure, and improves data purity, so that the generated structural feature vector reduces the interference of redundant data, and effectively improves the accuracy of the structural feature sequence in effectively representing the characteristics of the protein to be tested.

[0087] In one embodiment, Figure 5 As shown, a method for constructing a protein function identification model is provided, and the method is applied to Figure 1 The computer device 104 in the example is used as an example to illustrate, and the following steps are included:

[0088] Step S502, obtaining a protein secondary structure sequence and a corresponding protein function set.

[0089] Among them, the protein function set contains at least one protein function attribute information, and the protein function attribute information is used to characterize the biological function and attributes of the corresponding protein. The protein function attribute information includes but is not limited to GO (Gene Ontology) functional annotation. GO functional annotation includes three subcategories, namely: molecular function (MF), biological process (BP) and cell classification (CC). Usually, GO plus a 7-digit number represents a specific functional term.

[0090] Specifically, the computer device reads the protein secondary structure sequence and the corresponding protein function set from the database, wherein one protein secondary structure sequence can correspond to one protein function attribute information or can correspond to multiple protein attribute information.

[0091] Step S504, determining the structural arrangement information corresponding to each substructure in the protein secondary structure sequence based on the relative arrangement relationship of each substructure in the protein secondary structure sequence, and generating a corresponding structural feature vector based on the structural arrangement information.

[0092] Among them, the protein secondary structure sequence contains at least two substructure sequences, and each substructure is distributed in the form of a one-dimensional sequence. For example, the protein secondary structure information can be represented by the sequence CCBECHHHHH, where the protein secondary structure can be divided into 8 different substructure types, usually represented by 8 letters ['C', 'B', 'E', 'G', 'I', 'H', 'S', 'T']; the structural arrangement information is used to characterize the relative position relationship between each substructure in the structural feature sequence. Since the structural arrangement information is in the form of a one-dimensional sequence distribution, the structural arrangement information is information that characterizes the front-to-back arrangement relationship between each substructure. For example, in the protein secondary structure sequence CCBECHHHHH, the third substructure type is a "B" structure, and its structural arrangement information is the position of position number 3. Therefore, a structural feature vector corresponding to the substructure "B" is generated. The structural feature vector can characterize the corresponding position information of the substructure "B" structure in the entire protein secondary structure sequence and the type of the substructure.

[0093] Step S506: generating a functional feature vector corresponding to the protein function set based on the relative arrangement relationship between the sub-functions in the protein function set.

[0094] Specifically, the computer device traverses each sub-function in the protein function set, obtains the sub-function type corresponding to each sub-function in the set and the relative position relationship (relative arrangement relationship) corresponding to each sub-function according to the traversal result, and then generates a functional feature vector corresponding to the protein function set according to the relative arrangement relationship. For example, when the protein function set is C = {GO1, GO2, GO4}, it means that the protein corresponding to the set has three functional attributes, namely GO1, GO2, and GO4, then the corresponding flag position in the generated functional feature vector is 1, that is, the generated functional feature vector c = [1,1,0,1,0], wherein the dimension of the functional vector can be flexibly set as needed.

[0095] Step S508 , based on the training sample data determined by the structural feature vectors and functional feature vectors of each substructure, the initial protein function recognition model is trained to obtain a target protein function recognition model.

[0096] The initial protein function recognition model includes but is not limited to a convolutional neural network, and the target protein function recognition model is used to recognize the corresponding protein function according to the input protein data.

[0097] Specifically, the computer device uses the structural feature vectors of each substructure as input sample data for model training, and uses the functional feature vectors as input sample data or standard output sample data for model training. The input sample data and the output sample data together constitute the training sample data, and the initial protein function recognition model is trained to obtain the target protein function recognition model.

[0098] In this embodiment, by obtaining the protein secondary structure sequence and the corresponding protein function set, the structural arrangement information corresponding to each substructure in the protein secondary structure sequence is determined based on the relative arrangement relationship of each substructure in the protein secondary structure sequence, and the corresponding structural feature vector is generated based on the structural arrangement information, and then the functional feature vector corresponding to the protein function set is generated based on the relative arrangement relationship of each subfunction in the protein function set, and finally the initial protein function recognition model is trained based on the training sample data determined by the structural feature vectors of each substructure and the functional feature vector to obtain the target protein function recognition model. In this way, by constructing sample data reflecting the corresponding relationship between the spatial feature information of the protein and the functional attributes of the protein according to the spatial structural feature information of the protein, the protein function recognition model obtained by training can fully learn the spatial structural information of the protein, so that the model constructed by itself is more accurate in identifying the protein function, and can also well mine and process a large number of spatial structural features of the protein, and well avoid the defect that a large number of cumbersome biological experiments must be performed on the protein in the traditional technology to achieve effective identification of the protein function, thereby effectively improving the efficiency of protein function identification.

[0099] In one embodiment, Figure 6 As shown, the protein secondary structure sequence and the corresponding protein function set are obtained, including:

[0100] Step S602: Acquire a first corresponding relationship.

[0101] The first correspondence is used to characterize the correspondence between the protein amino acid sequence and the corresponding protein function set.

[0102] Specifically, the computer device obtains the first corresponding relationship in a manner including but not limited to extracting protein amino acid sequences and GO functional annotation information from the UniProt database.

[0103] Step S604: obtaining a second corresponding relationship.

[0104] The second corresponding relationship is used to characterize the corresponding relationship between the protein amino acid sequence and the corresponding protein secondary structure sequence.

[0105] Specifically, the computer device acquires the second corresponding relationship in a manner including but not limited to extracting the corresponding relationship between the protein amino acid sequence and the protein secondary structure sequence from cullPDB.

[0106] Step S606: determining the protein secondary structure sequence and the corresponding protein function set based on the first corresponding relationship and the second corresponding relationship.

[0107] Specifically, the computer device obtains the protein function set corresponding to each protein secondary structure sequence according to the first corresponding relationship and the second corresponding relationship determined in the aforementioned steps.

[0108] In this embodiment, by obtaining the first correspondence between the protein amino acid sequence and the corresponding protein function set, and the second correspondence between the protein amino acid sequence and the protein secondary structure sequence, the protein secondary structure sequence and the corresponding protein function set are determined, thereby achieving effective acquisition of data on the protein secondary structure sequence and the corresponding protein function set, and improving the efficiency of data acquisition.

[0109] In one embodiment, Figure 7 As shown, based on the relative arrangement relationship of each sub-function in the protein function set, a functional feature vector corresponding to the protein function set is generated, including:

[0110] Step S702: when the number of each sub-function in the protein function set is greater than a preset threshold, a target sub-function is determined from each sub-function according to the preset threshold.

[0111] The preset threshold may be set by technicians based on experience, for example, determined based on the complexity of the protein structure or function.

[0112] Specifically, the computer device compares the number of each sub-function in the protein function set with a preset threshold. When the number of each sub-function in the protein function set is greater than the preset threshold, redundant sub-functions are eliminated from each sub-function and a preset number of target sub-functions are retained.

[0113] Step S704: generating a functional feature vector corresponding to the protein function set based on the relative arrangement relationship of the target sub-functions.

[0114] Specifically, the computer device determines each target sub-function according to the aforementioned steps, and then generates a functional feature vector corresponding to the protein function set according to the relative arrangement order of each target sub-function in the protein function set.

[0115] In this embodiment, when the number of each sub-function in the protein function set is greater than a preset threshold, the target sub-function is determined from each sub-function according to the preset threshold, and based on the relative arrangement relationship of the target sub-function, a functional feature vector corresponding to the protein function set is generated. This allows for the effective elimination of redundant functional attributes when there are too many functions corresponding to certain proteins, and only retains a certain number of main protein functions, thereby effectively improving the reliability of the generated functional feature vector.

[0116] The present application also provides an application scenario, which applies the above-mentioned protein function identification method, and the method is applied to the scenario of identifying protein functional attributes based on protein structural feature information. Specifically, the application of protein function identification in this application scenario is as follows:

[0117] In this embodiment, the protein function identification method comprises the following steps:

[0118] (1) Construction of a database of protein secondary structure and GO functional annotation

[0119] The computer equipment extracts the protein primary sequence (protein amino acid sequence) and GO functional annotation information from the UniProt database, and also extracts the corresponding relationship between the protein primary sequence and secondary structure from the cullPDB database, and then cross-screens the above two databases to find the corresponding relationship between the protein secondary structure and GO functional annotation information.

[0120] Proteins are composed of 20 amino acids, usually represented by 20 letters ['A', 'C', 'E', 'D', 'G', 'F', 'I', 'H', 'K', 'M', 'L', 'N', 'Q', 'P', 'S', 'R', 'T', 'W', 'V', 'Y'], as shown in the left figure below. Protein secondary structure can be divided into 8 different structural types, usually represented by 8 letters ['C', 'B', 'E', 'G', 'I', 'H', 'S', 'T']. GO functional annotation includes three subcategories: molecular function (MF), biological process (BP) and cell classification (CC). A specific functional term is usually represented by GO plus a 7-digit number.

[0121] (2) Constructing a neural network model for deep learning training

[0122] Based on the linear characteristics of protein secondary structure sequences, the Keras framework and the convolutional neural network (CNN) model were used, which included the convolutional layer Conv1D and the pooling layer MaxPooling1D.

[0123] The one-to-many data of protein secondary structure sequences and GO functional annotations are first converted into a one-hot matrix. The length of the protein secondary structure sequences is uniformly extended to 2000, and the empty set is represented by 0. The GO categories are also processed one-hot.

[0124] For example, the sequence of ss1 is: CCBECHHHHH, and its functional classifications are GO1, GO2, and GO4. Its one-hot matrix is ​​shown in Table 1 below:

[0125] Table 1

[0126] ss1 C B E G I H S T 0 1 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 2 0 1 0 0 0 0 0 0 3 0 0 1 0 0 0 0 0 4 1 0 0 0 0 0 0 0 5 0 0 0 0 0 1 0 0 6 0 0 0 0 0 1 0 0 7 0 0 0 0 0 1 0 0 8 0 0 0 0 0 1 0 0 9 0 0 0 0 0 1 0 0

[0127] The corresponding functional feature vectors are shown in Table 2 below:

[0128] Table 2

[0129] GO1 GO2 GO3 GO4 GO5 ss1 1 1 0 1 0

[0130] (3) Select an appropriate database for verification

[0131] In order to verify the effectiveness of the algorithm, two types of organisms with low primary sequence similarity and public GO functional classification data pool were selected, one as a training set and the other as a test set for verification. The specific content is shown in the following example:

[0132] Embodiment 1:

[0133] The intersection database of UniProt and cullPDB was used, with 90% of it as the training set and 10% as the test set for training and evaluation. The evaluation results are shown in Fig. 9 As shown in the figure, all_aa represents the baseline for GO functional classification based on primary sequence. It can be seen that the Precision-Recall curve of the secondary structure represented by all_ss for GO annotation is basically at a similar level to the annotation result of all_aa.

[0134] Embodiment 2:

[0135] The eukaryotic (mouse) proteins were selected as the training set, and the prokaryotic (E. coli) proteins were selected as the test set. The evaluation results are as follows Fig.10 As shown in the figure, it can be seen that the prediction of GO functional annotation by ss secondary structure is significantly better than that by aa primary sequence (the curve corresponding to ss is more inclined to the upper right position). All other comprehensive index scores (Fmax & AUPR) also show that ss is more effective.

[0136] In this embodiment, the secondary structure sequence of the protein to be tested is obtained, and based on the arrangement relationship of each substructure in the secondary structure sequence, the structural arrangement information corresponding to each substructure is determined, and the structural feature vector corresponding to each substructure is generated according to the structural arrangement information, and then the input data to be tested is generated based on the structural feature vector of each substructure, and input into the trained convolutional neural network model for identification, and then the GO functional annotation information corresponding to the protein to be tested is determined. In this way, the spatial structural feature information of the protein to be tested can be effectively characterized by the feature vector constructed based on the arrangement information of each substructure in the secondary structure sequence of the protein to be tested, and the protein GO functional annotation information can be identified based on its spatial structural feature information, thereby realizing rapid and effective identification of protein functions, and effectively improving the efficiency of protein function identification.

[0137] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0138] In one embodiment, Fig.11 As shown, a protein function identification device is provided. The device can use a software module or a hardware module, or a combination of the two to become a part of a computer device. The device specifically includes: an acquisition module 1102, a data processing module 1104, and an identification module 1106, wherein:

[0139] An acquisition module 1102 is used to acquire a secondary structure sequence corresponding to the protein to be tested;

[0140] The data processing module 1104 is used to determine the structural arrangement information corresponding to each substructure in the secondary structure sequence based on the relative arrangement relationship of each substructure in the secondary structure sequence, and generate a structural feature vector corresponding to each substructure based on the structural arrangement information;

[0141] The identification module 1106 is used to input the input data to be tested determined based on the structural feature vectors of each substructure into the protein function identification model, and output the protein function feature vector; according to the arrangement information of each element in the protein function feature vector, the functional attribute corresponding to the protein to be tested is determined.

[0142] In one embodiment, the acquisition module 1102 is also used to obtain the protein amino acid sequence, protein secondary structure sequence and fusion weight of the protein to be tested; fuse the protein amino acid sequence and protein secondary structure sequence based on the fusion weight to obtain fusion input data; and use the fusion input data as the structural feature sequence.

[0143] In one embodiment, the data processing module 1104 is also used to classify each substructure in the secondary structure sequence to obtain at least three classification sets corresponding to preset structure categories; obtain at least three corresponding target substructures based on the classification sets; determine the structural arrangement information corresponding to each substructure in the secondary structure sequence based on the relative arrangement relationship of each target substructure, and generate a corresponding structural feature vector based on the structural arrangement information.

[0144] The above-mentioned protein function identification device obtains the structural feature sequence of the protein to be tested, and determines the structural arrangement information corresponding to each substructure based on the arrangement relationship of each substructure in the structural feature sequence, and generates the structural feature vector corresponding to each substructure according to the structural arrangement information, and then generates the input data to be tested based on the structural feature vector of each substructure, and inputs it into the protein function identification model for identification, thereby determining the functional attributes corresponding to the protein to be tested. In this way, the spatial structural feature information of the protein to be tested can be effectively characterized by the feature vector constructed based on the arrangement information of each substructure in the structural feature sequence of the protein to be tested, and the protein function is identified based on its spatial structural feature information, thereby realizing rapid and effective identification of protein functions, and effectively improving the efficiency of protein function identification.

[0145] In one embodiment, Fig.12 As shown, a protein function recognition model construction device is provided. The device can use a software module or a hardware module, or a combination of the two to become a part of a computer device. The device specifically includes: an acquisition module 1202, a conversion module 1204, and a training module 1206, wherein:

[0146] An acquisition module 1202 is used to acquire a protein secondary structure sequence and a corresponding protein function set;

[0147] The conversion module 1204 is used to determine the structural arrangement information corresponding to each substructure in the protein secondary structure sequence based on the relative arrangement relationship of each substructure in the protein secondary structure sequence, and generate a corresponding structural feature vector based on the structural arrangement information; based on the relative arrangement relationship corresponding to each subfunction in the protein function set, generate a functional feature vector corresponding to the protein function set;

[0148] The training module 1206 is used to train the initial protein function recognition model based on the training sample data determined by the structural feature vectors and functional feature vectors of each substructure to obtain a target protein function recognition model, which is used to identify the corresponding protein function according to the input protein data.

[0149] In one embodiment, the acquisition module 1202 is also used to obtain a first correspondence, wherein the first correspondence is used to characterize the correspondence between the protein amino acid sequence and the corresponding protein function set; obtain a second correspondence, wherein the second correspondence is used to characterize the correspondence between the protein amino acid sequence and the corresponding protein secondary structure sequence; based on the first correspondence and the second correspondence, determine the protein secondary structure sequence and the corresponding protein function set.

[0150] In one embodiment, the conversion module 1204 is also used to determine the target sub-function from each sub-function according to the preset threshold when the number of each sub-function in the protein function set is greater than a preset threshold; and generate a functional feature vector corresponding to the protein function set based on the relative arrangement relationship of the target sub-function.

[0151] The above-mentioned protein function recognition model construction device obtains the protein secondary structure sequence and the corresponding protein function set, determines the structural arrangement information corresponding to each substructure in the protein secondary structure sequence based on the relative arrangement relationship of each substructure in the protein secondary structure sequence, and generates the corresponding structural feature vector based on the structural arrangement information, and then generates the functional feature vector corresponding to the protein function set based on the relative arrangement relationship of each subfunction in the protein function set, and finally trains the initial protein function recognition model based on the training sample data determined by the structural feature vectors of each substructure and the functional feature vector to obtain the target protein function recognition model. In this way, by constructing sample data reflecting the corresponding relationship between protein spatial feature information and protein functional attributes according to the spatial structure feature information of the protein, the trained protein function recognition model can fully learn the protein spatial structure information, so that the model constructed by itself is more accurate in identifying the protein function, and can also well mine and process a large number of protein spatial structure features, and well avoid the defect that a large number of cumbersome biological experiments must be carried out on the protein in the traditional technology to achieve effective identification of the protein function, thereby effectively improving the efficiency of protein function identification.

[0152] The specific definition of the protein function identification device can be found in the definition of the protein function identification method above, which will not be repeated here. Each module in the above-mentioned protein function identification device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0153] The specific definition of the protein function identification model construction device can be found in the definition of the protein function identification model construction method above, which will not be repeated here. Each module in the above-mentioned protein function identification model construction device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0154] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig.13 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store protein secondary structure sequences and corresponding protein function set data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a protein function identification method or a protein function identification model construction method is implemented.

[0155] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Fig.14As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a protein function identification method or a protein function identification model construction method is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device housing, or an external keyboard, touchpad or mouse, etc.

[0156] Those skilled in the art will understand that Fig.13 or Fig.14 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0157] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0158] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0159] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program includes computer instructions, the computer instructions are 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 performs the steps in the above-mentioned method embodiments.

[0160] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0161] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0162] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for identifying protein function, characterized in that: The method comprises: Obtaining the structural characteristic sequence of the protein to be tested; Determine the structural arrangement information corresponding to each substructure in the structural feature sequence based on the relative arrangement relationship of each substructure in the structural feature sequence, and generate the structural feature vector corresponding to each substructure based on the structural arrangement information; Inputting the input data to be tested determined based on the structural feature vectors of each substructure into a protein function recognition model, and outputting a protein function feature vector; According to the arrangement information of each element in the protein function feature vector, the functional attribute corresponding to the protein to be tested is determined.

2. The method according to claim 1, characterized in that The step of obtaining the structural characteristic sequence of the protein to be tested comprises: Obtaining the protein amino acid sequence, protein secondary structure sequence and fusion weight of the protein to be tested; Based on the fusion weight, the protein amino acid sequence and the protein secondary structure sequence are fused to obtain fusion input data; The fused input data is used as the structural feature sequence.

3. The method according to claim 1, characterized in that: The structural feature sequence is a secondary structure sequence of the protein to be tested, and the step of determining the structural arrangement information corresponding to each substructure in the structural feature sequence based on the relative arrangement relationship of each substructure in the structural feature sequence, and generating a corresponding structural feature vector based on the structural arrangement information includes: Classifying each substructure in the secondary structure sequence to obtain at least three classification sets corresponding to preset structure categories; Obtaining corresponding at least three target substructures based on the classification set; The structural arrangement information corresponding to each substructure in the secondary structure sequence is determined based on the relative arrangement relationship of each target substructure, and a corresponding structural feature vector is generated based on the structural arrangement information.

4. A method for constructing a protein function identification model, characterized in that: The method comprises: Obtain protein secondary structure sequences and corresponding protein function sets; Determining the structural arrangement information corresponding to each substructure in the protein secondary structure sequence based on the relative arrangement relationship of each substructure in the protein secondary structure sequence, and generating a corresponding structural feature vector based on the structural arrangement information; Based on the relative arrangement relationship of each sub-function in the protein function set, generating a functional feature vector corresponding to the protein function set; Based on the structural feature vectors of each substructure and the training sample data determined by the functional feature vectors, the initial protein function recognition model is trained to obtain a target protein function recognition model, which is used to recognize the corresponding protein function according to the input protein data.

5. The method according to claim 4, characterized in that The obtaining of the protein secondary structure sequence and the corresponding protein function set includes: Acquire a first correspondence relationship, where the first correspondence relationship is used to characterize a correspondence relationship between a protein amino acid sequence and a corresponding protein function set; Acquire a second corresponding relationship, where the second corresponding relationship is used to characterize the corresponding relationship between the protein amino acid sequence and the corresponding protein secondary structure sequence; Based on the first corresponding relationship and the second corresponding relationship, a protein secondary structure sequence and a corresponding protein function set are determined.

6. The method according to claim 4, characterized in that The generating of the functional feature vector corresponding to the protein function set based on the relative arrangement relationship of each sub-function in the protein function set comprises: When the number of each sub-function in the protein function set is greater than a preset threshold, determining a target sub-function from the sub-functions according to the preset threshold; Based on the relative arrangement relationship of the target sub-functions, a functional feature vector corresponding to the protein function set is generated.

7. A protein function identification device, characterized in that: The device comprises: An acquisition module is used to obtain the secondary structure sequence corresponding to the protein to be tested; A data processing module, used to determine the structural arrangement information corresponding to each substructure in the secondary structure sequence based on the relative arrangement relationship of each substructure in the secondary structure sequence, and generate a structural feature vector corresponding to each substructure based on the structural arrangement information; The identification module is used to input the input data to be tested determined based on the structural feature vectors of each substructure into the protein function identification model and output the protein function feature vector; and determine the functional attributes corresponding to the protein to be tested according to the arrangement information of each element in the protein function feature vector.

8. A protein function identification model construction device, characterized in that: The device comprises: An acquisition module is used to obtain a protein secondary structure sequence and a corresponding protein function set; A conversion module, used to determine the structural arrangement information corresponding to each substructure in the protein secondary structure sequence based on the relative arrangement relationship of each substructure in the protein secondary structure sequence, and generate a corresponding structural feature vector based on the structural arrangement information; based on the relative arrangement relationship corresponding to each subfunction in the protein function set, generate a functional feature vector corresponding to the protein function set; The training module is used to train the initial protein function recognition model based on the structural feature vectors of each substructure and the training sample data determined by the functional feature vectors to obtain a target protein function recognition model, wherein the target protein function recognition model is used to identify the corresponding protein function according to the input protein data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images