Protein functional annotation method and apparatus, electronic device, and storage medium

Abstract

The present disclosure provides a protein functional annotation method and apparatus, an electronic device, and a storage medium. The method comprises: separately acquiring a sequence feature, a pre-trained sequence feature, and a preset annotation feature on the basis of a protein sequence; performing feature extraction on the preset annotation feature to obtain a first learning feature; performing feature extraction on the sequence feature and the pre-trained sequence feature to obtain a second learning feature; fusing the first learning feature and the second learning feature to obtain a first fused feature; and processing the first fused feature to obtain protein functional annotation information. Compared with the related art, the present disclosure relates to acquiring a sequence feature, a pre-trained sequence feature, and a preset annotation feature on the basis of a protein sequence, and fusing the three different aspects of features, such that protein functional annotation information obtained is more accurate.

Description

Protein function annotation method and device, electronic device and storage medium Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a method and device for protein function annotation, an electronic device, and a storage medium. Background Art

[0002] The order of amino acids in a protein determines its spatial structure, which in turn determines its biological function. The goal of protein functional annotation is to use algorithms to predict a protein's potential functions. Currently, protein functional annotation algorithms are primarily categorized into three types: those based solely on sequence information, those based on both sequence and structure information, and those based on both sequence and interaction networks.

[0003] Because protein structural and interaction network information requires complex experiments and is relatively difficult to obtain, while amino acid sequence information is the most readily available, algorithms based solely on sequence information are the most widely used. However, compared to other algorithms that utilize additional information that is difficult to obtain from sequences, protein function annotation algorithms based solely on sequence information have lower prediction accuracy.

[0004] Summary of the Invention

[0005] The present disclosure provides a method and apparatus, electronic device, and storage medium for protein function annotation, the main purpose of which is to solve the problem of low prediction accuracy of protein function annotation algorithms based only on sequence information.

[0006] According to a first aspect of the present disclosure, a method for protein function annotation is provided, comprising:

[0007] According to the protein sequence, sequence features, pre-trained sequence features and preset annotation features are obtained respectively;

[0008] Performing feature extraction processing on the preset annotation feature to obtain a first learning feature;

[0009] Performing feature extraction processing on the sequence feature and the pre-trained sequence feature to obtain a second learning feature;

[0010] Fusing the first learning feature with the second learning feature to obtain a first fused feature;

[0011] The first fusion feature is processed to obtain protein function annotation information.

[0012] Optionally, after respectively obtaining sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence, the method further includes:

[0013] Based on a first preset number of fully connected layers, feature extraction processing is performed on the preset annotation features to obtain a third learning feature;

[0014] Based on a second preset number of convolutional layers, feature extraction processing is performed on the sequence features and the pre-trained sequence features to obtain a fourth learning feature and a fifth learning feature;

[0015] fusing the third learning feature, the fourth learning feature, and the fifth learning feature to obtain a second fused feature; wherein the fusion process is an addition process;

[0016] Based on a third preset number of fully connected layers, the second fusion features are classified and calculated to obtain protein function annotation information.

[0017] Optionally, the obtaining of sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence includes:

[0018] Based on a preset encoding algorithm, encoding the protein sequence to obtain the sequence feature;

[0019] Based on a preset protein language model, pre-training the sequence features to obtain the pre-trained sequence features;

[0020] In a preset protein database, the preset annotation feature corresponding to the protein sequence is searched.

[0021] Optionally, performing feature extraction processing on the preset annotation feature to obtain the first learning feature includes the following steps:

[0022] Inputting the preset annotation features into a first preset number of residual fully connected modules;

[0023] The output features of the first residual fully connected module are used as the input features of the second residual fully connected module for feature extraction;

[0024] The above steps are looped for a first predetermined number of times to obtain the first learning feature, wherein the first predetermined number of times is equal to the first preset number.

[0025] Optionally, the first residual fully connected module inputting the preset annotation feature into the residual fully connected module includes:

[0026] Inputting the preset annotation features into a normalization layer for normalization processing to obtain the normalized preset annotation features;

[0027] Inputting the normalized preset annotation features into a nonlinear layer for nonlinear mapping to obtain the preset annotation features after nonlinear mapping;

[0028] Inputting the preset annotation features after the nonlinear mapping into a fully connected layer for feature extraction to obtain a first preset annotation feature;

[0029] Inputting the first preset annotation feature into the normalization layer, the nonlinear layer, and the fully connected layer in sequence for processing to obtain a second preset annotation feature;

[0030] The preset annotation feature and the second preset annotation feature are added together to obtain a first output feature.

[0031] Optionally, based on a fourth preset number of convolutional layers, the sequence features are processed to obtain a sixth learning feature, and the pre-trained sequence features are processed to obtain a seventh learning feature;

[0032] Adding the sixth learning feature and the seventh learning feature to obtain a third fusion feature;

[0033] The performing feature extraction processing on the sequence feature and the pre-trained sequence feature to obtain the second learning feature comprises the following steps:

[0034] Inputting the third fusion feature into a second preset number of residual convolution modules;

[0035] Using the output features of the first residual convolution module as input features of the second residual convolution module for feature extraction processing;

[0036] The above steps are repeated a second predetermined number of times to obtain the second learning feature, wherein the second predetermined number of times is equal to the second preset number.

[0037] Optionally, the step of inputting the third fusion feature into the first residual convolution module in the residual convolution module includes:

[0038] Inputting the third fused feature into a normalization layer for normalization processing to obtain a normalized third fused feature;

[0039] Inputting the normalized third fusion feature into a nonlinear layer for nonlinear mapping to obtain the nonlinearly mapped third fusion feature;

[0040] Inputting the third fused feature after the nonlinear mapping into the convolution layer for feature learning to obtain a second output feature;

[0041] Inputting the second output feature into the normalization layer, the nonlinear layer, and the convolution layer in sequence for processing to obtain a third output feature;

[0042] The third fusion feature and the third output feature are added together to obtain a fourth output feature.

[0043] Optionally, fusing the first learning feature with the second learning feature to obtain a first fused feature includes:

[0044] The first learning feature and the second learning feature are concatenated to obtain a first fusion feature.

[0045] Optionally, the processing the first fusion feature to obtain protein function annotation information includes:

[0046] Inputting the first fusion feature into an integration module for integration processing to obtain a protein fusion feature;

[0047] Based on a fifth preset number of fully connected layers, the protein fusion features are classified and processed to obtain the protein function annotation information.

[0048] Optionally, the step of inputting the first fusion feature into an integration module for integration processing to obtain a protein fusion feature includes:

[0049] Inputting the first fused feature into a normalization layer for normalization processing to obtain a normalized first fused feature;

[0050] Inputting the normalized first fusion feature into a nonlinear layer for nonlinear mapping to obtain the first fusion feature after nonlinear mapping;

[0051] Inputting the first fusion feature after the nonlinear mapping into a fully connected layer for feature extraction to obtain a first protein fusion feature;

[0052] Inputting the first protein fusion feature into the normalization layer, the nonlinear layer, and the fully connected layer in sequence for processing to obtain a second protein fusion feature;

[0053] The first fusion feature and the second protein fusion feature are added together to obtain the protein fusion feature.

[0054] Optionally, before respectively obtaining sequence features, pre-trained sequence features, and preset annotation features based on the protein sequence, the method further includes:

[0055] The protein function annotation model is trained based on the training protein sequences.

[0056] Optionally, inputting the training protein sequence into the protein function annotation model for training includes:

[0057] According to the training protein sequence, respectively obtaining a training sequence feature, a training first sequence feature, and a training preset annotation feature;

[0058] Performing feature extraction processing on the preset annotation features for training to obtain first learning features for training;

[0059] performing feature extraction processing on the training sequence feature and the first training sequence feature to obtain a second training learning feature;

[0060] Fusing the first training learning feature with the second training learning feature to obtain a first training fused feature;

[0061] The first training fusion feature is processed to obtain training protein function annotation information.

[0062] Optionally, after respectively obtaining the training sequence features, the training first sequence features, and the training preset annotation features according to the training protein sequence, the method further includes:

[0063] Based on the first preset number of training fully connected layers, feature extraction processing is performed on the training preset annotation features to obtain third training learning features;

[0064] Based on the second preset number of training convolutional layers, feature extraction processing is performed on the training sequence features and the training first sequence features to obtain fourth training learning features and fifth training learning features;

[0065] fusing the third training learning feature, the fourth training learning feature, and the fifth training learning feature to obtain a second training fused feature; wherein the fusion process is an addition process;

[0066] Based on a third preset number of fully connected layers for training, classification calculation is performed on the second training fusion features to obtain training protein function annotation information.

[0067] According to a second aspect of the present disclosure, a device for protein function annotation is provided, comprising:

[0068] An acquisition unit, used to acquire sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence;

[0069] A first processing unit is configured to perform feature extraction processing on the preset annotation feature to obtain a first learning feature;

[0070] A second processing unit is used to perform feature extraction processing on the sequence feature and the pre-trained sequence feature to obtain a second learning feature;

[0071] a first fusion unit, configured to fuse the first learning feature with the second learning feature to obtain a first fused feature;

[0072] The third processing unit is used to process the first fusion feature to obtain protein function annotation information.

[0073] Optionally, the device further includes:

[0074] a fourth processing unit, configured to, after respectively obtaining sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence, perform feature extraction processing on the preset annotation features based on a first preset number of fully connected layers to obtain third learning features;

[0075] a fifth processing unit, configured to perform feature extraction processing on the sequence feature and the pre-trained sequence feature based on a second preset number of convolutional layers to obtain a fourth learning feature and a fifth learning feature;

[0076] a second fusion unit, configured to fuse the third learning feature, the fourth learning feature, and the fifth learning feature to obtain a second fused feature; wherein the fusion process is an addition process;

[0077] The sixth processing unit is configured to perform classification calculation on the second fusion features based on a fully connected layer with a third preset number of layers to obtain protein function annotation information.

[0078] Optionally, the acquiring unit includes:

[0079] An encoding module, configured to encode the protein sequence based on a preset encoding algorithm to obtain the sequence feature;

[0080] A pre-training module, configured to perform pre-training processing on the sequence features based on a preset protein language model to obtain the pre-trained sequence features;

[0081] The search module is used to search for the preset annotation feature corresponding to the protein sequence in a preset protein database.

[0082] Optionally, the first processing unit includes:

[0083] A first input module, configured to input the preset annotation features into a first preset number of residual fully connected modules;

[0084] A first processing module is used to perform feature extraction processing on the output features of the first residual fully connected module as input features of the second residual fully connected module;

[0085] The first loop module is configured to loop and execute a first predetermined number of times according to the processing method of the first processing module to obtain the first learning feature, wherein the first predetermined number of times is equal to the first preset number.

[0086] Optionally, the first input module is further used to:

[0087] Inputting the preset annotation features into a normalization layer for normalization processing to obtain the normalized preset annotation features;

[0088] Inputting the normalized preset annotation features into a nonlinear layer for nonlinear mapping to obtain the preset annotation features after nonlinear mapping;

[0089] Inputting the preset annotation features after the nonlinear mapping into a fully connected layer for feature extraction to obtain a first preset annotation feature;

[0090] Inputting the first preset annotation feature into the normalization layer, the nonlinear layer, and the fully connected layer in sequence for processing to obtain a second preset annotation feature;

[0091] The preset annotation feature and the second preset annotation feature are added together to obtain a first output feature.

[0092] Optionally, the second processing unit includes:

[0093] A second processing module is configured to process the sequence features based on a fourth preset number of convolutional layers to obtain a sixth learning feature, and to process the pre-trained sequence features to obtain a seventh learning feature;

[0094] a first adding module, configured to add the sixth learning feature and the seventh learning feature to obtain a third fusion feature;

[0095] A second input module, configured to input the third fusion feature into a second preset number of residual convolution modules;

[0096] A first extraction module, configured to perform feature extraction processing on the output features of the first residual convolution module as input features of the second residual convolution module;

[0097] The second loop module is configured to loop and execute the processing method of the first extraction module a second predetermined number of times to obtain the second learning feature, wherein the second predetermined number of times is equal to the second preset number.

[0098] Optionally, the second input module is further used to:

[0099] Inputting the third fused feature into a normalization layer for normalization processing to obtain a normalized third fused feature;

[0100] Inputting the normalized third fusion feature into a nonlinear layer for nonlinear mapping to obtain the nonlinearly mapped third fusion feature;

[0101] Inputting the third fused feature after the nonlinear mapping into the convolution layer for feature learning to obtain a second output feature;

[0102] Inputting the second output feature into the normalization layer, the nonlinear layer, and the convolution layer in sequence for processing to obtain a third output feature;

[0103] The third fusion feature and the third output feature are added together to obtain a fourth output feature.

[0104] Optionally, the first fusion unit is further configured to:

[0105] The first learning feature and the second learning feature are concatenated to obtain a first fusion feature.

[0106] Optionally, the third processing unit includes:

[0107] An integration module, configured to input the first fusion feature into the integration module for integration processing to obtain a protein fusion feature;

[0108] A classification module is used to classify the protein fusion features based on a fifth preset number of fully connected layers to obtain the protein function annotation information.

[0109] Optionally, the integration module is further configured to:

[0110] Inputting the first fused feature into a normalization layer for normalization processing to obtain a normalized first fused feature;

[0111] Inputting the normalized first fusion feature into a nonlinear layer for nonlinear mapping to obtain the first fusion feature after nonlinear mapping;

[0112] Inputting the first fusion feature after the nonlinear mapping into a fully connected layer for feature extraction to obtain a first protein fusion feature;

[0113] Inputting the first protein fusion feature into the normalization layer, the nonlinear layer, and the fully connected layer in sequence for processing to obtain a second protein fusion feature;

[0114] The first fusion feature and the second protein fusion feature are added together to obtain the protein fusion feature.

[0115] Optionally, the device further includes:

[0116] The training unit is used to train the protein function annotation model based on the training protein sequence before respectively obtaining sequence features, pre-trained sequence features and preset annotation features according to the protein sequence.

[0117] Optionally, the training unit includes:

[0118] A training acquisition module is used to obtain training sequence features, training first sequence features and training preset annotation features according to the training protein sequence;

[0119] a first processing module for training, configured to perform feature extraction processing on the preset annotation features for training to obtain first learning features for training;

[0120] a second training processing module, configured to perform feature extraction processing on the training sequence features and the first training sequence features to obtain second training learning features;

[0121] a first training fusion module, configured to fuse the first training learning feature with the second training learning feature to obtain a first training fusion feature;

[0122] The third processing module for training is used to process the first fusion feature for training to obtain protein function annotation information for training.

[0123] Optionally, the training unit further includes:

[0124] a fourth training processing module, configured to, after respectively obtaining a training sequence feature, a training first sequence feature, and a training preset annotation feature based on a training protein sequence, perform feature extraction processing on the training preset annotation feature based on a first preset number of training fully connected layers to obtain a third training learning feature;

[0125] a fifth training processing module, configured to perform feature extraction processing on the training sequence features and the first training sequence features based on a second preset number of training convolutional layers, to obtain fourth training learning features and fifth training learning features;

[0126] a second training fusion module, configured to fuse the third training learning feature, the fourth training learning feature, and the fifth training learning feature to obtain a second training fused feature; wherein the fusion process is an addition process;

[0127] The sixth processing module for training is used to perform classification calculation on the second training fusion feature based on the third preset number of training fully connected layers to obtain training protein function annotation information.

[0128] According to a third aspect of the present disclosure, there is provided an electronic device, including:

[0129] at least one processor; and

[0130] a memory communicatively connected to the at least one processor; wherein,

[0131] The memory stores instructions that can be executed by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform the method described in the first aspect.

[0132] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described in the first aspect.

[0133] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the method as described in the first aspect above.

[0134] The present disclosure provides a method and apparatus, electronic device, and storage medium for protein function annotation, which respectively obtain sequence features, pre-trained sequence features, and preset annotation features based on the protein sequence; based on a first preset number of residual fully connected modules, perform feature extraction processing on the preset annotation features to obtain a first learning feature; based on a second preset number of residual convolution modules, perform feature extraction processing on the sequence features and the pre-trained sequence features to obtain a second learning feature; fuse the first learning feature with the second learning feature to obtain a first fusion feature; process the first fusion feature to obtain protein function annotation information. Compared with the related art, the present disclosure obtains sequence features, pre-trained sequence features, and preset annotation features based on the protein sequence, and by fusing the features from the above three different angles, the obtained protein function annotation information is more accurate; and the prediction accuracy of protein function annotation is effectively improved.

[0135] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0136] The accompanying drawings are used to better understand the present invention and do not constitute a limitation of the present invention.

[0137] FIG1 is a schematic diagram of a method for protein function annotation provided by an embodiment of the present disclosure;

[0138] FIG2 is a schematic diagram of a process for another method of protein function annotation provided by an embodiment of the present disclosure;

[0139] FIG3 is a schematic diagram of the network architecture of the MergeNet algorithm, a protein function annotation model;

[0140] FIG4 is a schematic diagram of a protein function annotation information prediction result provided by the present disclosure;

[0141] FIG5 is a schematic diagram of a process for another method of protein function annotation provided by an embodiment of the present disclosure;

[0142] FIG6 is a schematic diagram of a protein sequence feature;

[0143] FIG7 is a schematic diagram of the network architecture of a protein function annotation model ResMergeNet algorithm;

[0144] FIG8 is a schematic diagram of a process for training a protein function annotation model according to an embodiment of the present disclosure;

[0145] FIG9 is a schematic diagram of a process for training another protein function annotation model provided by an embodiment of the present disclosure;

[0146] FIG10 is a schematic structural diagram of a device for protein function annotation provided by an embodiment of the present disclosure;

[0147] FIG11 is a schematic structural diagram of another apparatus for protein function annotation provided by an embodiment of the present disclosure;

[0148] FIG12 is a schematic block diagram of an example electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION

[0149] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0150] With the development of protein computational tools, it is now possible to calculate sequence features from different perspectives from protein sequence information. However, current protein function annotation algorithms based solely on sequence information often only use one of these sequence features. The inability to effectively integrate multiple protein sequence features is the main reason that affects the prediction performance of current protein function annotation algorithms based solely on sequence information.

[0151] The following describes the method and apparatus, electronic device, and storage medium for protein function annotation according to the embodiments of the present disclosure with reference to the accompanying drawings.

[0152] FIG1 is a schematic diagram of a process for protein function annotation provided in an embodiment of the present disclosure.

[0153] As shown in Figure 1, the method includes the following steps:

[0154] Step 101: According to the protein sequence, sequence features, pre-trained sequence features and preset annotation features are obtained respectively.

[0155] In the embodiment of the present disclosure, the protein sequence is a protein sequence that needs to be annotated with protein function. The sequence feature is a coding feature (One-Hot feature) generated according to the protein sequence, the pre-trained sequence feature is a pre-trained feature calculated by the protein language model ProtBert according to the One-Hot feature of the protein, and the preset annotation feature is an InterPro feature retrieved from the InterPro database based on the protein sequence. It should be noted that, in the embodiment of the present disclosure, the sequence feature is described by taking the One-Hot feature obtained by One-Hot encoding as an example, the pre-trained sequence feature is described by taking the feature calculated in advance according to the sequence feature using the protein language model as an example, and the preset annotation feature is described by taking the InterPro feature retrieved from the InterPro database according to the sequence using the InterProScan tool as an example, but this does not constitute a limitation on the sequence feature, the pre-trained sequence feature and the preset annotation feature.

[0156] Protein sequences that need to be annotated for protein function are encoded into a one-hot feature matrix of the protein sequence through one-hot encoding technology; the protein language model ProtBert can mine the internal pattern information of the protein sequence based on the one-hot feature and explicitly express the internal pattern information of the protein sequence; the InterPro feature integrates a variety of features that are strongly related to protein function, and the InterPro feature can be used to classify protein functions with high accuracy.

[0157] The DeepGOPlus algorithm uses the one-hot feature of the protein sequence as the algorithm's only input feature; however, the sequence one-hot feature is relatively low-level, and it is difficult for the algorithm model to efficiently learn functional category information from it, resulting in low model prediction accuracy. Although the InterPro feature is used as the only input feature of the DeepGOZero algorithm, the prediction accuracy is relatively high, but not all proteins have InterPro features, and the model using only InterPro features cannot classify all proteins. The disclosed embodiment can effectively improve the accuracy of protein function annotation by using the aforementioned three protein sequence features for learning.

[0158] Step 102: performing feature extraction processing on the preset annotation feature to obtain a first learning feature.

[0159] In the embodiment of the present disclosure, the pre-trained feature is calculated based on the One-Hot feature, and has little difference from the One-Hot feature; while the InterPro feature is very different from the One-Hot feature and the pre-trained feature. Simply adding the three features at the shallow level of the network of the protein function annotation model will destroy the individual characteristics of each feature, thereby affecting the prediction accuracy of the protein function annotation model. However, simply deepening the number of network layers of the protein function annotation model will cause the problem of gradient disappearance, affect the transmission of information within the network, and thus affect the prediction accuracy of the protein function annotation model. The embodiment of the present disclosure can alleviate the problem of gradient disappearance by inputting the preset annotation features (InterPro features) into the first preset number of residual fully connected modules for feature extraction processing, so that the protein function annotation model can deepen the network layer and well integrate different protein sequence features at the high level of the network.

[0160] In the protein function annotation model, using a larger number of residual fully connected modules will result in overfitting characteristics, affecting the prediction accuracy, and a smaller number of residual fully connected modules will also affect the prediction accuracy; specifically, it is more appropriate to set the first preset number to about 4; it should be noted that the embodiment of the present disclosure is described with 4 residual fully connected modules, but this does not constitute a limitation on the first preset number.

[0161] Step 103: Perform feature extraction processing on the sequence feature and the pre-trained sequence feature to obtain a second learning feature.

[0162] In the embodiment of the present disclosure, please refer to the description of step 102. In order to alleviate the problem of gradient disappearance in the protein function annotation model, the sequence features and the pre-trained sequence features are input into a second preset number of residual convolution modules for feature extraction processing.

[0163] In the protein function annotation model, using a larger number of residual convolution modules will result in overfitting characteristics, affecting the prediction accuracy, and a smaller number of residual convolution modules will also affect the prediction accuracy; specifically, it is more appropriate to set the second preset number to about 8; it should be noted that the embodiment of the present disclosure is described with 8 residual convolution modules, but this does not constitute a limitation on the second preset number.

[0164] Step 104: Fusing the first learning feature with the second learning feature to obtain a first fused feature.

[0165] In the disclosed embodiment, because InterPro features differ significantly from one-hot features and pre-trained features, the one-hot features, pre-trained features, and InterPro features are respectively input into a second preset number of residual convolutional modules and a first preset number of residual fully connected modules for feature extraction. The extracted first learned features and second learned features are then fused at the high-level network of the protein function annotation model to generate a first fused feature. Fusion of these different features can compensate for deficiencies between them, thereby improving the overall feature quality and enhancing the prediction accuracy of the protein function annotation model.

[0166] Step 105: Process the first fusion feature to obtain protein function annotation information.

[0167] In the disclosed embodiment, simply fusing the first and second learned features does not directly yield protein functional annotation information. It is necessary to perform feature extraction on the first fused feature to fully exploit the advantages of one-hot features, pre-trained features, and InterPro features, achieving information complementarity between the features. Protein functional annotation information is generated by performing feature extraction on the first fused feature.

[0168] The present disclosure provides a method for protein function annotation, which obtains sequence features, pre-trained sequence features, and preset annotation features based on the protein sequence; performs feature extraction processing on the preset annotation features based on a first preset number of residual fully connected modules to obtain a first learning feature; performs feature extraction processing on the sequence features and the pre-trained sequence features based on a second preset number of residual convolution modules to obtain a second learning feature; fuses the first learning feature with the second learning feature to obtain a first fused feature; and processes the first fused feature to obtain protein function annotation information. Compared with related technologies, the present disclosure obtains sequence features, pre-trained sequence features, and preset annotation features based on the protein sequence, and by fusing the features from the above three different angles, the obtained protein function annotation information has more protein function information, effectively improving the prediction accuracy of protein function annotation.

[0169] In order to clearly illustrate the embodiments of the present disclosure, the embodiments of the present disclosure provide a flowchart of another method for protein function annotation.

[0170] As shown in FIG2 , the method includes the following steps:

[0171] Step 201 : obtaining sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence.

[0172] For the detailed description of step 201 , please refer to the above embodiment, so it will not be described in detail here.

[0173] Step 202: Based on a first preset number of fully connected layers, feature extraction processing is performed on the preset annotation features to obtain third learning features.

[0174] Specifically, in the embodiments of the present disclosure, a protein function annotation model (MergeNet algorithm) is provided that fuses multiple sequence features at the shallow level of the network, which makes up for the shortcomings of the model algorithm using a single feature, improves the prediction accuracy of the algorithm using only one-hot features, and increases the universality of the algorithm using only InterPro features.

[0175] For details, please refer to Figure 3; MergeNet uses a fully connected layer (MLP) to perform preliminary learning on InterPro features and obtain the learned InterPro features Wherein d is the feature dimension of the hidden layer. In this embodiment, d is set to 1024, and the feature dimension d of all hidden layers remains consistent.

[0176] Step 203: Based on a second preset number of convolutional layers, feature extraction processing is performed on the sequence features and the pre-trained sequence features to obtain a fourth learning feature and a fifth learning feature.

[0177] Specifically, in the embodiment of the present disclosure, please continue to refer to FIG3 ; the one-hot feature H and the pre-trained feature E obtained according to the protein sequence are input into a one-dimensional convolutional layer (CNN) for preliminary feature learning, and the obtained Where d is the feature dimension of the convolutional layer, and then H and E retain the largest feature value in the residue dimension to perform the maximum pooling operation, and we get and

[0178] Step 204 : Fusing the third learning feature, the fourth learning feature, and the fifth learning feature to obtain a second fused feature; wherein the fusion process is an addition process.

[0179] Specifically, in the embodiment of the present disclosure, MergeNet fuses the three features by summing them up in the shallow layer of the network (i.e., after all three features have passed through a feature learning layer) to obtain the protein feature Exemplarily, the summing process is

[0180] Step 205: Based on a third preset number of fully connected layers, the second fusion features are classified and calculated to obtain protein function annotation information.

[0181] Specifically, in the embodiment of the present disclosure, the protein feature P is fed into the fully connected layer to calculate the logical probability of each functional category. (Protein function annotation information), where class is the total number of protein functions. In this example, class is 5901. A schematic diagram of the prediction results is shown in Figure 4. The figure selects three functions of ten proteins predicted by the algorithm. The first column is the predicted protein name, and the next three columns are the predicted functions for each protein. The "|" before the function indicates the number of the function in the Gene Ontology predicted by the algorithm, and the "|" after the function indicates the probability value of the protein having the function.

[0182] In order to clearly illustrate the embodiments of the present disclosure, the embodiments of the present disclosure provide a flowchart of another method for protein function annotation.

[0183] As shown in FIG5 , the method includes the following steps:

[0184] Step 301: Encode the protein sequence based on a preset encoding algorithm to obtain the sequence feature.

[0185] Step 302: Pre-train the sequence features based on a preset protein language model to obtain the pre-trained sequence features.

[0186] Specifically, in the embodiment of the present disclosure, the preset encoding algorithm is a One-Hot encoding algorithm, which is used to encode the protein sequence to be annotated to generate a One-Hot feature. It should be noted that the embodiment of the present disclosure does not limit the preset encoding algorithm. The One-Hot feature is pre-trained using the protein language model ProtBert. The protein language model can mine the internal pattern information of the protein sequence and explicitly express the internal pattern information of the protein sequence.

[0187] For example, see Figure 6, where the One-Hot feature of a protein sequence is calculated based on the sequence using One-Hot encoding. Where N is the number of amino acids in the protein, and 20 is the characteristic dimension of each residue, which is also the total number of amino acid categories. If the amino acid corresponds to the jth of the 20 amino acid categories, then H i,j =1,H i The values ​​at the other 19 positions are all 0; the OneHot feature of the sequence is sent to the pre-training model ProtBert to obtain the pre-training feature Where 30 is the feature dimension of the residues in the pre-trained feature. Unlike the One-Hot feature, the pre-trained feature E is not a binary feature matrix composed of 0 and 1, but a feature matrix composed of continuous floating-point numbers. Take the protein SKSP4_ASCTR composed of eight residues in SwissProt as an example. Its sequence is: CHYIFNTC. The One-Hot feature and pre-trained feature of the sequence are shown in Figure 6. The dimension of the One-Hot feature of the sequence is The dimension of its sequence pre-training feature is

[0188] Step 303: Searching for the preset annotation feature corresponding to the protein sequence in a preset protein database.

[0189] Specifically, in the embodiment of the present disclosure, please refer to Figure 7, which is a schematic diagram of the network architecture of the ResMergeNet model algorithm. The InterProScan tool is used to retrieve the InterPro features of proteins that are strongly correlated with functional categories from the InterPro database. Where M is the dimension of the protein InterPro feature, that is, the total number of categories of the InterPro feature.

[0190] Step 304: Input the preset annotation features into a first preset number of residual fully connected modules for feature extraction processing to obtain the first learning features.

[0191] The steps specifically include: inputting the preset annotation features into the first residual fully connected module in the residual fully connected module; using the output features of the first residual fully connected module as the input features of the second residual fully connected module for feature extraction processing; and looping the above steps for a first predetermined number of times to obtain the first learning feature, wherein the first predetermined number is equal to the first preset number.

[0192] The first residual fully connected module in which the preset annotation feature is input into the residual fully connected module includes: inputting the preset annotation feature into the normalization layer for normalization processing to obtain the normalized preset annotation feature; inputting the normalized preset annotation feature into the nonlinear layer for nonlinear mapping to obtain the nonlinear mapped preset annotation feature; inputting the nonlinear mapped preset annotation feature into the fully connected layer for feature extraction to obtain a first preset annotation feature; inputting the first preset annotation feature into the normalization layer, the nonlinear layer and the fully connected layer in sequence for processing to obtain a second preset annotation feature; and summing the preset annotation feature and the second preset annotation feature to obtain a first output feature.

[0193] Specifically in the embodiment of the present disclosure, please continue to refer to FIG7, and the InterPro feature (M is the total number of InterPro feature categories) is sent to 4 sequentially connected residual fully connected modules (ResMLP) for high-level feature extraction. In each residual fully connected module, the input features are first normalized by the normalization layer (Norm), then nonlinearly mapped by the nonlinear layer (ReLU), and then learned by the fully connected layer (MLP). Then, the features are learned sequentially by the normalization layer, nonlinear layer, and fully connected layer. At this time, the initial input features of the residual fully connected module are residually summed with the learned features to obtain the high-level InterPro features. (First learning feature).

[0194] Step 305: Based on a fourth preset number of convolutional layers, the sequence features are processed to obtain a sixth learning feature, and the pre-trained sequence features are processed to obtain a seventh learning feature.

[0195] Specifically, in the embodiment of the present disclosure, please continue to refer to FIG7 , the one-hot feature and the pre-trained feature are respectively input into a one-dimensional convolution layer for preliminary feature extraction, and the obtained and Where N is the number of amino acids in the protein, and d is the characteristic dimension of the convolution layer. It should be noted that the embodiment of the present disclosure is described by taking one one-dimensional convolution layer as an example, which does not constitute a limitation on the fourth preset number of layers.

[0196] Step 306: Add the sixth learning feature and the seventh learning feature to obtain a third fusion feature.

[0197] Specifically, in the embodiment of the present disclosure, after learning a one-dimensional convolution layer, the model performs sum processing on these two features, and the summed feature is the third fusion feature P s ,Right now

[0198] Step 307: Input the third fusion feature into a second preset number of residual convolution modules for feature extraction processing to obtain the second learning feature.

[0199] The steps specifically include: inputting the third fusion feature into the first residual convolution module in the residual convolution module; using the output feature of the first residual convolution module as the input feature of the second residual convolution module for feature extraction processing; and looping the above steps for a second predetermined number of times to obtain the second learning feature, wherein the second predetermined number is equal to the second preset number.

[0200] The step of inputting the third fusion feature into the first residual convolution module in the residual convolution module includes: inputting the third fusion feature into the normalization layer for normalization processing to obtain the third fusion feature after normalization processing; inputting the third fusion feature after normalization processing into the nonlinear layer for nonlinear mapping to obtain the third fusion feature after nonlinear mapping; inputting the third fusion feature after nonlinear mapping into the convolution layer for feature learning to obtain the second output feature; inputting the second output feature into the normalization layer, the nonlinear layer and the convolution layer in sequence for processing to obtain the third output feature; and summing the third fusion feature and the third output feature to obtain the fourth output feature.

[0201] Specifically in the embodiment of the present disclosure, please continue to refer to FIG7 , the third fusion feature P sThe input features are first normalized by the normalization layer (Norm), then nonlinearly mapped by the nonlinear layer (ReLU), and then learned by the convolution layer (CNN). The features are then learned by the normalization layer, nonlinear layer, and convolution layer in sequence. After all the residual convolution modules have learned, the fourth output feature with the largest eigenvalue in the residue dimension is retained for maximum pooling to obtain high-level sequence features. (Second learning characteristic).

[0202] Step 308: Concatenate the first learned feature and the second learned feature to obtain a first fusion feature.

[0203] Specifically, in the embodiment of the present disclosure, the high-level sequence feature P s (second learning feature) and high-level InterPro feature I (first learning feature) become more abstract, and the feature distribution becomes similar, both of which are highly correlated with protein function. s The obtained feature is called high-level protein feature (first fusion feature) by splicing it with high-level InterPro feature I. Where concatenate(,) means concatenating two vectors according to the feature dimension.

[0204] Step 309: Input the first fusion feature into an integration module for integration processing to obtain a protein fusion feature.

[0205] The steps specifically include: inputting the first fusion feature into a normalization layer for normalization processing to obtain a first fusion feature after normalization processing; inputting the first fusion feature after normalization processing into a nonlinear layer for nonlinear mapping to obtain a first fusion feature after nonlinear mapping; inputting the first fusion feature after nonlinear mapping into a fully connected layer for feature extraction to obtain a first protein fusion feature; inputting the first protein fusion feature into the normalization layer, the nonlinear layer and the fully connected layer in sequence for processing to obtain a second protein fusion feature; and summing the first fusion feature and the second protein fusion feature to obtain the protein fusion feature.

[0206] Specifically, in the embodiment of the present disclosure, please continue to refer to FIG7 , the high-level protein feature (first fusion feature) is The input residual integration module (Integrate) is used for further protein feature extraction. In the residual integration module, the input features are first normalized by the normalization layer (Norm), then nonlinearly mapped by the nonlinear layer (ReLU), and then learned by the fully connected layer (MLP). Then, the features are learned by the normalization layer, nonlinear layer, and fully connected layer in sequence. At this time, the initial input features of the residue fully connected module are residually summed with the learned features to obtain (Protein fusion signature).

[0207] Step 310 : Based on a fifth preset number of fully connected layers, the protein fusion features are classified to obtain the protein function annotation information.

[0208] Specifically, in the embodiment of the present disclosure, please continue to refer to FIG7 , the obtained protein fusion feature P is sent to the fully connected layer (MLP) to classify the protein function, and the logical probability of the protein functional category is obtained. (Protein function annotation information).

[0209] As a possible implementation of the embodiment of the present disclosure, before respectively obtaining sequence features, pre-trained sequence features, and preset annotation features based on the protein sequence, the method further includes:

[0210] The protein function annotation model is trained based on the training protein sequences.

[0211] Specifically, in the embodiment of the present disclosure, before using the protein function annotation model to perform protein function annotation, the protein sequence for training is input into the protein function annotation model for training to improve the prediction accuracy of the model.

[0212] In order to clearly illustrate the embodiments of the present disclosure, the embodiments of the present disclosure provide a flowchart of a method for training a protein function annotation model.

[0213] As shown in FIG8 , the method includes the following steps:

[0214] Step 401 : obtaining training sequence features, training first sequence features, and training preset annotation features respectively based on the training protein sequence.

[0215] Step 402: Perform feature extraction processing on the preset annotation features for training to obtain first learning features for training.

[0216] Step 403: Perform feature extraction processing on the training sequence feature and the first training sequence feature to obtain a second training learning feature.

[0217] Step 404: Fusing the first training learning feature with the second training learning feature to obtain a first training fused feature.

[0218] Step 405: Process the first training fusion feature to obtain training protein function annotation information.

[0219] For steps 101 to 105 , please refer to the detailed description in the above embodiment, and thus will not be described in detail here.

[0220] Furthermore, in a possible implementation of this embodiment, after obtaining training sequence features, training first sequence features, and training preset annotation features based on the training protein sequence, the embodiment of the present disclosure provides a flow chart of another method for training a protein function annotation model.

[0221] As shown in FIG9 , the method includes the following steps:

[0222] Step 501 : According to the training protein sequence, a training sequence feature, a training first sequence feature, and a training preset annotation feature are respectively obtained.

[0223] Step 502 : Based on the first preset number of training fully connected layers, feature extraction processing is performed on the training preset annotation features to obtain third training learning features.

[0224] Step 503 : Based on the second preset number of training convolutional layers, feature extraction processing is performed on the training sequence features and the training first sequence features to obtain fourth training learning features and fifth training learning features.

[0225] Step 504 : Fusing the third training learning feature, the fourth training learning feature, and the fifth training learning feature to obtain a second training fused feature; wherein the fusion process is an addition process.

[0226] Step 505 : Based on a third preset number of training fully connected layers, classification calculation is performed on the second training fusion features to obtain training protein function annotation information.

[0227] For steps 201 to 205 , please refer to the detailed description in the above embodiment, and thus will not be described in detail here.

[0228] Corresponding to the above-mentioned method for protein function annotation, the present invention also provides a device for protein function annotation. Since the device embodiment of the present invention corresponds to the above-mentioned method embodiment, any details not disclosed in the device embodiment can be referred to the above-mentioned method embodiment and will not be further described in this invention.

[0229] FIG10 is a schematic structural diagram of a protein function annotation apparatus provided by an embodiment of the present disclosure. As shown in FIG10 , the apparatus includes: an acquisition unit 61 , a first processing unit 62 , a second processing unit 63 , a first fusion unit 64 and a third processing unit 65 .

[0230] An acquisition unit 61 is used to acquire sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence;

[0231] A first processing unit 62 is configured to perform feature extraction processing on the preset annotation feature to obtain a first learning feature;

[0232] A second processing unit 63 is configured to perform feature extraction processing on the sequence feature and the pre-trained sequence feature to obtain a second learning feature;

[0233] A first fusion unit 64 is configured to fuse the first learning feature with the second learning feature to obtain a first fused feature;

[0234] The third processing unit 65 is configured to process the first fusion feature to obtain protein function annotation information.

[0235] The present disclosure provides a device for protein function annotation. The device obtains sequence features, pre-trained sequence features, and preset annotation features based on a protein sequence; performs feature extraction processing on the preset annotation features to obtain a first learning feature; performs feature extraction processing on the sequence features and the pre-trained sequence features to obtain a second learning feature; fuses the first learning feature with the second learning feature to obtain a first fused feature; and processes the first fused feature to obtain protein function annotation information. Compared with related technologies, the present disclosure obtains sequence features, pre-trained sequence features, and preset annotation features based on a protein sequence. By fusing the aforementioned three different features, the obtained protein function annotation information is more accurate, effectively improving the prediction accuracy of protein function annotation.

[0236] Furthermore, in a possible implementation of this embodiment, as shown in FIG11 , the apparatus further includes:

[0237] a fourth processing unit 66 configured to, after obtaining sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence, perform feature extraction processing on the preset annotation features based on a first preset number of fully connected layers to obtain third learning features;

[0238] a fifth processing unit 67 configured to perform feature extraction processing on the sequence feature and the pre-trained sequence feature based on a second preset number of convolutional layers to obtain a fourth learning feature and a fifth learning feature;

[0239] a second fusion unit 68 configured to fuse the third learning feature, the fourth learning feature, and the fifth learning feature to obtain a second fused feature; wherein the fusion process is an addition process;

[0240] The sixth processing unit 69 is configured to perform classification calculation on the second fusion features based on a third preset number of fully connected layers to obtain protein function annotation information.

[0241] Furthermore, in a possible implementation of this embodiment, as shown in FIG11 , the acquiring unit 61 includes:

[0242] An encoding module 611 is configured to encode the protein sequence based on a preset encoding algorithm to obtain the sequence feature;

[0243] A pre-training module 612 is configured to perform pre-training processing on the sequence features based on a preset protein language model to obtain the pre-trained sequence features;

[0244] The search module 613 is configured to search a preset protein database for the preset annotation feature corresponding to the protein sequence.

[0245] Furthermore, in a possible implementation of this embodiment, as shown in FIG11 , the first processing unit 62 includes:

[0246] A first input module 621 is configured to input the preset annotation features into a first preset number of residual fully connected modules;

[0247] A first processing module 622 is configured to perform feature extraction processing on the output features of the first residual fully connected module as input features of the second residual fully connected module;

[0248] The first loop module 623 is configured to loop and execute the processing method of the first processing module a first predetermined number of times to obtain the first learning feature, wherein the first predetermined number of times is equal to the first preset number.

[0249] Furthermore, in a possible implementation of this embodiment, the first input module 621 is further configured to:

[0250] Inputting the preset annotation features into a normalization layer for normalization processing to obtain the normalized preset annotation features;

[0251] Inputting the normalized preset annotation features into a nonlinear layer for nonlinear mapping to obtain the preset annotation features after nonlinear mapping;

[0252] Inputting the preset annotation features after the nonlinear mapping into a fully connected layer for feature extraction to obtain a first preset annotation feature;

[0253] Inputting the first preset annotation feature into the normalization layer, the nonlinear layer, and the fully connected layer in sequence for processing to obtain a second preset annotation feature;

[0254] The preset annotation feature and the second preset annotation feature are added together to obtain a first output feature.

[0255] Furthermore, in a possible implementation of this embodiment, as shown in FIG11 , the second processing unit 63 includes:

[0256] A second processing module 631 is configured to process the sequence features based on a fourth preset number of convolutional layers to obtain a sixth learning feature, and to process the pre-trained sequence features to obtain a seventh learning feature;

[0257] A first adding module 632 is configured to add the sixth learning feature and the seventh learning feature to obtain a third fusion feature;

[0258] A second input module 633 is used to input the third fusion feature into a second preset number of residual convolution modules;

[0259] A first extraction module 634 is configured to perform feature extraction processing on the output features of the first residual convolution module as input features of the second residual convolution module;

[0260] The second loop module 635 is configured to loop and execute the processing method of the first extraction module a second predetermined number of times to obtain the second learning feature, wherein the second predetermined number of times is equal to the second preset number.

[0261] Furthermore, in a possible implementation of this embodiment, the second input module 633 is further configured to:

[0262] Inputting the third fused feature into a normalization layer for normalization processing to obtain a normalized third fused feature;

[0263] Inputting the normalized third fusion feature into a nonlinear layer for nonlinear mapping to obtain the nonlinearly mapped third fusion feature;

[0264] Inputting the third fused feature after the nonlinear mapping into the convolution layer for feature learning to obtain a second output feature;

[0265] Inputting the second output feature into the normalization layer, the nonlinear layer, and the convolution layer in sequence for processing to obtain a third output feature;

[0266] The third fusion feature and the third output feature are added together to obtain a fourth output feature.

[0267] Furthermore, in a possible implementation of this embodiment, the first fusion unit 64 is further configured to:

[0268] The first learning feature and the second learning feature are concatenated to obtain a first fusion feature.

[0269] Furthermore, in a possible implementation of this embodiment, as shown in FIG11 , the third processing unit 65 includes:

[0270] Integration module 651, configured to input the first fusion feature into an integration module for integration processing to obtain a protein fusion feature;

[0271] The classification module 652 is configured to classify the protein fusion features based on a fifth preset number of fully connected layers to obtain the protein function annotation information.

[0272] Furthermore, in a possible implementation of this embodiment, the integration module 651 is further configured to:

[0273] Inputting the first fused feature into a normalization layer for normalization processing to obtain a normalized first fused feature;

[0274] Inputting the normalized first fusion feature into a nonlinear layer for nonlinear mapping to obtain the first fusion feature after nonlinear mapping;

[0275] Inputting the first fusion feature after the nonlinear mapping into a fully connected layer for feature extraction to obtain a first protein fusion feature;

[0276] Inputting the first protein fusion feature into the normalization layer, the nonlinear layer, and the fully connected layer in sequence for processing to obtain a second protein fusion feature;

[0277] The first fusion feature and the second protein fusion feature are added together to obtain the protein fusion feature.

[0278] Furthermore, in a possible implementation of this embodiment, as shown in FIG11 , the apparatus further includes:

[0279] The training unit 610 is used to train the protein function annotation model based on the training protein sequence before respectively obtaining the sequence features, the pre-trained sequence features and the preset annotation features according to the protein sequence.

[0280] Furthermore, in a possible implementation of this embodiment, as shown in FIG11 , the training unit 610 includes:

[0281] A training acquisition module 6101 is used to acquire training sequence features, training first sequence features, and training preset annotation features based on the training protein sequence;

[0282] A first training processing module 6102 is configured to perform feature extraction processing on the preset annotation features for training to obtain first learning features for training;

[0283] A second training processing module 6103 is configured to perform feature extraction processing on the training sequence features and the first training sequence features to obtain second training learning features;

[0284] A first training fusion module 6104 is configured to fuse the first training learning feature with the second training learning feature to obtain a first training fused feature;

[0285] The third training processing module 6105 is used to process the first training fusion feature to obtain training protein function annotation information.

[0286] Furthermore, in a possible implementation of this embodiment, as shown in FIG11 , the training unit 610 further includes:

[0287] a fourth training processing module 6106 for obtaining, based on the training protein sequence, a training sequence feature, a training first sequence feature, and a training preset annotation feature, and then performing feature extraction processing on the training preset annotation feature based on a first preset number of training fully connected layers to obtain a third training learning feature;

[0288] a fifth training processing module 6107, configured to perform feature extraction processing on the training sequence features and the first training sequence features based on a second preset number of training convolutional layers, to obtain fourth training learning features and fifth training learning features;

[0289] a second training fusion module 6108 for fusing the third training learning feature, the fourth training learning feature, and the fifth training learning feature to obtain a second training fused feature; wherein the fusion process is an addition process;

[0290] The sixth training processing module 6109 is configured to perform classification calculation on the second training fusion features based on a third preset number of training fully connected layers to obtain training protein function annotation information.

[0291] It should be noted that the above explanation of the method embodiment is also applicable to the device of this embodiment, and the principles are the same, which is not limited in this embodiment.

[0292] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0293] FIG12 shows a schematic block diagram of an example electronic device 700 that can be used to implement an embodiment of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0294] As shown in FIG12 , the device 700 includes a computing unit 701 that can perform various appropriate actions and processes based on a computer program stored in a ROM (Read-Only Memory) 702 or a computer program loaded from a storage unit 708 into a RAM (Random Access Memory) 703. Various programs and data required for the operation of the device 700 can also be stored in the RAM 703. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An I / O (Input / Output) interface 705 is also connected to the bus 704.

[0295] Various components in device 700 are connected to I / O interface 705, including an input unit 706, such as a keyboard, mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a magnetic disk, optical disk, etc.; and a communication unit 709, such as a network card, modem, wireless communication transceiver, etc. The communication unit 709 allows device 700 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0296] The computing unit 701 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a CPU (Central Processing Unit), a GPU (Graphic Processing Unit), various specialized AI (Artificial Intelligence) computing chips, various computing units that run machine learning model algorithms, a DSP (Digital Signal Processor), and any suitable processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as the protein function annotation method. For example, in some embodiments, the protein function annotation method can be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as the storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the method described above can be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to execute the aforementioned protein function annotation method in any other appropriate manner (eg, by means of firmware).

[0297] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, FPGAs (Field Programmable Gate Arrays), ASICs (Application-Specific Integrated Circuits), ASSPs (Application-Specific Standard Products), SOCs (System on Chips), CPLDs (Complex Programmable Logic Devices), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special-purpose or general-purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0298] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0299] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or apparatus, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, RAM, ROM, EPROM (Electrically Programmable Read-Only-Memory) or flash memory, optical fiber, CD-ROM (Compact Disc Read-Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0300] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (Cathode-Ray Tube) or LCD (Liquid Crystal Display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0301] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: LAN (Local Area Network), WAN (Wide Area Network), the Internet, and blockchain networks.

[0302] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. This client-server relationship is established by computer programs running on the respective computers, establishing a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and VPS services ("Virtual Private Servers" or simply "VPS"). The server may also be a server in a distributed system or a server integrated with blockchain.

[0303] It's important to note that artificial intelligence (AI) is the study of how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). This encompasses both hardware and software technologies. AI hardware technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, and big data processing. AI software technologies primarily encompass computer vision, speech recognition, natural language processing, machine learning / deep learning, big data processing, and knowledge graphs.

[0304] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0305] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for protein function annotation, characterized in that: include: According to the protein sequence, sequence features, pre-trained sequence features and preset annotation features are obtained respectively; Performing feature extraction processing on the preset annotation feature to obtain a first learning feature; Performing feature extraction processing on the sequence feature and the pre-trained sequence feature to obtain a second learning feature; Fusing the first learning feature with the second learning feature to obtain a first fused feature; The first fusion feature is processed to obtain protein function annotation information.

2. The method according to claim 1, characterized in that After respectively acquiring sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence, the method further includes: Based on the fully connected layer with the first preset number of layers, feature extraction processing is performed on the preset annotation feature to obtain a third learning feature; Based on the convolutional layer with a second preset number of layers, feature extraction processing is performed on the sequence feature and the pre-trained sequence feature to obtain a fourth learning feature and a fifth learning feature; The third learning feature, the fourth learning feature and the fifth learning feature are fused to obtain a second fused feature; wherein the fusion process is an addition process; Based on a third preset number of fully connected layers, the second fusion features are classified and calculated to obtain protein function annotation information.

3. The method according to claim 1, characterized in that The step of obtaining sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence comprises: Based on a preset encoding algorithm, encoding the protein sequence to obtain the sequence feature; Based on a preset protein language model, pre-training the sequence features to obtain the pre-trained sequence features; In a preset protein database, the preset annotation feature corresponding to the protein sequence is searched.

4. The method according to claim 1, characterized in that: The step of performing feature extraction processing on the preset annotation feature to obtain the first learning feature comprises the following steps: Inputting the preset annotation features into a first preset number of residual fully connected modules; The output features of the first residual fully connected module are used as the input features of the second residual fully connected module for feature extraction; The above steps are looped for a first predetermined number of times to obtain the first learning feature, wherein the first predetermined number of times is equal to the first preset number.

5. The method according to claim 4, characterized in that The first residual fully connected module that inputs the preset annotation feature into the residual fully connected module includes: Inputting the preset annotation features into a normalization layer for normalization processing to obtain the preset annotation features after normalization processing; Inputting the normalized preset annotation features into a nonlinear layer for nonlinear mapping to obtain the preset annotation features after nonlinear mapping; Inputting the preset annotation features after the nonlinear mapping into a fully connected layer for feature extraction to obtain a first preset annotation feature; Inputting the first preset annotation feature into the normalization layer, the nonlinear layer and the fully connected layer in sequence for processing to obtain a second preset annotation feature; The preset annotation feature and the second preset annotation feature are added together to obtain a first output feature.

6. The method according to claim 3, characterized in that Based on a fourth preset number of convolutional layers, the sequence features are processed to obtain a sixth learning feature, and the pre-trained sequence features are processed to obtain a seventh learning feature; Adding the sixth learning feature and the seventh learning feature to obtain a third fusion feature; The performing feature extraction processing on the sequence feature and the pre-trained sequence feature to obtain the second learning feature comprises the following steps: Inputting the third fused feature into a second preset number of residual convolution modules; Using the output features of the first residual convolution module as input features of the second residual convolution module for feature extraction processing; The above steps are looped for a second predetermined number of times to obtain the second learning feature, wherein the second predetermined number of times is equal to the second preset number.

7. The method according to claim 6, characterized in that The first residual convolution module that inputs the third fusion feature into the residual convolution module includes: Inputting the third fused feature into a normalization layer for normalization processing to obtain a normalized third fused feature; Inputting the normalized third fused feature into a nonlinear layer for nonlinear mapping to obtain the third fused feature after nonlinear mapping; Inputting the third fused feature after the nonlinear mapping into the convolution layer for feature learning to obtain a second output feature; Inputting the second output feature into the normalization layer, the nonlinear layer and the convolution layer in sequence for processing to obtain a third output feature; The third fusion feature and the third output feature are added together to obtain a fourth output feature.

8. The method according to claim 1, characterized in that The fusing the first learning feature with the second learning feature to obtain a first fused feature includes: The first learning feature and the second learning feature are concatenated to obtain a first fusion feature.

9. The method according to claim 1, characterized in that: The processing of the first fusion feature to obtain protein function annotation information includes: Inputting the first fusion feature into an integration module for integration processing to obtain a protein fusion feature; Based on a fifth preset number of fully connected layers, the protein fusion features are classified and processed to obtain the protein function annotation information.

10. The method according to claim 9, characterized in that The step of inputting the first fusion feature into an integration module for integration processing to obtain a protein fusion feature comprises: Inputting the first fused feature into a normalization layer for normalization processing to obtain a normalized first fused feature; Inputting the normalized first fused feature into a nonlinear layer for nonlinear mapping to obtain the first fused feature after nonlinear mapping; Inputting the first fusion feature after the nonlinear mapping into a fully connected layer for feature extraction to obtain a first protein fusion feature; Inputting the first protein fusion feature into the normalization layer, the nonlinear layer and the fully connected layer in sequence for processing to obtain a second protein fusion feature; The first fusion feature and the second protein fusion feature are added together to obtain the protein fusion feature.

11. The method according to claim 1, characterized in that Before respectively acquiring sequence features, pre-trained sequence features, and preset annotation features according to the protein sequence, the method further comprises: The protein function annotation model is trained based on the training protein sequences.

12. The method according to claim 11, characterized in that: The step of inputting the training protein sequence into the protein function annotation model for training comprises: According to the training protein sequence, respectively obtaining a training sequence feature, a training first sequence feature and a training preset annotation feature; Performing feature extraction processing on the preset annotation feature for training to obtain a first learning feature for training; Performing feature extraction processing on the training sequence feature and the first training sequence feature to obtain a second training learning feature; Fusing the first training learning feature with the second training learning feature to obtain a first training fused feature; The first training fusion feature is processed to obtain training protein function annotation information.

13. The method according to claim 12, characterized in that After respectively acquiring the training sequence features, the training first sequence features and the training preset annotation features according to the training protein sequence, the method further comprises: Based on the first preset number of training fully connected layers, performing feature extraction processing on the training preset annotation features to obtain third training learning features; Based on the second preset number of training convolutional layers, feature extraction processing is performed on the training sequence features and the training first sequence features respectively to obtain fourth training learning features and fifth training learning features; fusing the third training learning feature, the fourth training learning feature and the fifth training learning feature to obtain a second training fused feature; wherein the fusion process is an addition process; Based on a third preset number of training fully connected layers, the second training fusion features are classified and calculated to obtain training protein function annotation information.

14. A device for protein function annotation, characterized in that: include: An acquisition unit, used for acquiring sequence features, pre-trained sequence features and preset annotation features according to the protein sequence; A first processing unit, configured to perform feature extraction processing on the preset annotation feature to obtain a first learning feature; A second processing unit is used to perform feature extraction processing on the sequence feature and the pre-trained sequence feature to obtain a second learning feature; A first fusion unit, configured to fuse the first learning feature with the second learning feature to obtain a first fused feature; The third processing unit is used to process the first fusion feature to obtain protein function annotation information.

15. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 13.

16. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-13.

17. A computer program product, characterized in that It comprises a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 13.

Images