Prediction of protein stability based on graph neural network

Abstract

The invention provides a method for predicting protein stability based on a graph neural network. Specifically, the invention provides a method for predicting the stability of protein after mutation, which comprises the following steps: based on the sequence information of a to-be-predicted protein sample, carrying out feature extraction by using a pre-training model to obtain the sequence features of the to-be-predicted protein sample; obtaining structural features of the to-be-predicted protein sample based on the three-dimensional structural information of the to-be-predicted protein sample; and based on the sequence features and the structural features, using a prediction model based on a graph neural network to perform prediction, and determining a stability prediction result of the to-be-predicted protein sample after mutation. The invention further provides a construction method of the stability prediction model after protein mutation. By combining the natural language processing model with the graph neural network, the stability change caused by mutation is predicted, and the performance superior to that of an early-stage method is shown.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the priority of Chinese patent application No. 202310905338.0, filed with the Chinese Patent Office on July 21, 2023, and entitled “Predicting Protein Stability Based on Graph Neural Networks”, and incorporates its entire contents into this application by reference. Technical Field

[0003] The present disclosure relates to protein stability research, and more specifically to predicting protein stability based on a graph neural network (GNN). Background Art

[0004] Proteins can only perform specific functions by adopting specific three-dimensional structures, and their ability to fold into corresponding three-dimensional structures is affected by protein stability. (See Ittisoponpisan, Sirawit et al., 2019, “Can Predicted Protein 3D Structures Provide Reliable Insights into Whether Missense Variants Are Disease Associated?”, Journal of Molecular Biology 431(11):2197–2212 and Ofoegbu, Tochukwu C. et al., 2019, “Phyre Risk: A Dynamic Web Application to Bridge Genomics, Proteomics and 3D Structure Data to Guide Interpretation of Human Genetic Variants”, Journal of Molecular Biology 431(13):2460–66. The entire contents of the above documents are hereby incorporated into the present disclosure by reference and become part of the content of the present disclosure.) Accurately predicting the impact of mutations on protein stability (folding free energy) is crucial in many fields from biotechnology to medicine. For example, in protein engineering, targeted redesign of proteins makes it possible to optimize the biotechnology and biopharmaceutical processes involved in these proteins. (See Coluzza, Ivan, 2017, "Computational protein design: a review", Journal of Physics: Condensed Matter 29 (14): 143001 and Korendovych, Ivan V., and William F. DeGrado, 2020, "De Novo Protein Design, a Retrospective", Quarterly Reviews of Biophysics 53: e3. The entire contents of the above documents are hereby incorporated into the present disclosure by reference, making them part of the content of the present disclosure.) Stability prediction also plays a key role in explaining the impact of human genetic variation, which can help people better understand how these variations cause disease.(See Gunning, Adam C et al., 2021, “Assessing Performance of Pathogenicity Predictors Using Clinically Relevant Variant Datasets”, Journal of Medical Genetics 58(8):547–55 and Kopanos, Christos et al., 2019, “VarSome: The Human Genomic Variant Search Engine” ed. Jonathan Wren, Bioinformatics 35(11):1978–80. The entire contents of the above documents are hereby incorporated into the present disclosure by reference, making them part of the contents of the present disclosure.) Directed evolution and rational design are two common protein modification strategies. Directed evolution strategies have been successfully applied to the modification of enzyme properties such as enzyme activity, stability, substrate specificity, and stereoselectivity, and American scientist Frances H. Arnold won the 2018 Nobel Prize in Chemistry for this. However, directed evolution requires the construction of a large-scale mutant library and the establishment of a high-throughput screening method, which will consume a lot of manpower, material resources and financial resources. Directed evolution methods also have difficulty in completing a comprehensive search of sequence space and have fundamental defects. Rational design strategies rely on the understanding of the relationship between protein structure and function to predict possible mutants, and then construct mutants in the target gene by means of site-directed mutagenesis. In essence, rational design is more efficient in modifying proteins than directed evolution. In addition, rational design methods are universal, and an effective rational design strategy can be widely applied to the modification of a variety of proteins. However, the accuracy of rational design methods is generally low, and the scope of application is far less extensive than directed evolution. In the past few decades, many studies have been devoted to establishing newer and more effective methods, with the latest research and development direction of predicting the effects of mutations on protein stability based on artificial intelligence. In recent years, artificial intelligence-assisted protein engineering has gradually developed into a new and efficient protein design strategy, showing unique advantages in protein structure prediction, function prediction, stability prediction, and antibody affinity prediction, becoming another wave of technology after rational design and directed evolution.

[0005] In the past decade, artificial intelligence has made unprecedented improvements to various protein prediction methods, from predicting residue interactions at the protein sequence embedding layer to predicting protein structure and function. Among them, natural language processing is one of the hottest areas of artificial intelligence at present, and is often used for text generation, question answering and translation. Natural language processing itself is suitable for analyzing the structure and meaning of text, and the most basic nucleic acids and proteins in life are composed of sequences, which is intrinsically related to text. The main concept behind these methods is to interpret protein sequences as sentences and their components, amino acids, as single words. These methods have been able to extract specific features directly from the amino acid sequence of proteins, overcoming the limitations of standard feature-based machine learning (ML) methods. Now a variety of protein-based natural language models have been developed to characterize the features of protein sequences and used for downstream protein property prediction, showing excellent performance. (See Rao, Roshan et al., 2019, "Evaluating Protein Transfer Learning with TAPE", Advances in Neural Information Processing Systems 32: 9689–9701. The entire contents of the above-mentioned documents are incorporated into the present disclosure by reference, making them part of the contents of the present disclosure.)

[0006] In contrast, geometric deep learning methods, especially graph neural networks (GNNs), can predict various properties of proteins by performing convolution operations on protein structures. If the three-dimensional structure of a protein is regarded as a network in which each amino acid is directly connected according to a certain relationship, then the connection between each node and other nodes is the interaction between a certain amino acid and other surrounding amino acids, which reflects the characteristics of the protein structure. This feature is generally called the edge feature of the graph, and the weight can be reflected by the distance between points. For example, in the structure of a drug molecule, the edge between two atoms can be determined by the distance between them or the type of chemical bond. By considering the characteristics and weights of the edges, GNNs can more accurately capture the relationship and distance between nodes, thereby improving the expressiveness and performance of the model. GNNs have achieved great success in various problems, from learning features of quantitative structure-activity relationship models, to predicting the biochemical activity of drugs, to predicting the interface between protein pairs. Summary of the invention

[0007] Based on the above-mentioned problems with the current methods for predicting the effects of mutations on protein stability, the present disclosure proposes a fast and accurate method for predicting the effects of mutations on protein stability. It is based on the latest natural language processing method, using the T5 (i.e., Transfer Text-to-Text Transformer) pre-trained model to extract sequence features, combined with structure-based features, and input into a graph neural network model, preferably a graph convolutional neural network model, to predict changes in folding free energy caused by amino acid substitutions, and obtain prediction results for protein stability. The results show that the performance has been further improved over existing methods.

[0008] According to the first aspect of the present disclosure, a method for predicting protein stability after mutation is provided. The method may include the following steps: based on the sequence information of the protein sample to be predicted, using a pre-trained model to perform feature extraction to obtain the sequence features of the protein sample to be predicted; based on the three-dimensional structure information of the protein sample to be predicted, obtaining the structural features of the protein sample to be predicted; based on the sequence features and the structural features, using a prediction model based on a graph neural network to perform prediction, and determining the stability prediction result of the protein sample to be predicted after mutation.

[0009] In the method according to the first aspect of the present disclosure, the sequence information of the protein sample to be predicted may include amino acid sequence information of the protein sample to be predicted before mutation and amino acid sequence information of the protein sample to be predicted after mutation.

[0010] On the other hand, the sequence information of the protein sample to be predicted may include the amino acid sequence information of the protein sample to be predicted before mutation and the designated mutation information. Thus, based on the amino acid sequence information of the protein sample to be predicted before mutation and the designated mutation information, the amino acid sequence information of the protein sample to be predicted after mutation can be obtained.

[0011] Preferably, the specified mutation information includes a specified mutation site and / or a specified amino acid type after mutation.

[0012] In the method according to the first aspect of the present disclosure, the method of extracting features based on the sequence information of the protein sample to be predicted using a pre-trained model to obtain the sequence features of the protein sample to be predicted may further include: extracting features from the amino acid sequence information of the protein sample to be predicted before mutation and the amino acid sequence information after mutation using the pre-trained model respectively to obtain the sequence feature information before mutation and the sequence feature information after mutation; and splicing the sequence feature information before mutation and the sequence feature information after mutation to obtain the sequence features of the protein sample to be predicted.

[0013] In the method according to the first aspect of the present disclosure, the pre-trained model can be implemented by a combination of one or more of the following models: ESM-1b, UniRef, ProteinBert, TAPE, ProtGPT2, ProtTXL, ProtBert, ProtXLNet, ProtAlbert, ProtElectra, ProtT5-XL and ProtT5-XXL; preferably, the pre-trained model is ProtT5-XL or ProtT5-XXL; more preferably, the pre-trained model is ProtT5-XL.

[0014] In the method according to the first aspect of the present disclosure, the three-dimensional structure information of the protein sample to be predicted may include the three-dimensional structure information of the protein sample to be predicted obtained through at least one of the following databases or prediction software: PDB database, AlphaFold2, I-TASSER, RoseTTAFold, Modeller and Swiss-model.

[0015] In the method according to the first aspect of the present disclosure, the three-dimensional structural information may include an interaction network, a secondary structure, an amino acid residue distance or a physical environment. Preferably, the three-dimensional structural information includes an interaction network.

[0016] Preferably, the three-dimensional structural information of the protein sample to be predicted is obtained in the following manner: obtaining the coordinates of the three-dimensional structure of the protein sample to be predicted; calculating the distance between the amino acid residue pairs in the three-dimensional structure of the protein according to the obtained coordinates of the three-dimensional structure; when the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and it itself is recorded as 0, thereby obtaining the adjacency matrix of the entire protein sample to be predicted.

[0017] Accordingly, in the method according to the first aspect of the present disclosure, the three-dimensional structural information of the protein sample to be predicted, obtaining the structural features of the protein sample to be predicted may include: taking the mutation site as the center, intercepting an adjacency matrix of a specified size from the adjacency matrix of the entire protein sample to be predicted as the structural features of the protein sample to be predicted; or, according to the amino acids in the adjacency matrix of the entire protein sample to be predicted, calculating the distance between the amino acid residue pairs, when the alpha carbon atom distance between the two residue pairs is less than 10 angstroms, indicating that there is an edge relationship between the nodes, recorded as 1, otherwise 0, and itself recorded as 0, to obtain the second-order adjacency matrix of the protein sample to be predicted. In some embodiments, taking the mutation site as the center, an adjacency matrix of N*N (N is an integer greater than or equal to 1) is intercepted from the adjacency matrix of the entire protein. Intercepting the N*N adjacency matrix from the adjacency matrix of the entire protein can be intercepted based on the sequences on the left and right sides of the mutation site, or it can be intercepted based on the distance between the mutation site. In one embodiment, with the mutation site as the center, amino acids whose distance from the mutation site is less than 10 angstroms in the adjacency matrix of the entire protein sample to be predicted are selected as structural features of the protein sample to be predicted.

[0018] In the method according to the first aspect of the present disclosure, the prediction based on the sequence features and the structural features using a prediction model based on a graph neural network to determine the stability prediction result of the protein sample to be predicted after mutation can further include: inputting the sequence features and the structural features into the prediction model based on a graph neural network; updating and outputting the features using several layers of graph network structures through matrix multiplication and aggregation operations; merging the features outputted by several layers of graph network structures respectively; compressing the features using a pooling operation; and classifying the features for stability according to the output layer classification function. Preferably, the pooling operation can be average pooling, maximum pooling, or K-max pooling.

[0019] In the method according to the first aspect of the present disclosure, the prediction model based on the graph neural network can be constructed by the following steps: obtaining a training set, the training set including information of multiple training samples, the information of each training sample including three-dimensional structural information of the sample protein, amino acid sequence information of the sample protein before and after mutation, and a mutation stability label; based on the amino acid sequence information of the sample protein before and after mutation, performing feature extraction through a pre-training model to obtain sequence features of the multiple training samples; based on the three-dimensional structural information of the sample protein, obtaining structural features of the multiple training samples; based on the sequence features, the structural features and the mutation stability label, training the graph neural network to obtain the prediction model based on the graph neural network.

[0020] Preferably, the graph neural network-based prediction model is a classifier.

[0021] Preferably, the graph neural network is a graph convolutional network.

[0022] More preferably, the prediction model based on graph neural network includes one of the following models: MGC model, MGC_F model, MGC_S model, GINConv model and SageConv model.

[0023] Preferably, experimental data in a public database can be obtained as a training set. At the same time, the training set can be doubled in size by inverting the experimental data in the public database.

[0024] According to the second aspect of the present disclosure, a method for constructing a prediction model for stability after protein mutation is provided. The method may include: obtaining a training set, the training set including information of multiple training samples, the information of each training sample including three-dimensional structural information of the sample protein, amino acid sequence information of the sample protein before and after mutation, and mutation stability label; based on the amino acid sequence information of the sample protein before and after mutation, extracting features through a pre-training model to obtain sequence features of the multiple training samples; based on the three-dimensional structural information of the sample protein, obtaining structural features of the multiple training samples; based on the sequence features, the structural features, and the mutation stability label, training a graph neural network to obtain a prediction model based on the graph neural network.

[0025] In the method according to the second aspect of the present disclosure, the feature extraction based on the amino acid sequence information of the sample protein before and after mutation through a pre-trained model to obtain the sequence features of the multiple training samples can further include: using the pre-trained model to extract features from the amino acid sequence information of the sample protein before and after mutation, respectively, to obtain sequence feature information before mutation and sequence feature information after mutation; and splicing the sequence feature information before mutation and the sequence feature information after mutation to obtain the sequence features of the multiple training samples.

[0026] In the method according to the second aspect of the present disclosure, the pre-trained model can be implemented by a combination of one or more of the following models: ESM-1b, UniRef, ProteinBert, TAPE, ProtGPT2, ProtTXL, ProtBert, ProtXLNet, ProtAlbert, ProtElectra, ProtT5-XL and ProtT5-XXL; preferably, the pre-trained model is ProtT5-XL or ProtT5-XXL; more preferably, the pre-trained model is ProtT5-XL.

[0027] In the method according to the second aspect of the present disclosure, the three-dimensional structural information of the sample protein may include the three-dimensional structural information of the sample protein obtained through at least one of the following databases or prediction software: PDB database, AlphaFold2, I-TASSER, RoseTTAFold, Modeller and Swiss-model.

[0028] In the method according to the second aspect of the present disclosure, the three-dimensional structural information may include an interaction network, a secondary structure, an amino acid residue distance or a physical environment. Preferably, the three-dimensional structural information includes an interaction network.

[0029] Preferably, the three-dimensional structural information of the sample protein can be obtained in the following manner: obtaining the coordinates of the three-dimensional structure of the sample protein; calculating the distance between the amino acid residue pairs in the three-dimensional structure of the sample protein according to the obtained coordinates of the three-dimensional structure; when the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and it itself is recorded as 0, thereby obtaining the adjacency matrix of the entire sample protein.

[0030] Accordingly, in the method according to the second aspect of the present disclosure, the three-dimensional structural information of the sample protein based on which the structural features of multiple training samples are obtained may include: taking the mutation site as the center, intercepting an adjacency matrix of a specified size from the adjacency matrix of the entire sample protein as the structural feature of the sample protein; or, calculating the distance between the amino acid residue pairs based on the amino acids in the adjacency matrix of the entire protein sample to be predicted, when the alpha carbon atom distance between the two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and it itself is recorded as 0, thereby obtaining the second-order adjacency matrix of the protein sample to be predicted.

[0031] In the method according to the second aspect of the present disclosure, the training of the graph neural network based on the sequence features, the structural features and the mutation stability labels to obtain a prediction model based on the graph neural network may include: inputting the sequence features and the structural features into the graph neural network; updating and outputting the features using several layers of graph network structures through matrix multiplication and aggregation operations; merging the features outputted by several layers of graph network structures respectively; compressing the features using pooling operations; classifying the features for stability according to the output layer classification function; and adjusting the network parameters of the graph neural network using the mutation stability labels as output targets. Preferably, the pooling operation may be average pooling, maximum pooling or K-max pooling.

[0032] Preferably, the graph neural network-based prediction model is a classifier.

[0033] Preferably, the graph neural network is a graph convolutional network.

[0034] More preferably, the prediction model based on graph neural network includes one of the following models: MGC model, MGC_F model, MGC_S model, GINConv model and SageConv model.

[0035] Preferably, experimental data in a public database can be obtained as a training set. At the same time, the training set can be doubled in size by inverting the experimental data in the public database.

[0036] According to the third aspect of the present disclosure, a system for predicting the stability of a protein after mutation is provided. The system may include: an acquisition module for acquiring the sequence information and three-dimensional structure information of the protein sample to be predicted; a processing module for obtaining the stability prediction result of the protein sample to be predicted after mutation by inputting the sequence information and three-dimensional structure information of the protein sample to be predicted. The processing module further includes the following submodules: a sequence feature submodule for extracting features based on the sequence information of the protein sample to be predicted using a pre-trained model to obtain the sequence features of the protein sample to be predicted; a structural feature submodule for obtaining the structural features of the protein sample to be predicted based on the three-dimensional structure information of the protein sample to be predicted; a model prediction submodule for predicting based on the sequence features and the structural features using a prediction model based on a graph neural network to determine the stability prediction result of the protein sample to be predicted after mutation.

[0037] According to the third aspect of the present disclosure, the system for predicting protein stability based on graph neural network can be implemented by a computer. Preferably, the following operations can be implemented by executing a computer program: obtaining sequence information and three-dimensional structure information of the protein sample to be predicted; obtaining the stability prediction result of the protein sample to be predicted after mutation by inputting the sequence information and three-dimensional structure information of the protein sample to be predicted, and the operation further includes the following sub-operations: based on the sequence information of the protein sample to be predicted, using a pre-trained model to perform feature extraction to obtain the sequence characteristics of the protein sample to be predicted; based on the three-dimensional structure information of the protein sample to be predicted, obtaining the structural characteristics of the protein sample to be predicted; based on the sequence characteristics and the structural characteristics, using a prediction model based on graph neural network to perform prediction to determine the stability prediction result of the protein sample to be predicted after mutation.

[0038] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium is provided for storing a computer program. The computer program includes instructions. When the instructions are executed by a processor of an electronic device, the electronic device implements the method for predicting protein stability after mutation according to the first aspect of the present disclosure or the method for constructing a protein stability prediction model after mutation according to the second aspect of the present disclosure.

[0039] According to a fifth aspect of the present disclosure, a computer system is provided. The computer system comprises: a processor, a memory and a computer program. The computer program is stored in the memory and is configured to be executed by the processor. The computer program comprises instructions for implementing the method for predicting protein stability after mutation according to the first aspect of the present disclosure or the method for constructing a protein stability prediction model after mutation according to the second aspect of the present disclosure.

[0040] The above methods are based on advanced natural language processing models combined with graph neural networks, can be applied to different biological fields, and have shown better performance than earlier methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present disclosure will be more fully understood from the following detailed description taken in conjunction with the accompanying drawings, in which like elements are numbered in a similar manner, and in which:

[0042] Figure 1 is a flowchart of a method for predicting protein stability after mutation according to an embodiment of the present disclosure.

[0043] Figure 2 It is a schematic diagram of the association between the method for predicting protein stability after mutation and the method for constructing a model for predicting protein stability after mutation according to an embodiment of the present disclosure.

[0044] Figure 3 It is a flowchart of the method for constructing a protein stability prediction model after mutation used in the present invention.

[0045] Figure 4 It is a diagram of the adjacency matrix.

[0046] Figure 5 It is a schematic diagram of forming a node feature matrix.

[0047] Figure 6 It is a schematic diagram of the construction of the structure-based adjacency matrix.

[0048] Figure 7 It is a schematic diagram of constructing a sequence-based node feature matrix.

[0049] Figure 8 It is a schematic diagram of the framework of the prediction model.

[0050] Fig. 9 Yes Figure 6-8 A comprehensive schematic diagram.

[0051] Fig.10 A schematic block diagram of a system for predicting protein stability after mutation according to the present disclosure is shown. DETAILED DESCRIPTION

[0052] Unless otherwise defined, technical and scientific terms used in the present disclosure have the same meaning as commonly understood by one of ordinary skill in the art to which the present disclosure belongs.

[0053] The technical scheme of the present disclosure is further described in detail below by way of examples and in conjunction with the accompanying drawings. Unless otherwise specified, the methods and materials of the embodiments described below are conventional products that can be purchased from the market. It will be understood by those skilled in the art that the present disclosure belongs to that the methods and materials described below are merely exemplary and should not be construed as limiting the scope of the present disclosure.

[0054] At present, the methods for predicting the effect of mutations on protein stability are mainly divided into structure-based and sequence-based methods. The performance of structure-based methods is usually slightly better than that of sequence-based methods, while sequence-based methods are not limited by the lack of structure. Although only about 0.2% of proteins in the UniProt database have available three-dimensional structures (see Quan, Lijun, Qiang Lv and Yang Zhang, 2016, "STRUM: Structure-Based Prediction of Protein Stability Changes upon Single-Point Mutation", Bioinformatics 32 (19): 2936-46, which is hereby incorporated into the present disclosure in its entirety by reference, making it part of the content of the present disclosure), the emergence of software tools such as AlphaFold2 has greatly compensated for the limitations of the application of structure-based methods due to the lack of three-dimensional structure.

[0055] At this stage, the input features generated by the natural language model only contain sequence information to obtain excellent performance (for example, see Chinese patent application CN202211708226.8, all the contents recorded in the above patent application are incorporated into the present disclosure by reference, making it a part of the content of the present disclosure), and using deep neural networks to extract structural features from the three-dimensional structure of proteins can also obtain good performance. However, there are still great difficulties in further improving it. Therefore, introducing graph neural networks into protein stability prediction and combining them with protein natural language models will be a potential method.

[0056] From the overall perspective of forecasting methods, Figure 1 A flow chart of a method for predicting protein stability after mutation according to an embodiment of the present disclosure is shown.

[0057] like Figure 1 As described in, the method 100 for predicting protein stability after mutation according to an embodiment of the present disclosure begins at step S110. In this step, based on the sequence information of the protein sample to be predicted, a pre-trained model is used to perform feature extraction to obtain the sequence features of the protein sample to be predicted.

[0058] The sequence information of the protein sample to be predicted mentioned here may include the amino acid sequence information of the protein sample to be predicted before mutation and the amino acid sequence information of the protein sample to be predicted after mutation.

[0059] In another case, the sequence information of the protein sample to be predicted may also include the amino acid sequence information of the protein sample to be predicted before mutation and the designated mutation information. In this case, the amino acid sequence information of the protein sample to be predicted after mutation can be obtained according to the amino acid sequence information of the protein sample to be predicted before mutation and the designated mutation information. The designated mutation information may include a designated mutation site and / or a designated amino acid type after mutation. For example, in the case where the designated mutation information is a designated mutation site, the amino acid at the site may be changed to form the amino acid sequence information of the protein sample to be predicted after mutation. In the case where the designated mutation information also includes the designated amino acid type after mutation, the amino acid sequence information after mutation may be an amino acid sequence. In the case where the designated mutation information does not include the designated amino acid type after mutation, the amino acid sequence information after mutation may be 19 amino acid sequences in which the amino acid at the mutation site is replaced with other 19 amino acids. In another case, the designated mutation information only includes the designated amino acid type after mutation. In this case, each site of the sequence before mutation needs to be sequentially changed with amino acids to form N sequences after mutation (N is the number of sites on the sequence). In addition, another extreme case is that the specified mutation sites are multiple sites or even all sites, or the specified amino acid types after mutation are multiple types or all other 19 types. In this case, it is necessary to traverse all mutation possibilities to form a maximum of 19*N mutated sequences.

[0060] Those skilled in the art should understand that in subsequent steps, it is necessary to extract features from the sequences before and after mutation respectively, and finally splice them into features about the sequences. Therefore, no matter what the protein sequence information initially obtained is, when executing step S110, it is necessary to form the sequences before and after mutation.

[0061] Specifically, in step S110, the amino acid sequence information of the protein sample to be predicted before mutation and the amino acid sequence information after mutation are respectively subjected to feature extraction using the pre-trained model to obtain the sequence feature information before mutation and the sequence feature information after mutation. Then, still in step S110, the sequence feature information before mutation and the sequence feature information after mutation are spliced ​​to obtain the sequence feature of the protein sample to be predicted.

[0062] In the field of natural language processing research, with the continuous increase in computer computing power, more and more pre-trained models (PTMs) of general language representation have gradually emerged. This is very helpful for downstream pre-processing tasks. It can learn the knowledge of the language itself from massive unlabeled data, and then fine-tune on a small amount of labeled data, so that downstream tasks can better learn the characteristics of the language itself and the knowledge of specific tasks. By training on protein or nucleic acid sequence corpora, pre-trained models that can be used in the biological field can be obtained. The existing models trained on protein sequence corpora mainly include ESM-1b, UniRef, ProteinBert, TAPE, ProtGPT2, ProtTXL, ProtBert, ProtXLNet, ProtAlbert, ProtElectra, ProtT5-XL and ProtT5-XXL. Although it is also possible to obtain pre-trained models by training on protein sequence corpora, the performance of self-trained models is usually limited by cost and time issues. It is usually not possible to reach the level of the above models.

[0063] As mentioned above, the pre-trained model is essentially a natural language model. In the biological field, a pre-trained model that is improved and optimized for the characteristics and uses of the biological field is generally used. The pre-trained model has learned biological field-related knowledge from a large amount of unlabeled sequence data in the existing protein corpus, and can be used to extract mutation-related features from the protein sequence information of the training sample; then based on the extracted mutation-related features, training is performed, and only a small amount of labeled training samples are needed to optimize the parameters of the model. For example, the pre-trained model described in the present disclosure can be implemented using a combination of one or more of the models mentioned above and listed again below: ESM-1b, UniRef, ProteinBert, TAPE, ProtGPT2, ProtTXL, ProtBert, ProtXLNet, ProtAlbert, ProtElectra, ProtT5-XL and ProtT5-XXL. Preferably, the pre-trained model can be ProtT5-XL or ProtT5-XXL. More preferably, the pre-trained model is ProtT5-XL. That is, the pre-trained model used here is preferably a T5 (ie, Transfer Text-to-Text Transformer) pre-trained model based on the latest natural language processing method.

[0064] The purpose of step S110 is to obtain the sequence characteristics of the protein sample to be predicted, while the purpose of step S120 to be described next is to obtain the structural characteristics of the protein sample to be predicted.

[0065] In step S120, the structural features of the protein sample to be predicted are obtained based on the three-dimensional structural information of the protein sample to be predicted.

[0066] In addition to obtaining relevant features through pre-trained models, relevant features can also be obtained by combining the three-dimensional structure information of proteins for supplementary information. The three-dimensional structure information can be obtained through the PDB database (https: / / www.rcsb.org / ) or through prediction software, such as the following prediction software (references or sources are in brackets):

[0067] AlphaFold2(https: / / alphafold.ebi.ac.uk / );

[0068] I-TASSER (https: / / zhanggroup.org / I-TASSER / );

[0069] RoseTTAFold (see Minkyung Baeketal., "Accurate prediction of protein structures and interactions using a three-track neural network", Science373, 871-876 (2021). DOI: 10.1126 / science.abj8754);

[0070] Modeller(https: / / salilab.org / modeller / );

[0071] Swiss-model (https: / / swissmodel.expasy.org / ) etc.

[0072] The available protein three-dimensional structural information usually includes secondary structure, interaction network, amino acid residue distance, physical environment, and energy information obtained through three-dimensional structure, etc. Among them, the interaction network is one of the most commonly used features in structure-based prediction methods because it contains the contact information between all residue pairs.

[0073] According to a preferred embodiment of the present disclosure, the three-dimensional structural information used is the interaction network. More specifically, it is based on the characteristics of the interaction network that the GNN, more specifically the prediction model of the graph convolutional network (GCN) is selected.

[0074] Those skilled in the art should understand that the present disclosure does not limit the order of step S110 and step S120. That is, the acquisition of sequence features and structural features can be performed in any order or simultaneously.

[0075] Finally, in step S130, based on the sequence features obtained in step S110 and the structural features obtained in step S120, a prediction model based on a graph neural network is used to perform prediction to determine the stability prediction result of the protein sample to be predicted after mutation.

[0076] The prediction model based on graph neural network is actually a classifier.

[0077] In step S130, the sequence-based features and structure-based features are placed in a classifier to predict a specific task. Classifiers are generally divided into machine learning methods and deep learning methods. Machine learning methods are generally commonly used: Linear Regression algorithm, Support Vector Machine (SVM) algorithm, Nearest Neighbor / k-Nearest Neighbor (KNN) algorithm, Logistic Regression (LR) algorithm, Decision Tree (Decision Tree) algorithm, K-Means algorithm, Random Forest (Random Forest) algorithm, Naive Bayes (Naive Bayes) algorithm, Gradient Boosting (Gradient Boosting) algorithm, Ensemble (Ensemble Learning) method. Deep learning methods generally include: Convolutional Neural Network (Convolutional Neural Network), Recurrent Neural Network (Recurrent Neural Network), Recursive Neural Network (Recursive Neural Network), Long Short-Term Memory Network (LSTM). Graph Neural Network (GNN) is a deep learning model based on graph structured data, which is used to process structured data of nodes and edges. Graph Convolutional Network (GCN) is a GNN model based on Convolutional Neural Network (CNN). It achieves effective representation and learning of node features by aggregating the information of neighboring nodes. Graph Attention Network (GAT) is a GNN model based on attention mechanism. It achieves adaptive aggregation of node features by learning the attention weights of neighboring nodes, and can better handle the relationship between different nodes. GraphSAGE (Graph Sampling and Aggregation, SageConv) is a GNN model based on sampling. It achieves effective learning and representation of node features by sampling and aggregating neighboring nodes. GraphSAGE can process large-scale graph data while ensuring the efficiency and accuracy of the model. Graph Isomorphism Network (GINConv) is a GNN model based on graph isomorphism. It achieves effective learning and representation of node features by aggregating the features of neighboring nodes and combining the aggregation results with the features of the node itself.Diffusion Convolutional Neural Network (DCNN) is a GNN model based on the diffusion process. It achieves effective learning and representation of node features by diffusing and aggregating node features. Multi-Graph Convolutional (MGC) can perform graph convolution on different types of graphs, merge them together, and make full use of the information of different types of graphs. MGC-F represents that the input adjacency matrix is ​​a first-order adjacency matrix, while MFC-S represents that the input adjacency matrix is ​​a second-order adjacency matrix. These methods have their own advantages and scope of application. You can choose the appropriate method according to the specific task and data characteristics.

[0078] In a preferred embodiment of the present disclosure, the graph neural network is a graph convolutional network.

[0079] In a more preferred embodiment of the present disclosure, the prediction model based on the graph neural network includes one of the following models: MGC model, MGC_F model, MGC_S model, GINConv model and SageConv model. For more details, please refer to the detailed description below.

[0080] For the method of constructing the prediction model, please refer to the detailed description below.

[0081] As mentioned above, the existing methods for predicting the effect of mutations on protein stability are mainly divided into structure-based and sequence-based methods. The input features generated by the natural language model only contain sequence information to achieve excellent performance, while the use of deep neural networks to extract structural features from the three-dimensional structure of proteins can also achieve good performance.

[0082] The difference between the technology provided by the present disclosure and the above-mentioned prior art is that the technology disclosed in the present disclosure introduces graph neural networks, more specifically graph convolutional networks, into the prediction of protein stability after mutation, and combines it with the mature protein natural language model (pre-training model) in the prior art to jointly determine the prediction results. In other words, the sequence features and structural features of the protein sample to be predicted are combined and used as the input of the prediction model based on the graph neural network, thereby outputting the prediction results of the stability of the protein after mutation. The prediction results determined in this way are better than the prediction results determined by the methods of the prior art.

[0083] Figure 2 A schematic diagram 200 is provided for the association between the method for predicting protein stability after mutation and the method for constructing a protein stability prediction model after mutation according to an embodiment of the present disclosure. Figure 2 In the schematic diagram 200 shown, the left side of the dotted line shows a method for building a model, and the right side of the dotted line shows a method for using the model for prediction.

[0084] Figure 2 The process on the right side of the dotted line is actually Figure 1 The method for predicting protein stability after mutation according to an embodiment of the present disclosure is shown. Figure 2 The process on the left side of the dashed line will be described in detail below (for example Figure 3 and its corresponding text description). Figure 2 It can be seen that both model construction and the actual prediction process require the use of two feature extraction steps, namely: the extraction of sequence features and structural features. For the training sample, it includes the three-dimensional structural information of the sample protein, the amino acid sequence information of the sample protein before and after mutation, and the mutation stability label. As introduced later, the mutation stability label can be an unstable mutation and a stable mutation, where the unstable mutation includes a decrease in stability or a loss of stability, and the stable mutation includes an increase in stability or a flat stability. Therefore, in the process of model construction, the sequence information in the training sample is used to extract sequence features. Structural features are extracted through three-dimensional structural information. The GNN-based prediction model, i.e., the classifier, is fully trained using sequence features and structural features as input data and mutation stability labels in the training sample as output data, thereby obtaining the final GNN-based protein post-mutation stability prediction model. For the constructed prediction model, it can be put into the actual work of protein stability prediction after mutation, i.e. Figure 2 The process on the right side of the dotted line. In addition, it should be noted that Figure 2 In the figure, for sequence information and sequence features, hollow arrows are used to represent their signal transmission process; for three-dimensional structure information and structure features, different solid arrows are used to represent their signal transmission process; in addition, other signal flows and control flows are also represented by solid arrows.

[0085] The following is a detailed description of the process of the method for constructing a protein stability prediction model after mutation according to an embodiment of the present disclosure.

[0086] Figure 3 3 is a flow chart of a method 300 for constructing a protein stability prediction model after mutation used in the present disclosure.

[0087] Those skilled in the art should know that Figure 3 As shown in , in step S310, a training set should first be obtained, and the training set includes information of multiple training samples. Specifically, the information of each training sample includes the three-dimensional structure information of the sample protein, the amino acid sequence information of the sample protein before and after mutation, and the mutation stability label.

[0088] Artificial intelligence (AI) algorithms rely heavily on data. The quantity and quality of the initial data determine the generalization performance of the trained model, which is most directly reflected in the prediction accuracy of the model. Insufficient sample quantity or low quality of the data set will lead to overfitting or underfitting problems in the model. To solve this problem, the most effective way is to obtain more experimental data as a training set. The main source of experimental data is public databases. For the task of predicting the effect of mutations on proteins based on sequence-based methods, the data set can be inverted to double the original size by inverting the data in the database. Inversion means that under the same environmental conditions, the change in folding free energy (ΔΔG) is reversely symmetric when the free energy is mutated. More precisely, if protein B is a mutant of protein A, ΔΔG (A→B) = -ΔΔG (B→A), which conforms to the law of conservation of energy in physics. According to the above definition, the original data can be merged with the inverted data to double the original data volume. By inverting this method, the original data set is expanded and balanced at the same time, making the final prediction model more robust.

[0089] It should be noted that this method of expanding the dataset by reversing the original dataset is only implemented for the training set.

[0090] In the specific embodiments below, the acquisition of the training data set will be described in more detail.

[0091] Mutation stability labels can be used to reflect the trend of the influence of amino acid mutations on the thermal stability of proteins. In some embodiments, mutation stability labels can be determined based on the folding free energy change value (ΔΔG) of the mutant of the protein. The folding free energy change value (ΔΔG) caused by the mutation is closely related to the thermal stability of the mutant, wherein ΔΔG = folding free energy of the protein before mutation - folding free energy of the protein after mutation; ΔΔG less than 0 indicates that the stability of the protein after mutation is improved, and vice versa, it indicates that the mutation will cause a decrease in protein stability. Exemplarily, mutation stability labels can be unstable mutations and stable mutations, wherein unstable mutations include decreased stability or loss of stability, and stable mutations include improved stability or flat stability. It can be understood that using mutation stability information as a label for training samples allows the prediction model obtained through training to output prediction results related to the effect of amino acid mutations on protein thermal stability.

[0092] In step S320, based on the amino acid sequence information of the sample protein before and after mutation, feature extraction is performed through a pre-training model to obtain sequence features of the multiple training samples.

[0093] Specifically, in step S320, the amino acid sequence information of the sample protein before and after mutation is respectively subjected to feature extraction using the pre-training model to obtain sequence feature information before mutation and sequence feature information after mutation. Then, still in step S320, the sequence feature information before mutation and the sequence feature information after mutation are concatenated to obtain sequence features of the plurality of training samples.

[0094] As mentioned above, the pre-trained model is implemented by a combination of one or more of the following models: ESM-1b, UniRef, ProteinBert, TAPE, ProtGPT2, ProtTXL, ProtBert, ProtXLNet, ProtAlbert, ProtElectra, ProtT5-XL and ProtT5-XXL. Preferably, the pre-trained model is ProtT5-XL or ProtT5-XXL. More preferably, the pre-trained model is ProtT5-XL. That is, the pre-trained model used here is preferably a T5 (i.e., Transfer Text-to-Text Transformer) pre-trained model based on the latest natural language processing method.

[0095] The purpose of step S320 is to obtain the sequence characteristics of the training sample, while the purpose of step S330 to be described next is to obtain the structural characteristics of the training sample.

[0096] In step S330, structural features of a plurality of training samples are obtained based on the three-dimensional structural information of the sample protein.

[0097] As mentioned above, the three-dimensional structure information of the sample protein can be obtained through at least one of the following databases or prediction software: PDB database, AlphaFold2, I-TASSER, RoseTTAFold, Modeller and Swiss-model, etc.

[0098] The available protein three-dimensional structural information usually includes secondary structure, interaction network, amino acid residue distance, physical environment, and energy information obtained through three-dimensional structure, etc. Among them, the interaction network is one of the most commonly used features in structure-based prediction methods because it contains the contact information between all residue pairs.

[0099] According to a preferred embodiment of the present disclosure, the three-dimensional structural information used here is an interaction network. More specifically, it is based on the characteristics of the interaction network that GNN is selected, more specifically, the prediction model of the graph convolutional network (GCN).

[0100] Those skilled in the art should understand that the present disclosure does not limit the order of step S320 and step S330. That is, the acquisition of sequence features and structural features can be performed in any order or simultaneously.

[0101] In step S340, the graph neural network is trained based on the sequence features, the structural features, and the mutation stability labels to obtain a prediction model based on the graph neural network.

[0102] As mentioned earlier, the prediction model based on graph neural network is actually a classifier.

[0103] In a preferred embodiment of the present disclosure, the graph neural network is a graph convolutional network (GCN).

[0104] In a more preferred embodiment of the present disclosure, the prediction model based on the graph neural network includes one of the following models: MGC model, MGC_F model, MGC_S model, GINConv model and SageConv model. For more details, please refer to the detailed description below.

[0105] For the method of constructing the prediction model, please refer to the detailed description in the embodiments below.

[0106] In addition, it should be noted that: for sequence features, by placing the processed protein stability change dataset into the pre-trained model, the pre-trained model can be migrated to the new task through transfer learning. In transfer learning, there are two common application methods: feature extraction and fine-tuning. In feature extraction, after the network structure is pre-trained, a simple classifier can be modified or added, and the pre-trained network on the original task can be used as a feature extractor for another target task. Only the last added classifier parameters are re-learned, and the pre-trained network parameters will not be modified or frozen. The preferred embodiment of the present disclosure adopts a transfer learning method of feature extraction.

[0107] During the model training process, sequence features and structural features are used as input data of the GNN-based prediction model, and mutation stability labels are used as output targets of the prediction model to fully train the model. As the model parameters are continuously optimized, the output data of the prediction model will be consistent with the prediction results represented by the mutation stability labels.

[0108] More specifically, in some embodiments of the present disclosure, the prediction model (classifier) ​​can be supervised trained based on the sequence features, structural features and mutation stability labels of multiple training samples to obtain a trained classifier. Supervised training is the process of adjusting the parameters of the classifier using a set of samples of known categories to achieve the required performance. In some embodiments, a loss function, such as a cross entropy loss function, can be constructed to reflect the difference between the post-mutation stability prediction result predicted by the classifier and the actual mutation stability label of the training sample, and then the parameters of the classifier are adjusted according to the difference to obtain the model optimization parameters of the classifier. When the loss function meets the predetermined conditions, such as the loss function converges, or the loss function value is less than the preset value, the classifier training is completed. In some embodiments, the classifier parameters can be adjusted using a parameter search algorithm, for example, a search algorithm such as a grid search, a Bayesian search, etc. In some embodiments, supervised training is the process of training a prediction model based on a stability label.

[0109] It will be understood by those skilled in the art that by, for example Figure 3 The prediction model obtained by this method can be used Figure 1 The prediction method shown in .

[0110] Specific example embodiments will be given below.

[0111] Examples Methods and Materials

[0112] 1. Training Dataset

[0113] In order to overcome the problem of insufficient samples in the data sets used in previous methods, the recently published MPTherm (https: / / www.iitm.ac.in / bioinfo / mptherm / ) (see Kulandaisamy, A., R. Sakthivel and M. Michael Gromiha, 2021, "MPTherm: Database for Membrane Protein Thermodynamics for Understanding Folding and Stability", Briefings in Bioinformatics 22(2):2119–25, the entire contents of which are hereby incorporated into the present disclosure by reference and become part of the present disclosure), ProThermDB (https: / / web.iitm.ac.in / bioinfo2 / prothermdb / index.html) (see Nikam, Rahul et al., 2021, "ProThermDB: Thermodynamic Database for Proteins and Mutants Revisited after 15Years", Nucleic Acids Research 49(D1):D420–24, the entire contents of which are hereby incorporated by reference into the present disclosure and become part of the content of the present disclosure), ThermoMutDB (http: / / biosig.unimelb.edu.au / thermomutdb / ) (see Xavier, Joicymara S et al., 2021, “ThermoMutDB: A Thermodynamic Database for Missense Mutations”, Nucleic Acids Research 49(D1):D475–79, the entire contents of which are hereby incorporated by reference into the present disclosure and become part of the content of the present disclosure) and FireProtDB (https: / / loschmidt.chemi.muni.cz / fireprotdb / ) (Stourac, Jan et al., 2021, "FireProtDB: Database of Manually Curated Protein Stability Data", Nucleic Acids Research 49(D1):D319–24, the entire contents of which are incorporated into the present disclosure by reference, making them part of the present disclosure) 5917 experimental samples containing thermal stability changes were re-screened from the experimental database, including 1451 samples of stable mutations and 4466 samples of unstable mutations. The specific screening criteria are as follows:.

[0114] 1) Remove all samples that do not contain the difference in folding free energy (ΔΔG) between the wild-type and mutant proteins

[0115] 2) Remove all samples with mutation information other than single-point mutations

[0116] 3) Remove duplicate samples

[0117] 4) Reverse and merge the data

[0118] The dataset used to train the model contains 11,834 samples.

[0119] 2. Test Dataset

[0120] 2.1Tm141

[0121] The source of the test data is consistent with the training set, and the screening criteria are as follows:

[0122] 1) Remove all experimental Tm changes (ΔTm) that do not include those before and after the mutation

[0123] 2) Remove all samples with mutation information other than single-point mutations

[0124] 3) Remove duplicate samples

[0125] 4) |Δtm| is greater than or equal to 15 degrees Celsius. The data set used to test the model contains 141 experimental data. It should be noted that there are no repeated samples between the test data set and the training set.

[0126] 2.2S669

[0127] In order to strictly test the difference between the method disclosed in the present invention and other methods, a test data set S669 was introduced, which contains 669 single-point mutations. When testing this data set, duplicate data were removed from the training set. This test data set is widely used in the following literature: Pancotti, Corrado et al., 2022, "Predicting Protein Stability Change supon Single-Point Mutation: A Thorough Comparison of the Available Tools on a New Dataset", Briefings in Bioinformatics 23 (2): bbab555. By citing here, the entire contents of the above-mentioned documents are incorporated into the present disclosure and become part of the contents of the present disclosure.

[0128] 3. Pre-trained Model

[0129] The ProtT5-XL (ProtT5-XL-U50 / ProtT5-XL-UniRef50) model used in this method was released by Ahmed Elnaggar et al. in 2021. The T5 model was first trained on the BFD dataset, and then the model was fine-tuned and optimized using the UniRef50 dataset to obtain a pre-trained protein model. The model contains 24 layers of attention structure, each layer consisting of 32 self-attention heads and 1024 neurons. See Elnaggar, Ahmed et al., 2021, "ProtTrans: Towards Cracking the Language of Lifes Code Through Self-Supervised Deep Learning and High Performance Computing", IEEE Transactions on Pattern Analysis and Machine Intelligence: 1-1. By reference hereto, the entire contents of the above-mentioned documents are incorporated into the present disclosure to make them part of the contents of the present disclosure.

[0130] 4. Graph Convolutional Neural Network Model Construction

[0131] This preferred embodiment is based on variants of graph convolutional networks (GCN): MGC, MGC_F, MGC_S, GINConv and SageConv. The difference between the MGC method and GCN is the combination of different graph convolutional layers with different propagation rules. Compared with other graph neural network methods, the performance is excellent (see Vladimir et al., 2021, “Structure-Based Protein Function Prediction Using Graph Convolutional Networks”, Nature Communications 12(1): 3168. The entire contents of the above-mentioned document are hereby incorporated into the present disclosure by reference, making it a part of the contents of the present disclosure).

[0132] 4.1MGC Model

[0133] 4.1.1 Adjacency Matrix

[0134] The three-dimensional structure of each protein was obtained through prediction software such as PDB database, AlphaFold2, I-TASSER, RoseTTAFold, Modeller and Swiss-model. The distance between the amino acid residue pairs in the protein three-dimensional structure was calculated according to the coordinates of the obtained protein three-dimensional structure. When the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0 (itself is 0), which is the adjacency matrix of the entire protein.

[0135] In this embodiment, three amino acids are taken on each side of the mutation site, that is, 7 consecutive amino acids are taken as the residue sequence, and a 7*7 adjacency matrix is ​​intercepted from the adjacency matrix of the entire protein.

[0136] Figure 4 It is a schematic diagram of the adjacency matrix. As shown in the figure, the adjacency matrix of length 100 amino acids, where the mutation site is assumed to be residue 5, and the box is the 7*7 adjacency matrix of the residue.

[0137] 4.1.2 Node feature matrix

[0138] After a protein sequence information is input into the pre-trained model, the model first encodes each amino acid in the protein sequence through an encoder. Each amino acid will correspond to a vector of length 1024 after encoding, that is, a protein sequence of length N will obtain a vector of length N*1024. The model has a 24-layer structure. In order to obtain the most information about the mutation, the last layer can be selected as the output, and the 1024-dimensional feature vectors corresponding to the mutation positions of the wild type and mutant types are extracted respectively, and the feature vectors of the wild type (i.e. before mutation) and mutant type (i.e. after mutation) are spliced ​​into 2048 dimensions in sequence. With 7 as the sliding window size and the mutation site as the center, the three amino acid residues before and after are spliced, and finally a 7*2048-dimensional feature is obtained as the node feature matrix, with the following structure. Figure 5 shown.

[0139] 4.1.3 Model Construction

[0140] The matrices generated in 4.1.1 and 4.1.2 are input into the graph convolutional neural network as structural features and sequence features respectively. Through matrix multiplication and aggregation operations, each layer of the graph network structure updates the node features into a 512-dimensional feature vector, and the node features output by the three-layer graph network are merged into a 7*1536 (1536=512*3)-dimensional feature vector through the residual network structure. Then the average pooling operation is used to compress the updated feature matrix into a 1*1536-dimensional feature vector to represent the feature representation of the 7 amino acid residue sequences. The specific parameters are: the output dimensions of the three-layer graph convolution are 512, 512, 512 respectively; the number of neurons in the fully connected layer is 256 and 2; the optimizer uses Adam; the loss function is binary cross entropy loss; 'relu' is used as the activation function; 'softmax' is used as the output layer classification function; the l2 regularization term coefficient is 0.0001, the dropout rate is 0.5; the learning rate is 5e-05. Those skilled in the art should understand that the average pooling operation here can also be replaced by any pooling operation such as maximum pooling or K-max pooling.

[0141] 4.2MGC_F model construction

[0142] 4.2.1 First-order adjacency matrix

[0143] According to 4.1.1, the adjacency matrix of the entire protein is obtained, and then all amino acids whose spatial distance from the mutation site is less than 10 angstroms are selected with the mutation site residue as the center to obtain the adjacency matrix composed of these amino acids.

[0144] 4.2.2 Node feature matrix

[0145] After a protein sequence is input into the pre-trained model, the model first encodes each amino acid in the protein sequence through an encoder. Each amino acid is encoded into a vector of length 1024, that is, a protein sequence of length N will obtain a vector of length N*1024. The model has a 24-layer structure. In order to obtain the most information about mutations, this embodiment selects the last layer as the output, and simultaneously extracts the 1024-dimensional feature vectors of the wild type and mutant corresponding to the mutation position, and splices the wild type and mutant feature vectors into 2048 dimensions in turn, and finally obtains N f *2048-dimensional features as node feature matrix, where N f is the number of corresponding amino acids obtained from 4.2.1. The order is the same. The number of amino acids obtained from different proteins may be different. The maximum N f As a benchmark, fill in the missing 0 so that all proteins can obtain a matrix of the same size.

[0146] 4.2.3 Model Construction

[0147] The matrices generated in 4.2.1 and 4.2.2 are input into the graph convolutional neural network as structural features and sequence features respectively. Through matrix multiplication and aggregation operations, each layer of the graph network structure updates the node features into a 512-dimensional feature vector, and the node features output by the three-layer graph network are merged into N through the residual network structure. f *1536 (1536 = 512 * 3) dimensional feature vector. Then the updated feature matrix is ​​compressed into a 1 * 1536 dimensional feature vector represented by N by using the average pooling operation. f The specific parameters are: the output dimensions of the three-layer graph convolution are 512, 512, 512 respectively; the number of neurons in the fully connected layer is 128, 2; the optimizer uses Adam; the loss function is binary cross entropy loss; 'relu' is used as the activation function; 'softmax' is used as the output layer classification function; the l2 regularization term coefficient is 0.0005, the dropout rate is 0.5; and the learning rate is 0.0001. Those skilled in the art should understand that the average pooling operation here can also be replaced by any pooling operation such as maximum pooling or K-max pooling.

[0148] 4.3MGC_S model construction

[0149] 4.3.1 Second-order adjacency matrix

[0150] Construct a first-order adjacency matrix as shown in method 4.2.1, and then take the amino acids in the first-order adjacency matrix to calculate the first-order adjacency matrix again, that is, calculate the distance between the amino acid residue pairs again. When the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it means that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and it is recorded as 0 itself. After removing the duplicate amino acids in all matrices, a new matrix is ​​generated, namely the second-order adjacency matrix.

[0151] 4.3.2 Node feature matrix

[0152] After a protein sequence is input into the pre-trained model, the model first encodes each amino acid in the protein sequence through an encoder. Each amino acid is encoded into a vector of length 1024, that is, a protein sequence of length N will obtain a vector of length N*1024. The model has a 24-layer structure. In order to obtain the most information about mutations, this embodiment selects the last layer as the output, and simultaneously extracts the 1024-dimensional feature vectors of the wild type and mutant corresponding to the mutation position, and splices the wild type and mutant feature vectors into 2048 dimensions in turn, and finally obtains N s *2048-dimensional features as node feature matrix, where N sis the number of corresponding amino acids obtained from 4.3.1. The order is the same. The number of amino acids obtained from different proteins may be different. The maximum N s As a benchmark, fill in the missing 0 so that all proteins can obtain a matrix of the same size.

[0153] 4.3.3 Model Construction

[0154] The matrices generated in 4.3.1 and 4.3.2 are input into the graph convolutional neural network as structural features and sequence features respectively. Through matrix multiplication and aggregation operations, each layer of the graph network structure updates the node features into a 512-dimensional feature vector, and the node features output by the three-layer graph network are merged into N through the residual network structure. s *1536 (1536 = 512 * 3) dimensional feature vector. Then the updated feature matrix is ​​compressed into a 1 * 1536 dimensional feature vector represented by N by using the average pooling operation. s The characteristic representation of the amino acid residue sequence. The representation is input into the fully connected layer. The network consists of a hidden layer composed of 256 neurons and an output layer composed of 2 neurons, and finally outputs the probabilities predicted to be stable mutations and unstable mutations respectively. The specific parameters are: the output dimensions of the three-layer graph convolution are 512, 512, 512 respectively; the number of neurons in the fully connected layer is 256 and 2; the optimizer uses Adam; the loss function is binary cross entropy loss; 'relu' is used as the activation function; 'softmax' is used as the output layer classification function; the l2 regularization term coefficient is 0.0001, the dropout rate is 0.3; and the learning rate is 5e-05. Those skilled in the art should understand that the average pooling operation here can also be replaced by any pooling operation such as maximum pooling or K-max pooling.

[0155] 4.4 GINConv model construction

[0156] 4.4.1 First-order adjacency matrix

[0157] Same as 4.2.1 method.

[0158] 4.4.2 Node feature matrix

[0159] Same as 4.2.2 method.

[0160] 4.4.3 Model Construction

[0161] The matrices generated in 4.4.1 and 4.4.2 are input into the graph convolutional neural network as structural features and sequence features respectively. Through matrix multiplication and aggregation operations, each layer of the graph network structure updates the node features into a 64-dimensional feature vector, and the node features output by the three-layer graph network are merged into N through the residual network structure. f*192-dimensional feature vector. Then, the updated feature matrix is ​​compressed into a 1*192-dimensional feature vector represented by N by using the addition and pooling operation. f The specific parameters are as follows: the output dimensions of the three-layer graph convolution are 64, 64, 64 respectively; the number of neurons in each layer of the fully connected layer is 64, 64, 2; the aggregator_type of GINConv is 'max'; the initial init_eps value is 0.01; learn_eps is set to False; the optimizer is Adam; the loss function is binary cross entropy loss; 'relu' is used as the activation function; 'softmax' is used as the output layer classification function; and the learning rate is 0.01.

[0162] 4.5SageConv model construction

[0163] 4.5.1 First-order adjacency matrix

[0164] Same as 4.2.1 method.

[0165] 4.5.2 Node Feature Matrix

[0166] Same as 4.2.2 method.

[0167] 4.5.3 Model Construction

[0168] The matrices generated in 4.5.1 and 4.5.2 are input into the graph convolutional neural network as structural features and sequence features respectively. Through matrix multiplication and aggregation operations, each layer of the graph network structure updates the node features into a 128-dimensional feature vector, and the node features output by the three-layer graph network are merged into N through the residual network structure. f *384-dimensional feature vector. Then, the updated feature matrix is ​​compressed into a 1*384-dimensional feature vector represented by N by using the addition and pooling operation. f The specific parameters are as follows: the output dimensions of the three-layer graph convolution are 128, 128, 128 respectively; the number of neurons in each layer of the fully connected layer is 256, 256, 2; in the SAGEConv model, aggregator_type is 'gcn'; the optimizer is Adam; the loss function is binary cross entropy loss; 'relu' is used as the activation function; 'softmax' is used as the output layer classification function; the learning rate is 0.01 (SageConv(128,256,'gcn',0.01)).

[0169] Some summary is given below.

[0170] Figure 6 A schematic diagram showing the structure-based adjacency matrix construction. Figure 7 A schematic diagram showing the construction of a sequence-based node feature matrix. Figure 8 A schematic diagram of the prediction model framework is shown. Fig. 9 Yes Figure 6-8 A comprehensive schematic diagram.

[0171] According to the above embodiments, it can be seen that in the process of constructing a prediction model, the three-dimensional structural information of the sample protein can be obtained in the following manner: obtaining the coordinates of the three-dimensional structure of the sample protein; calculating the distance between the amino acid residue pairs in the three-dimensional structure of the sample protein according to the obtained coordinates of the three-dimensional structure; when the alpha carbon atom distance between the two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and itself is recorded as 0, and the adjacency matrix of the entire sample protein is obtained. Based on the three-dimensional structural information of the sample protein, obtaining the structural features of multiple training samples may include: taking the mutation site as the center, intercepting an adjacency matrix of a specified size from the adjacency matrix of the entire sample protein as the structural feature of the sample protein; or, according to the amino acids in the adjacency matrix of the entire protein sample to be predicted, calculating the distance between the amino acid residue pairs, when the alpha carbon atom distance between the two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and itself is recorded as 0, and the second-order adjacency matrix of the protein sample to be predicted is obtained. Based on sequence features, structural features and mutation stability labels, the graph neural network is trained to obtain a prediction model based on the graph neural network, which may further include: inputting the sequence features and the structural features into the graph neural network; updating and outputting the features using several layers of graph network structures through matrix multiplication and aggregation operations; merging the features output by several layers of graph network structures; compressing the features using pooling operations; classifying the features for stability according to the output layer classification function; and adjusting the network parameters of the graph neural network using the mutation stability label as the output target. The pooling operation mentioned here may be average pooling, maximum pooling or K-max pooling.

[0172] Accordingly, in the process of applying the prediction model to protein stability prediction, the three-dimensional structural information of the protein sample to be predicted is obtained in the following manner: obtaining the coordinates of the three-dimensional structure of the protein sample to be predicted; calculating the distance between the amino acid residue pairs in the three-dimensional structure of the protein according to the coordinates of the obtained three-dimensional structure; when the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and itself is recorded as 0, and the adjacency matrix of the entire protein sample to be predicted is obtained. Based on the three-dimensional structural information of the protein sample to be predicted, obtaining the structural features of the protein sample to be predicted includes: taking the mutation site as the center, intercepting an adjacency matrix of a specified size from the adjacency matrix of the entire protein sample to be predicted as the structural features of the protein sample to be predicted; or, according to the amino acids in the adjacency matrix of the entire protein sample to be predicted, calculating the distance between the amino acid residue pairs, when the alpha carbon atom distance between the two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and itself is recorded as 0, and the second-order adjacency matrix of the protein sample to be predicted is obtained. Based on the sequence features and the structural features, prediction is performed using a prediction model based on a graph neural network, and determining the stability prediction result of the protein sample to be predicted after mutation further includes: inputting the sequence features and the structural features into the prediction model based on a graph neural network; updating and outputting the features using a plurality of layers of graph network structures through matrix multiplication and aggregation operations; merging the features outputted by the plurality of layers of graph network structures respectively; compressing the features using an average pooling operation; and classifying the features for stability according to the output layer classification function. Similarly, the pooling operation mentioned here can be average pooling, maximum pooling, or K-max pooling.

[0173] 5. Performance indicators

[0174] This example uses the following indicators for evaluation, namely, area under the curve (AUC), precision, and recall. The specific definitions are as follows:

[0175] True Positive (TP)

[0176] False Positive (FP)

[0177] True Negative (TN)

[0178] False Negative (FN)

[0179] Among them, TP (True Positive) and TN (True Negative) represent the number of correctly classified positive samples and negative samples, FP (False Positive) and FN (False Negative) represent the number of incorrectly classified positive samples and negative samples, and P and N represent the number of positive samples and negative samples.

[0180] AUC is a popular parameter-independent metric used to describe binary classifiers. AUC is the area under the receiver operating characteristic (ROC) curve. The larger the value, the better the classifier performance, and the maximum value is 1.

[0181] Precision refers to the proportion of correctly predicted results among all the results predicted as positive samples. The precision calculation formula is: precision = TP / (TP+FP), where TP (True Positive) is the positive sample judged to be correctly classified, and FP (False Positive) is the positive sample judged to be incorrectly classified.

[0182] Recall refers to the ratio of the number of correctly predicted positive samples to the total number of positive samples. The calculation formula for recall is: recall = TP / (TP+FN), where FN (False Negative) is the negative sample that is judged to be misclassified.

[0183] Results Evaluation

[0184] 1. Model Performance

[0185] The model was trained on a self-designed balanced dataset containing 11,834 single-point mutation experimental ΔΔG values ​​of 328 different proteins from 4 databases. In the five-fold cross validation, the model performance is shown in Table 1 below.

[0186] method AUC Accuracy Recall MGC 0.87 0.80 0.79 MGC_F 0.83 0.76 0.75 MGC_S 0.82 0.73 0.76

[0187] Table 1: Five-fold cross-validation performance indicators of the prediction model disclosed in the present invention on the training set

[0188] 2. Comparison with other methods

[0189] In order to evaluate the difference between the method of the present invention and other prediction methods, the model of the present invention was compared with other existing excellent prediction methods on the experimental test set, including FoldX (see Guerois, Raphael, Jens Erik Nielsen and Luis Serrano, 2002, "Predicting Changes in the Stability of Proteins and Protein Complexes: A Study of More than 1000 Mutations", Journal of Molecular Biology 320(2):369-87, the entire contents of which are hereby incorporated into the present disclosure by reference and become part of the content of the present disclosure), Rosetta (Leman, Julia Koehler et al., 2020, "Macromolecular Modeling and Design in Rosetta: Recent Methods and Frameworks", Nature Methods 17(7):665-80, the entire contents of which are hereby incorporated into the present disclosure by reference and become part of the content of the present disclosure), Thermnet (Ofoeg Bu, Tochukwu C. et al., 2019, "Phyre Risk: A Dynamic Web Application to Bridge Genomics, Proteomics and 3D Structure Data to Guide Interpretation of Human Genetic Variants", Journal of Molecular Biology 431(13): 2460–66, the entire contents of which are hereby incorporated by reference into the present disclosure and become part of the content of the present disclosure), PremPS (Chen, Yuting et al., 2020, "PremPS: Predicting the Impact of Missense Mutation on Protein Stability", PLoS computational biology 16(12): e1008543, the entire contents of which are hereby incorporated by reference into the present disclosure and become part of the content of the present disclosure), I-Mutant (Capriotti, E., P. Fariselli and R. Casadio, 2005, "I-Mutant 2.0: Predicting Stability Change supon Mutation from the Protein Sequence or Structure", Nucleic Acids Research 33 (Web Server): W306–10, the entire contents of the above documents are hereby incorporated into the present disclosure by reference, making it a part of the content of the present disclosure) and DDGun (Montanucci, Ludovica et al., 2022, "DDGun: An Untrained Predictor of Protein Stability Change supon Amino Acid Variants", Nucleic Acids Research 50 (W1): W222–27, the entire contents of the above documents are hereby incorporated into the present disclosure by reference, making it a part of the content of the present disclosure). The samples in the experimental test set did not appear in the training set of this method and the above prediction method, and the results are shown in Table 2 below.

[0190] method AUC Accuracy Recall MGC 0.88 0.76 0.67 MGC_F 0.89 0.73 0.67 MGC_S 0.87 0.72 0.66 GINConv 0.86 0.71 0.62 SageConv 0.86 0.59 0.54 PremPS 0.88 0.58 0.29 DDGun 0.77 0.34 0.50 I-Mutant 0.86 0.60 0.50 FoldX 0.81 0.52 0.71 Rosetta 0.80 0.50 0.50 Thermnet 0.81 0.33 0.79

[0191] Table 2: Performance comparison of the model disclosed in this paper and other methods on the experimental test set Tm141

[0192] As can be seen from the above table, the graph-based method has a significant advantage in accuracy over the previous methods, and the recall rate remains high.

[0193] method AUC Accuracy Recall MGC 0.78 0.57 0.48 MGC_F 0.78 0.55 0.45 GINConv 0.80 0.57 0.58 SageConv 0.79 0.56 0.51 MGC_S 0.75 0.60 0.39 PremPS 0.64 0.43 0.23 DDGun 0.71 0.43 0.42 I-Mutant 0.66 0.48 0.21 FoldX 0.62 0.34 0.41 Rosetta 0.64 0.46 0.25 Thermnet 0.68 0.36 0.51

[0194] Table 3: Performance comparison of the model disclosed in this paper and other methods on the experimental test set S669

[0195] The results show that the performance of the method disclosed in the present invention on the test set is outstanding compared with other methods.

[0196] 3. Experimental Validation of the Prediction Model

[0197] In order to verify the practical application of the model disclosed in the experiment, additional verification was performed by predicting the experimental results already available in the literature. The results are shown in Table 4 below:

[0198]

[0199] Table 4: Test results of the method disclosed in the actual application

[0200] Prediction Process

[0201] Here, two cases are taken as examples to describe the process of using a prediction model to predict protein stability.

[0202] One case is that the input protein sequence information includes a single protein sequence and mutation site information. First, the protein mutation site preprocessing is performed based on the input single protein sequence and mutation site information. Specifically, the input protein sequence is recorded as the original sequence, and the mutated sequence is obtained based on the mutation site information. Then, the amino acid embedding layer is extracted, which is the extraction of sequence information mentioned above. That is, the original sequence is input into the natural language processing T5 pre-training model to read the embedded features; the mutated protein sequence is input into the natural language processing T5 pre-training model to read the embedded features; the amino acid embedding features are extracted based on the mutation site, and the features before and after the mutation are spliced ​​as the input of the subsequent model. Next is the construction of the graph network, which requires the protein embedding layer feature network and the protein residue interaction feature network. The above feature network is applied to the MGC prediction model to predict the changes in protein thermal stability before and after the amino acid mutation. Finally, the prediction results are output and displayed. For this example, the single point mutation prediction results can be output.

[0203] Another situation is that the input protein sequence information only includes a single protein sequence. First, according to the input single protein sequence, the protein mutation site preprocessing is performed. Specifically, the input protein sequence is recorded as the original sequence, and the amino acid mutation of each site of the protein sequence is recorded as the other 19 amino acids as the mutated sequence. Then, the amino acid embedding layer is extracted, which is the extraction of sequence information mentioned above. That is, the original sequence is input into the natural language processing T5 pre-training model to read the embedded features; the mutated protein sequence is input into the natural language processing T5 pre-training model to read the embedded features; the amino acid embedding features are extracted according to the mutation site, and the features before and after the mutation are spliced ​​as the input of the subsequent model. Next is the construction of the graph network, which requires the protein embedding layer feature network and the protein residue interaction feature network. The above feature network is applied to the MGC prediction model to predict the change in protein thermal stability before and after the amino acid mutation. Finally, the prediction results are output and displayed. For this example, the probability table of all stability mutations and / or the ranking of all stability mutation possibilities can be output as the final prediction result.

[0204] in conclusion

[0205] More stable protein mutants are necessary for academia and industry to find more applications for these proteins. This disclosure proposes a new method for discovering the effects of amino acid mutations on protein stability. The method disclosed in this disclosure is based on advanced natural language processing models combined with graph neural networks, can be applied to different biological fields, and has shown performance superior to earlier methods. This method has been successfully applied to predict stability changes caused by mutations in a variety of experiments, demonstrating the applicability of the method disclosed in this disclosure in protein engineering.

[0206] The reasons why the method of the present disclosure is superior to the prior art are discussed below.

[0207] From the above description we can see that:

[0208] 1. The method disclosed in the present invention is based on the existing latest natural language processing model ProtT5-XL, combined with structure-based features, and successfully applied to protein stability prediction, achieving results that are superior to existing technologies and verified by experiments.

[0209] 2. The method disclosed in this paper is based on graph neural network, which greatly improves the prediction performance of the model.

[0210] 3. The training set of the method disclosed in the present invention is collected from a variety of the latest databases, and the sample size is nearly twice the previous one.

[0211] 4. After the training data set used in the method disclosed in the present invention is inverted, the data set is expanded and the stable mutation and unstable mutation data are balanced, thereby ensuring the predictive robustness of the model.

[0212] Those skilled in the art should understand that based on the method for predicting protein stability after mutation disclosed in the present invention, a system for predicting protein stability after mutation can be developed. Fig.10 A schematic block diagram of a system for predicting protein stability after mutation according to the present disclosure is shown. Specifically, the system 1000 for predicting protein stability after mutation includes an acquisition module 1010 and a processing module 1020 .

[0213] The acquisition module 1010 is used to acquire the sequence information and three-dimensional structure information of the protein sample to be predicted.

[0214] The processing module 1020 is used to obtain the stability prediction result of the protein sample to be predicted after mutation by inputting the sequence information and three-dimensional structure information of the protein sample to be predicted. Fig.10As shown in, the processing module 1020 may further include: a sequence feature submodule 1021, which is used to extract features based on the sequence information of the protein sample to be predicted using a pre-trained model to obtain the sequence features of the protein sample to be predicted; a structural feature submodule 1022, which is used to obtain the structural features of the protein sample to be predicted based on the three-dimensional structure information of the protein sample to be predicted; a model prediction submodule 1023, which is used to predict based on the sequence features and the structural features using a prediction model based on a graph neural network to determine the stability prediction result of the protein sample to be predicted after mutation.

[0215] Those skilled in the art should understand that the above-mentioned system for predicting protein stability after mutation can be implemented by a computer. For example, the following operations can be implemented by executing a computer program: obtaining the sequence information and three-dimensional structure information of the protein sample to be predicted; obtaining the stability prediction result of the protein sample to be predicted after mutation by inputting the sequence information and three-dimensional structure information of the protein sample to be predicted, and the operation further includes the following sub-operations: based on the sequence information of the protein sample to be predicted, using a pre-trained model to perform feature extraction to obtain the sequence features of the protein sample to be predicted; based on the three-dimensional structure information of the protein sample to be predicted, obtaining the structural features of the protein sample to be predicted; based on the sequence features and the structural features, using a prediction model based on a graph neural network to perform prediction, and determining the stability prediction result of the protein sample to be predicted after mutation.

[0216] That is to say, although in the system for predicting protein stability based on graph neural network according to the present disclosure, the various operating functions corresponding to the prediction method are described as modules or sub-modules, those skilled in the art should understand that such modules or sub-modules may be circuit components or other physical components, or may not be composed of circuit components or other physical components, but may be functional modules constructed by computer programs.

[0217] In addition, it should be recognized by those skilled in the art that the method of the present disclosure can be implemented as a computer program. As described above in conjunction with the accompanying drawings, the method of the above embodiment is executed by one or more programs, and the instructions in the program cause the computer or processor to execute the algorithm described in conjunction with the accompanying drawings. These programs can be stored and provided to the computer or processor using various types of non-transient computer-readable media. Non-transient computer-readable media include various types of tangible storage media. Examples of non-transient computer-readable media include magnetic recording media (such as floppy disks, tapes, and hard disk drives), magneto-optical recording media (such as magneto-optical disks), CD-ROMs (compact disk read-only memories), CD-Rs, CD-R / Ws, and semiconductor memories (such as ROMs, PROMs (programmable ROMs), EPROMs (erasable PROMs), flash ROMs, and RAMs (random access memories)). Further, these programs can be provided to the computer using various types of transient computer-readable media. Examples of transient computer-readable media include electrical signals, optical signals, and electromagnetic waves. Transient computer-readable media can be used to provide programs to the computer through wired communication paths or wireless communication paths such as wires and optical fibers.

[0218] For example, according to one embodiment of the present disclosure, a non-transitory computer-readable storage medium may be provided for storing a computer program, wherein the computer program includes instructions that, when executed by a processor of an electronic device, enable the electronic device to implement the method for predicting stability after protein mutation as described above.

[0219] In addition, according to the content disclosed in the present disclosure, a computer system can also be provided, the computer system comprising: a processor; a memory; and a computer program. The computer program is stored in the memory and is configured to be executed by the processor. The computer program includes instructions for implementing the method for predicting protein stability after mutation described above.

[0220] On the other hand, according to one embodiment of the present disclosure, a non-transitory computer-readable storage medium may be provided for storing a computer program, wherein the computer program includes instructions, which, when executed by a processor of an electronic device, enable the electronic device to implement the method for constructing a protein stability prediction model after mutation as described above.

[0221] In addition, according to the content disclosed in the present disclosure, a computer system can also be provided, the computer system comprising: a processor; a memory; and a computer program. The computer program is stored in the memory and is configured to be executed by the processor. The computer program includes instructions for implementing the method for constructing the protein mutation stability prediction model described above.

[0222] The implementation methods of the present disclosure are not limited to the above-mentioned embodiments. Without departing from the spirit and scope of the present disclosure, ordinary technicians in this field can make various changes and improvements to the present disclosure in form and details, and these are considered to fall within the protection scope of the present disclosure.

Claims

1. A method for predicting protein stability after mutation, characterized in that: The method comprises: Based on the sequence information of the protein sample to be predicted, a pre-trained model is used to perform feature extraction to obtain the sequence features of the protein sample to be predicted; Based on the three-dimensional structure information of the protein sample to be predicted, obtaining the structural characteristics of the protein sample to be predicted; Based on the sequence features and the structural features, a prediction model based on a graph neural network is used to perform predictions to determine the stability prediction results of the protein sample to be predicted after mutation.

2. The method according to claim 1, characterized in that: The sequence information of the protein sample to be predicted includes the amino acid sequence information of the protein sample to be predicted before mutation and the amino acid sequence information of the protein sample to be predicted after mutation.

3. The method according to claim 1, characterized in that The sequence information of the protein sample to be predicted includes the amino acid sequence information of the protein sample to be predicted before mutation and the designated mutation information.

4. The method according to claim 3, characterized in that The method further comprises: obtaining the amino acid sequence information of the protein sample to be predicted after the mutation according to the amino acid sequence information of the protein sample to be predicted before the mutation and the designated mutation information.

5. The method according to claim 3, characterized in that: The specified mutation information includes a specified mutation site and / or a specified amino acid type after mutation.

6. The method according to claim 2 or 4, characterized in that: The method of extracting features based on the sequence information of the protein sample to be predicted using a pre-trained model to obtain the sequence features of the protein sample to be predicted further includes: Using the pre-trained model to extract features from the amino acid sequence information of the protein sample to be predicted before mutation and the amino acid sequence information after mutation, respectively, to obtain sequence feature information before mutation and sequence feature information after mutation; The sequence feature information before the mutation and the sequence feature information after the mutation are spliced ​​to obtain the sequence feature of the protein sample to be predicted.

7. The method according to claim 1, characterized in that The pre-trained model is implemented by a combination of one or more of the following models: ESM-1b, UniRef, ProteinBert, TAPE, ProtGPT2, ProtTXL, ProtBert, ProtXLNet, ProtAlbert, ProtElectra, ProtT5-XL and ProtT5-XXL; preferably, the pre-trained model is ProtT5-XL or ProtT5-XXL; more preferably, the pre-trained model is ProtT5-XL.

8. The method according to claim 1, characterized in that: The three-dimensional structure information of the protein sample to be predicted includes the three-dimensional structure information of the protein sample to be predicted obtained through at least one of the following databases or prediction software: PDB database, AlphaFold2, I-TASSER, RoseTTAFold, Modeller and Swiss-model.

9. The method according to claim 1, characterized in that: The three-dimensional structural information includes interaction network, secondary structure, amino acid residue distance or physical environment; preferably, the three-dimensional structural information includes interaction network.

10. The method according to claim 9, characterized in that The three-dimensional structure information of the protein sample to be predicted is obtained by: Obtaining the coordinates of the three-dimensional structure of the protein sample to be predicted; Calculate the distances between pairs of amino acid residues in the three-dimensional structure of the protein based on the coordinates of the acquired three-dimensional structure; When the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and the node itself is recorded as 0, thus obtaining the adjacency matrix of the entire protein sample to be predicted.

11. The method according to claim 10, characterized in that The step of obtaining the structural features of the protein sample to be predicted based on the three-dimensional structural information of the protein sample to be predicted includes: Taking the mutation site as the center, intercepting an adjacency matrix of a specified size from the adjacency matrix of the entire protein sample to be predicted as the structural feature of the protein sample to be predicted; or According to the amino acids in the adjacency matrix of the entire protein sample to be predicted, the distance between the amino acid residue pairs is calculated. When the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1; otherwise, it is 0, and it itself is recorded as 0, thus obtaining the second-order adjacency matrix of the protein sample to be predicted.

12. The method according to claim 1, characterized in that The prediction based on the sequence features and the structural features using a prediction model based on a graph neural network to determine the stability prediction result of the protein sample to be predicted after mutation further includes: Inputting the sequence features and the structural features into a prediction model based on a graph neural network; Through matrix multiplication and aggregation operations, the features are updated and output using several layers of graph network structures; Merge the features output by several layers of graph network structures; Use pooling operation to compress features; The features are classified for stability according to the output layer classification function.

13. The method according to claim 1, characterized in that The graph neural network-based prediction model is constructed through the following steps: Acquire a training set, wherein the training set includes information of a plurality of training samples, and the information of each training sample includes three-dimensional structure information of the sample protein, amino acid sequence information of the sample protein before and after mutation, and a mutation stability label; Based on the amino acid sequence information of the sample protein before and after mutation, feature extraction is performed through a pre-training model to obtain sequence features of the plurality of training samples; Based on the three-dimensional structural information of the sample protein, obtaining structural features of the plurality of training samples; Based on the sequence features, the structural features and the mutation stability labels, the graph neural network is trained to obtain the graph neural network-based prediction model.

14. The method according to claim 13, characterized in that The graph neural network-based prediction model is a classifier.

15. The method according to claim 13, characterized in that The graph neural network is a graph convolutional network.

16. The method according to claim 15, characterized in that The prediction model based on graph neural network includes one of the following models: MGC model, MGC_F model, MGC_S model, GINConv model and SageConv model.

17. The method according to claim 13, characterized in that The step of obtaining the training set includes: Obtain experimental data from a public database as a training set; The training set is doubled in size by inverting the experimental data in the public database.

18. A method for constructing a protein stability prediction model after mutation, characterized in that: The method comprises: Acquire a training set, wherein the training set includes information of a plurality of training samples, and the information of each training sample includes three-dimensional structure information of the sample protein, amino acid sequence information of the sample protein before and after mutation, and a mutation stability label; Based on the amino acid sequence information of the sample protein before and after mutation, feature extraction is performed through a pre-training model to obtain sequence features of the plurality of training samples; Based on the three-dimensional structural information of the sample protein, the structural features of multiple training samples are obtained; Based on the sequence features, the structural features and the mutation stability labels, a graph neural network is trained to obtain a prediction model based on the graph neural network.

19. The method according to claim 18, characterized in that The method of extracting features based on the amino acid sequence information of the sample protein before and after mutation by using a pre-training model to obtain sequence features of the plurality of training samples includes: The amino acid sequence information of the sample protein before and after mutation is respectively subjected to feature extraction using the pre-trained model to obtain sequence feature information before and after mutation; The sequence feature information before the mutation and the sequence feature information after the mutation are concatenated to obtain the sequence features of the multiple training samples.

20. The method according to claim 18, characterized in that The pre-trained model is implemented by a combination of one or more of the following models: ESM-1b, UniRef, ProteinBert, TAPE, ProtGPT2, ProtTXL, ProtBert, ProtXLNet, ProtAlbert, ProtElectra, ProtT5-XL and ProtT5-XXL; preferably, the pre-trained model is ProtT5-XL or ProtT5-XXL; more preferably, the pre-trained model is ProtT5-XL.

21. The method according to claim 18, characterized in that The three-dimensional structural information of the sample protein includes the three-dimensional structural information of the sample protein obtained through at least one of the following databases or prediction software: PDB database, AlphaFold2, I-TASSER, RoseTTAFold, Modeller and Swiss-model.

22. The method according to claim 18, characterized in that The three-dimensional structural information includes interaction network, secondary structure, amino acid residue distance or physical environment; preferably, the three-dimensional structural information includes interaction network.

23. The method according to claim 22, characterized in that The three-dimensional structural information of the sample protein is obtained by: Obtain the coordinates of the three-dimensional structure of the sample protein; Calculate the distances between pairs of amino acid residues in the three-dimensional structure of the sample protein based on the coordinates of the acquired three-dimensional structure; When the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1, otherwise it is 0, and the node itself is recorded as 0, and the adjacency matrix of the entire sample protein is obtained.

24. The method according to claim 23, characterized in that The step of obtaining the structural features of multiple training samples based on the three-dimensional structural information of the sample protein includes: Taking the mutation site as the center, extracting an adjacency matrix of a specified size from the adjacency matrix of the entire sample protein as the structural feature of the sample protein; or According to the amino acids in the adjacency matrix of the entire protein sample to be predicted, the distance between the amino acid residue pairs is calculated. When the alpha carbon atom distance between two residue pairs is less than 10 angstroms, it indicates that there is an edge relationship between the nodes, which is recorded as 1; otherwise, it is 0, and it itself is recorded as 0, thus obtaining the second-order adjacency matrix of the protein sample to be predicted.

25. The method according to claim 18, characterized in that The training of the graph neural network based on the sequence features, the structural features and the mutation stability labels to obtain a prediction model based on the graph neural network includes: Inputting the sequence features and the structural features into a graph neural network; Through matrix multiplication and aggregation operations, the features are updated and output using several layers of graph network structures; Merge the features output by several layers of graph network structures; Use pooling operation to compress features; Classify the stability of features according to the output layer classification function; The mutation stability label is used as an output target to adjust the network parameters of the graph neural network.

26. The method according to claim 18, characterized in that The graph neural network-based prediction model is a classifier.

27. The method according to claim 18, characterized in that The graph neural network is a graph convolutional network.

28. The method according to claim 27, characterized in that The prediction model based on graph neural network includes one of the following models: MGC model, MGC_F model, MGC_S model, GINConv model and SageConv model.

29. The method according to claim 18, characterized in that The step of obtaining the training set includes: Obtain experimental data from a public database as a training set; The training set is doubled in size by inverting the experimental data in the public database.

30. A system for predicting protein stability after mutation, characterized in that: The system comprises: An acquisition module is used to obtain the sequence information and three-dimensional structure information of the protein sample to be predicted; The processing module is used to obtain the stability prediction result of the protein sample to be predicted after mutation by inputting the sequence information and three-dimensional structure information of the protein sample to be predicted. The processing module further includes the following submodules: A sequence feature submodule, used to extract features based on the sequence information of the protein sample to be predicted using a pre-trained model to obtain the sequence features of the protein sample to be predicted; A structural feature submodule, used for obtaining the structural features of the protein sample to be predicted based on the three-dimensional structural information of the protein sample to be predicted; The model prediction submodule is used to predict the stability of the protein sample to be predicted after mutation based on the sequence features and the structural features using a prediction model based on a graph neural network.

31. A computer-implemented system for predicting protein stability after mutation, wherein the computer-implemented system implements the following operations by executing a computer program: Obtaining the sequence information and three-dimensional structure information of the protein sample to be predicted; By inputting the sequence information and three-dimensional structure information of the protein sample to be predicted, the stability prediction result of the protein sample to be predicted after mutation is obtained. This operation further includes the following sub-operations: Based on the sequence information of the protein sample to be predicted, using a pre-trained model to perform feature extraction to obtain sequence features of the protein sample to be predicted; Based on the three-dimensional structural information of the protein sample to be predicted, obtaining the structural characteristics of the protein sample to be predicted; Based on the sequence features and the structural features, a prediction model based on a graph neural network is used to perform predictions to determine the stability prediction results of the protein sample to be predicted after mutation.

32. A non-transitory computer-readable storage medium for storing a computer program, wherein the computer program includes instructions, which, when executed by a processor of an electronic device, causes the electronic device to implement the method for predicting protein stability after mutation as described in any one of claims 1-17 or the method for constructing a protein stability prediction model after mutation as described in any one of claims 18-29.

33. A computer system, comprising: processor; Memory; and A computer program, wherein the computer program is stored in the memory and is configured to be executed by the processor, the computer program comprising instructions for implementing the method for predicting protein stability after mutation as described in any one of claims 1 to 17 or the method for constructing a protein stability prediction model after mutation as described in any one of claims 18 to 29.

Images