Methods, devices, electronic equipment and media for predicting the relationship between diseases and genes

Abstract

This invention relates to the fields of digital healthcare and artificial intelligence, disclosing a method, device, electronic device, and storage medium for predicting the relationship between diseases and genes. The method includes: constructing a disease similarity network, a gene-disease relationship network, a first integrated network, and a second integrated network; extracting representation vectors from the first integrated network, the second integrated network, and the gene-disease relationship network to obtain a first representation vector, a second representation vector, and relationship node representation vectors; performing a weighted summation of the first representation vector, the second representation vector, and the relationship node representation vectors to obtain a first target representation vector and a second target representation vector; performing a weighted summation of the first target representation vector and the second target representation vector to obtain a target representation vector; and performing linear calculations on the target representation vector to obtain the probability that the gene data to be analyzed and the disease data to be analyzed are correlated in the target representation vector. This invention can improve the intelligence and accuracy of predicting the relationship between genes and diseases.

Description

Technical Field

[0001] This invention relates to the fields of digital healthcare and artificial intelligence, and in particular to a method, apparatus, electronic device, and readable storage medium for predicting the relationship between diseases and genes. Background Technology

[0002] With the advancement of medical technology, people have gradually discovered that there may be some connection between diseases and genes. However, due to the wide variety of diseases and the huge number of gene types, predicting the relationship between genes and diseases is a huge and extremely resource-intensive project. In addition, since human intervention plays a major role in the process of predicting the relationship between genes and diseases, it leads to large errors, resulting in low accuracy in predicting the relationship between genes and diseases. Summary of the Invention

[0003] This invention provides a method, apparatus, electronic device, and readable storage medium for predicting the relationship between diseases and genes, with the aim of improving the intelligence and accuracy of predicting the relationship between genes and diseases.

[0004] To achieve the above objectives, the present invention provides a method for predicting the relationship between diseases and genes, the method comprising:

[0005] Acquire gene data and disease data to be analyzed, construct a first gene regulatory network based on protein interaction data in the gene data to be analyzed, and construct a second gene regulatory network based on the relationship between transcription factors and messenger RNA in the gene data to be analyzed.

[0006] Calculate the semantic similarity between the disease data to be analyzed, and construct a disease similarity network based on the semantic similarity;

[0007] Extract the relationship between the gene data to be analyzed and the disease data to be analyzed, and construct a gene-disease relationship network based on the relationship;

[0008] The first gene regulation network and the disease similarity network are integrated to obtain a first integrated network. The representation vectors of each node in the first integrated network and the gene-disease relationship network are extracted to obtain a first representation vector and a relationship node representation vector. According to a preset first weight, the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector are weighted and summed to obtain a first target representation vector.

[0009] The second gene regulation network and the disease similarity network are integrated to obtain a second integrated network. The representation vectors of each node in the second integrated network are extracted to obtain a second representation vector. According to the preset second weight, the representation vectors representing similar genes and disease relationships in the second representation vector and the relationship node representation vectors are weighted and summed to obtain a second target representation vector.

[0010] According to the preset third weight, the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector are weighted and summed to obtain the target representation vector;

[0011] Linear calculations are performed on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed.

[0012] Optionally, the step of performing linear calculations on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed includes:

[0013] The target representation vector is linearly transformed using the first preset parameter, the second preset parameter, and the third preset parameter in the fully connected layer of the pre-trained prediction model to obtain the query vector, the key vector, and the numerical vector.

[0014] The similarity matrix is ​​obtained by multiplying the query vector with the transpose of the key vector.

[0015] The similarity matrix is ​​normalized to obtain a normalized matrix;

[0016] The normalization matrix is ​​calculated using a preset activation function to obtain the activation matrix;

[0017] The activation matrix is ​​multiplied by the numerical vector to obtain the probability that the gene data to be analyzed and the disease data to be analyzed contained in the target representation vector are related.

[0018] Optionally, calculating the semantic similarity between the disease data to be analyzed includes:

[0019] Obtain the disease name data from the disease data to be analyzed;

[0020] Extract the disease feature vector from the disease name data;

[0021] The semantic similarity between each disease feature vector is calculated using the cosine similarity formula.

[0022] Optionally, the step of calculating the semantic similarity between each disease feature vector using the cosine similarity calculation formula includes:

[0023] The semantic similarity between each disease feature vector is calculated using the following formula. :

[0024]

[0025] in, Represents any one of the disease feature vectors, Indicates except Any disease feature vector other than, Indicates except The transpose of any disease feature vector other than .

[0026] Optionally, the step of extracting the representation vectors of each node in the first integrated network and the gene-disease relationship network to obtain the first representation vector and the relationship node representation vector includes:

[0027] The feature vectors of the first integrated network and the gene-disease relationship network are extracted using the pre-trained prediction model to obtain the first feature vector and the gene-disease feature vector.

[0028] The pre-trained prediction model is used to perform position index encoding on the first feature vector and the gene disease feature vector respectively to obtain the first position encoding vector and the gene disease position encoding vector.

[0029] The first position encoding vector and the gene disease position encoding vector are combined with the corresponding first feature vector and the gene disease feature vector to obtain the first point encoding vector and the gene disease encoding vector, respectively.

[0030] The convolutional layers in the pre-trained prediction model are used to extract features from the first point encoding vector and the gene disease encoding vector to obtain the first representation vector and the relationship node representation vector.

[0031] Optionally, before extracting the feature vectors of the first integrated network and the gene-disease relationship network using the pre-trained prediction model to obtain the first feature vector and the gene-disease feature vector, the method further includes:

[0032] Acquire historical gene data, historical disease data, and the true relationship between historical diseases and genes; extract the relationship between the historical gene data and the historical disease data; and construct a historical gene-disease relationship network based on the relationship.

[0033] A historical gene regulatory network was constructed based on the aforementioned historical gene data;

[0034] Calculate the historical semantic similarity between the historical disease data, and construct a historical disease similarity network based on the historical semantic similarity;

[0035] A preliminary prediction model is constructed, integrating the historical gene regulation network and the historical disease similarity network to obtain a historical integrated network. The representation vectors of each node in the historical integrated network and the historical gene-disease relationship network are extracted to obtain the historical node representation vector and the historical relationship node representation vector. Based on the preset historical weights, the representation vectors representing similar genes and disease relationships in the historical node representation vector and the historical relationship node representation vector are weighted and summed to obtain the historical target node representation vector.

[0036] The pooling layer in the preliminary prediction model is used to perform feature dimensionality reduction on the historical target node representation vector to obtain the historical dimensionality-reduced node representation vector.

[0037] Linear calculation is performed on the historical dimensionality reduction node representation vector to obtain the relationship probability of each node in the historical gene-disease relationship network. Based on the true relationship between the historical disease and the gene, the loss value of the relationship probability is calculated using a preset loss function.

[0038] The steps are as follows: adjusting the parameters of the preliminary prediction model based on the loss value, returning to the step of encoding the nodes in the historical gene-disease relationship network using the encoding layer in the preliminary prediction model to obtain the node encoding matrix, iterating and updating the preliminary prediction model until the loss value is less than a preset threshold, and obtaining the pre-trained prediction model.

[0039] Optionally, the step of weighted summing of the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector according to a preset first weight, to obtain the first target representation vector, includes:

[0040] The similarity between the first representation vector and the relationship node representation vector is calculated using cosine similarity, and the representation vectors with similarity higher than a preset threshold between the first representation vector and the relationship node representation vector are used as the representation vectors of similar genes and disease relationships.

[0041] Based on the representation vectors of the similar genes and disease relationships and the preset first weights, the first target representation vector is calculated using the following formula. :

[0042]

[0043] in, This indicates the weight percentage of the first representation vector in the preset first weight. This represents the representation vector in the first representation vector that represents the relationship between similar genes and diseases. This indicates the weight percentage of the relation node representation vector in the preset first weight. This represents the representation vector in the relationship node representation vector that represents the relationship between similar genes and diseases in the first representation vector.

[0044] To address the above problems, the present invention also provides a disease-gene relationship prediction device, the device comprising:

[0045] A network construction module is used to acquire gene data and disease data to be analyzed, construct a first gene regulation network based on protein interaction data in the gene data to be analyzed, construct a second gene regulation network based on the relationship between transcription factors and messenger RNA in the gene data to be analyzed, calculate the semantic similarity between the disease data to be analyzed, construct a disease similarity network based on the semantic similarity, extract the relationship between the gene data to be analyzed and the disease data to be analyzed, and construct a gene-disease relationship network based on the relationship.

[0046] The representation vector calculation module is used to integrate the first gene regulation network and the disease similarity network to obtain a first integrated network, extract the representation vectors of each node in the first integrated network and the gene-disease relationship network to obtain a first representation vector and a relationship node representation vector, and perform a weighted summation of the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector according to a preset first weight to obtain a first target representation vector. The module then integrates the second gene regulation network and the disease similarity network to obtain a second integrated network, extracts the representation vectors of each node in the second integrated network to obtain a second representation vector, and performs a weighted summation of the representation vectors representing similar genes and disease relationships in the second representation vector and the relationship node representation vector according to a preset second weight to obtain a second target representation vector. Finally, it performs a weighted summation of the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector according to a preset third weight to obtain a target representation vector.

[0047] The relevant probability calculation module is used to perform linear calculations on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed.

[0048] To address the above problems, the present invention also provides an electronic device, the electronic device comprising:

[0049] Memory, storing at least one computer program; and

[0050] The processor executes the computer program stored in the memory to implement the disease-gene relationship prediction method described above.

[0051] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one computer program, which is executed by a processor in an electronic device to implement the disease-gene relationship prediction method described above.

[0052] This invention acquires preset gene and disease data to be analyzed, and constructs a first gene regulation network, a second gene regulation network, a disease similarity network, and a gene-disease relationship network based on the gene and disease data. This reduces the risk of low error tolerance in predictions using single data, making the prediction results more convincing. Secondly, it integrates different gene and disease data to be analyzed, and extracts the representation vectors of the integrated data and the representation vectors of each node in the gene-disease relationship network, thereby preventing overfitting caused by single data and improving the accuracy of gene-disease relationship prediction. Finally, it uses the fully connected layer in the pre-trained prediction model to perform linear calculations on the target representation vector, obtaining the probability that the gene and disease data contained in the target representation vector are related, and thus determining whether there is a correlation between the gene and disease data. This achieves disease-gene relationship prediction, reduces human intervention, improves the intelligence level of predicting gene-disease relationships, and reduces resource consumption. Therefore, the disease-gene relationship prediction method, device, equipment, and storage medium provided by this invention can improve the intelligence and accuracy of predicting gene-disease relationships. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating a method for predicting the relationship between diseases and genes according to an embodiment of the present invention.

[0054] Figures 2 to 3 This is a detailed flowchart of one step in a disease-gene relationship prediction method provided in an embodiment of the present invention;

[0055] Figure 4 This is a schematic diagram of a disease-gene relationship prediction device provided in an embodiment of the present invention;

[0056] Figure 5 A schematic diagram of the internal structure of an electronic device for implementing a method for predicting the relationship between diseases and genes, according to an embodiment of the present invention;

[0057] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0058] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0059] This invention provides a method for predicting the relationship between diseases and genes. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application embodiment: a server, a terminal, etc. In other words, the method for predicting the relationship between diseases and genes can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server can include an independent server, or it can include a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0060] Reference Figure 1 The flowchart shown is a disease-gene relationship prediction method provided in an embodiment of the present invention. In this embodiment, the disease-gene relationship prediction method includes:

[0061] S1. Obtain gene data and disease data to be analyzed, construct a first gene regulatory network based on protein interaction data in the gene data to be analyzed, and construct a second gene regulatory network based on the relationship between transcription factors and messenger RNA in the gene data to be analyzed.

[0062] In this embodiment of the invention, the gene data to be analyzed can be nucleotide sequences already discovered in organisms and typically stored in gene databases, such as gene data for Y chromosome microdeletion genes, cystic fibrosis transmembrane regulatory genes, and mitochondrial genes. The disease data to be analyzed can be biological disease data already discovered in humans and typically stored in disease databases, such as disease data for chronic bronchitis, mycoplasma pneumonia, and adult respiratory distress syndrome. The protein-protein interaction data can be each protein involved in various stages of life processes such as biological signal transduction, gene expression regulation, energy and material metabolism, and cell cycle regulation. The first gene regulatory network can be a network formed by the interactions between individual proteins in the protein-protein interaction data. The transcription factor can be a key molecule for gene expression regulation.

[0063] In an optional embodiment of the present invention, gene data and disease data to be analyzed can be obtained from publicly available gene and disease databases on the Internet. A first gene regulatory network is constructed based on the relationships between proteins in the gene data to be analyzed, and a second gene regulatory network is constructed based on the related transcription factors and messenger RNAs in the gene data to be analyzed. This completes the basic construction of the gene network, thereby providing gene data for model training. Furthermore, it allows for the prediction of the relationship between gene data and disease data in the first and second gene regulatory networks. Moreover, creating two gene regulatory networks reduces the risk of low error tolerance when using single data for prediction, making the prediction results more convincing.

[0064] S2. Calculate the semantic similarity between the disease data to be analyzed, and construct a disease similarity network based on the semantic similarity.

[0065] In this embodiment of the invention, the semantic similarity may be the similarity between the names of various diseases in the disease data to be analyzed.

[0066] This invention calculates the semantic similarity between the disease data to be analyzed, determines the similarity between each disease, thereby determining the possible relationships between each disease and ensuring the completion of the construction of the disease similarity network.

[0067] Furthermore, as an optional embodiment of the present invention, reference is made to... Figure 2 As shown, calculating the semantic similarity between the disease data to be analyzed includes steps S21-S23:

[0068] S21. Obtain the disease name data from the disease data to be analyzed;

[0069] S22. Extract the disease feature vector from the disease name data;

[0070] S23. Calculate the semantic similarity between each disease feature vector using the cosine similarity calculation formula.

[0071] In this embodiment of the invention, the disease name data can be the disease name of each disease.

[0072] In an optional embodiment of the present invention, since similar diseases often have a certain correlation in disease naming, the degree of correlation between each disease can be calculated from the disease name data in the disease data to be analyzed, thereby making it easier to predict the relationship between genes and diseases.

[0073] Furthermore, the calculation of semantic similarity between each disease feature vector using the cosine similarity formula includes:

[0074] The semantic similarity between each disease feature vector is calculated using the following formula. :

[0075]

[0076] in, Represents any one of the disease feature vectors, Indicates except Any disease feature vector other than, Indicates except The transpose of any disease feature vector other than .

[0077] In this embodiment of the invention, by calculating the semantic similarity between each disease feature vector, the association between each disease node in the disease similarity network is guaranteed, the probability of predicting the relationship between genes and diseases is increased, and thus the efficiency of predicting the relationship between genes and diseases is improved.

[0078] In this embodiment of the invention, a disease similarity network is constructed based on the semantic similarity to ensure that there is a correlation between the disease data in the disease similarity network. This increases the likelihood of predicting the relationship between genes and diseases, reduces the time spent blindly searching for similar diseases when a gene has an impact on a disease, and speeds up the efficiency of predicting the relationship between genes and diseases.

[0079] S3. Extract the relationship between the gene data to be analyzed and the disease data to be analyzed, and construct a gene-disease relationship network based on the relationship.

[0080] In this embodiment of the invention, the gene-disease relationship network can be constructed from the gene data to be analyzed and the disease data to be analyzed, including gene data and disease data that have positive or negative correlations.

[0081] In an optional embodiment of the present invention, a web crawler can be used to crawl known gene data and disease data with positive or negative correlations in the network. Then, the gene data and disease data with positive or negative correlations are matched with the gene data and disease data to be analyzed to obtain target gene data and target historical disease data with positive or negative correlations in the gene data and disease data to be analyzed. A relationship network is constructed based on the correlation between the target gene data and the target historical disease data to obtain the gene-disease relationship network. This ensures that there are real values ​​as references during the model training process, making the model results more accurate and improving the accuracy of predicting the relationship between genes and diseases.

[0082] S4. Integrate the first gene regulation network and the disease similarity network to obtain a first integrated network. Extract the representation vectors of each node in the first integrated network and the gene-disease relationship network to obtain a first representation vector and a relationship node representation vector. According to a preset first weight, perform a weighted summation on the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector to obtain a first target representation vector.

[0083] In this embodiment of the invention, the preset first weight can be the proportion weight of the first representation vector and the proportion weight of the relationship node representation vector, respectively. The similar gene and disease relationship can be the relationship between the same gene and similar diseases, the relationship between similar genes and the same disease, and the relationship between the same gene and the same disease.

[0084] In an optional embodiment of the present invention, gene-disease integrated data is obtained by splicing each gene data in the first gene regulation network with each disease data in the disease similarity network. All gene-disease integrated data are then integrated to obtain a first integrated network. This ensures that each gene data in the first gene regulation network is spliced ​​with each disease data in the disease similarity network at least once, thereby ensuring that gene data and disease data with positive or negative correlations are not missed, thus improving the accuracy of predicting the relationship between genes and diseases.

[0085] This invention extracts the representation vectors of each node in the first integrated network and the gene-disease relationship network to obtain the first representation vector and the relationship node representation vector, which facilitates the model's prediction of the relationship between genes and diseases, thereby reducing the degree of human intervention and further improving the intelligence level of predicting the relationship between genes and diseases.

[0086] Furthermore, as an optional embodiment of the present invention, reference is made to... Figure 3 As shown, the step of extracting the representation vectors of each node in the first integration network and the gene-disease relationship network to obtain the first representation vector and the relationship node representation vector includes steps S41-S44:

[0087] S41. Use the pre-trained prediction model to extract the feature vectors of the first integrated network and the gene-disease relationship network to obtain the first feature vector and the gene-disease feature vector.

[0088] S42. Using the pre-trained prediction model, the first feature vector and the gene disease feature vector are respectively encoded by position index to obtain the first position encoding vector and the gene disease position encoding vector.

[0089] S43. Combine the first position encoding vector and the gene disease position encoding vector with the corresponding first feature vector and the gene disease feature vector respectively to obtain the first point encoding vector and the gene disease encoding vector.

[0090] S44. Using the convolutional layers in the pre-trained prediction model, feature extraction is performed on the first point encoding vector and the gene disease encoding vector to obtain the first representation vector and the relationship node representation vector.

[0091] In this embodiment of the invention, the pre-trained prediction model can be a GCN (Graph Convolutional Network) model. The GCN model mainly studies the representation of graph nodes, graph edge structure prediction tasks, and graph classification. This invention is mainly used for predicting graph edge structure tasks.

[0092] According to a preset first weight, the first representation vector and the representation vectors representing similar genes and disease relationships in the relationship node representation vector are weighted and summed to obtain a first target representation vector. This neutralizes the known gene data and disease data with correlation in the gene-disease relationship network, thereby increasing the weight ratio of the same gene to similar diseases and improving the accuracy of predicting the relationship between diseases and genes.

[0093] In this embodiment of the invention, the preset first weight includes the weight of the first representation vector and the weight of the relationship node representation vector.

[0094] Furthermore, as an optional embodiment of the present invention, before extracting the feature vectors of the first integrated network and the gene-disease relationship network using the pre-trained prediction model to obtain the first feature vector and the gene-disease feature vector, the method further includes:

[0095] Acquire historical gene data, historical disease data, and the true relationship between historical diseases and genes; extract the relationship between the historical gene data and the historical disease data; and construct a historical gene-disease relationship network based on the relationship.

[0096] A historical gene regulatory network was constructed based on the aforementioned historical gene data;

[0097] Calculate the historical semantic similarity between the historical disease data, and construct a historical disease similarity network based on the historical semantic similarity;

[0098] A preliminary prediction model is constructed, integrating the historical gene regulation network and the historical disease similarity network to obtain a historical integrated network. The representation vectors of each node in the historical integrated network and the historical gene-disease relationship network are extracted to obtain the historical node representation vector and the historical relationship node representation vector. Based on the preset historical weights, the representation vectors representing similar genes and disease relationships in the historical node representation vector and the historical relationship node representation vector are weighted and summed to obtain the historical target node representation vector.

[0099] The pooling layer in the preliminary prediction model is used to perform feature dimensionality reduction on the historical target node representation vector to obtain the historical dimensionality-reduced node representation vector.

[0100] Linear calculation is performed on the historical dimensionality reduction node representation vector to obtain the relationship probability of each node in the historical gene-disease relationship network. Based on the true relationship between the historical disease and the gene, the loss value of the relationship probability is calculated using a preset loss function.

[0101] The parameters of the preliminary prediction model are adjusted according to the loss value, and the steps of using the pooling layer in the preliminary prediction model to perform feature dimensionality reduction on the historical target node representation vector to obtain the historical dimensionality-reduced node representation vector are returned. The preliminary prediction model is iteratively updated until the loss value is less than a preset threshold, and the pre-trained prediction model is obtained.

[0102] In this embodiment of the invention, the nodes in the gene-disease relationship network can be known gene nodes and disease nodes that have a correlation.

[0103] In an optional embodiment of the present invention, by training the constructed preliminary prediction model using known gene data and disease data with known correlations, the accuracy and precision of the pre-trained prediction model are ensured, thus laying the foundation for subsequent prediction of the probability of the relationship between genes and diseases. Further, as an optional embodiment of the present invention, the step of weighted summation of the representation vectors representing similar gene and disease relationships in the first representation vector and the relationship node representation vector according to a preset first weight, to obtain the first target representation vector, includes:

[0104] The similarity between the first representation vector and the relationship node representation vector is calculated using cosine similarity, and the representation vectors with similarity higher than a preset threshold between the first representation vector and the relationship node representation vector are used as the representation vectors of similar genes and disease relationships.

[0105] Based on the representation vectors of the similar genes and disease relationships and the preset first weights, the first target representation vector is calculated using the following formula. :

[0106]

[0107] in, This indicates the weight percentage of the first representation vector in the preset first weight. This represents the representation vector in the first representation vector that represents the relationship between similar genes and diseases. This indicates the weight percentage of the relation node representation vector in the preset first weight. This represents the representation vector in the relationship node representation vector that represents the relationship between similar genes and diseases in the first representation vector.

[0108] In an optional embodiment of the present invention, the weight proportion of the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector is calculated using the cosine similarity calculation formula and the weighted summation formula, thereby making the representation vectors of similar genes and disease relationships more accurate, and thus improving the accuracy of predicting the relationship between diseases and genes.

[0109] S5. Integrate the second gene regulation network and the disease similarity network to obtain a second integrated network. Extract the representation vector of each node in the second integrated network to obtain a second representation vector. According to the preset second weight, perform a weighted summation on the second representation vector and the representation vectors of the relationship nodes that represent similar genes and disease relationships to obtain a second target representation vector.

[0110] In this embodiment of the invention, the preset second weight can be the proportion weight of the second node representation vector and the proportion weight of the relationship node representation vector, respectively.

[0111] In an optional embodiment of the present invention, gene-disease integrated data is obtained by splicing each gene data in the second gene regulation network with each disease data in the disease similarity network. All gene-disease integrated data are then integrated to obtain a second integrated network. This ensures that each gene data in the second gene regulation network is spliced ​​with each disease data in the disease similarity network at least once, thereby ensuring that gene data and disease data with positive or negative correlations are not missed, thus improving the accuracy of predicting the relationship between genes and diseases.

[0112] This invention extracts the representation vectors of each node in the second integrated network to obtain a second representation vector, which facilitates the model's prediction of the relationship between genes and diseases, thereby reducing the degree of human intervention and further improving the intelligence level of predicting the relationship between genes and diseases.

[0113] Furthermore, the extraction of representation vectors from each node in the second integrated network to obtain the second representation vector is similar to the extraction of representation vectors from each node in the first integrated network and the gene-disease relationship network to obtain the first representation vector and the relationship node representation vector, so it will not be elaborated further. In this embodiment of the invention, based on a preset second weight, the representation vectors representing similar gene-disease relationships in the second representation vector and the relationship node representation vector are weighted and summed to obtain the second target representation vector. This neutralizes known gene and disease data with known associations in the gene-disease relationship network, thereby increasing the weight of the same gene for similar diseases and improving the accuracy of predicting the relationship between diseases and genes.

[0114] Furthermore, as an optional embodiment of the present invention, the step of weighting and summing the representation vectors representing similar genes and disease relationships in the second representation vector and the relationship node representation vector according to the preset second weight to obtain the second target representation vector is similar to the step of weighting and summing the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector according to the preset first weight to obtain the first target representation vector, so it will not be described in detail here.

[0115] S6. Based on the preset third weight, the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector are weighted and summed to obtain the target representation vector.

[0116] In this embodiment of the invention, the preset third weight can be the proportion weight of the first target representation vector and the proportion weight of the second target representation vector, respectively.

[0117] According to the embodiments of the present invention, the representation vectors representing the relationship between similar genes and diseases in the first target representation vector and the second target representation vector are weighted and summed according to a preset third weight to obtain the target representation vector, thereby preventing overfitting caused by a single data point and improving the accuracy of the relationship between genes and diseases.

[0118] Furthermore, in an optional embodiment of the present invention, the step of weighting and summing the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector according to a preset third weight to obtain the target representation vector is similar to the step of weighting and summing the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector according to a preset first weight to obtain the first target representation vector, so it will not be described in detail here.

[0119] S7. Perform linear calculations on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed.

[0120] This invention can use the activation function of the fully connected layer in the pre-trained prediction model to perform linear calculation on the target representation vector, thereby obtaining the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed, and thus determining whether there is a correlation between the gene data to be analyzed and the disease data to be analyzed. This realizes the prediction of the relationship between diseases and genes, reduces human intervention, improves the intelligence level of predicting the relationship between genes and diseases, and reduces resource consumption.

[0121] Further, as an optional embodiment of the present invention, the step of performing linear calculations on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed includes:

[0122] The target representation vector is linearly transformed using the first preset parameter, the second preset parameter, and the third preset parameter in the fully connected layer of the pre-trained prediction model to obtain the query vector, the key vector, and the numerical vector.

[0123] The similarity matrix is ​​obtained by multiplying the query vector with the transpose of the key vector.

[0124] The similarity matrix is ​​normalized to obtain a normalized matrix;

[0125] The normalization matrix is ​​calculated using a preset activation function to obtain the activation matrix;

[0126] The activation matrix is ​​multiplied by the numerical vector to obtain the probability that the gene data to be analyzed and the disease data to be analyzed contained in the target representation vector are related.

[0127] In this embodiment of the invention, the first preset parameter, the second preset parameter, and the third preset parameter can be parameter matrices obtained through multiple training and optimization processes. The preset activation function can be the softmax activation function.

[0128] In an optional embodiment of the present invention, by performing score normalization calculation on the similarity matrix, the matrix gradient becomes more stable and reliable, thereby improving the calculation efficiency of the probability that the gene data to be analyzed and the disease data to be analyzed contained in the target representation vector are related.

[0129] like Figure 4 The diagram shown is a functional block diagram of the disease-gene relationship prediction device of the present invention.

[0130] The disease-gene relationship prediction device 100 of the present invention can be installed in an electronic device. Depending on the functions implemented, the disease-gene relationship prediction device 100 may include a network construction module 101, a representation vector calculation module 102, and a related probability calculation module 103. The module mentioned in the present invention may also be referred to as a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0131] In this embodiment, the functions of each module / unit are as follows:

[0132] The network construction module 101 is used to acquire gene data to be analyzed and disease data to be analyzed, construct a first gene regulation network based on protein interaction data in the gene data to be analyzed, construct a second gene regulation network based on the relationship between transcription factors and messenger RNA in the gene data to be analyzed, calculate the semantic similarity between the disease data to be analyzed, construct a disease similarity network based on the semantic similarity, extract the relationship between the gene data to be analyzed and the disease data to be analyzed, and construct a gene-disease relationship network based on the relationship.

[0133] The representation vector calculation module 102 is used to integrate the first gene regulation network and the disease similarity network to obtain a first integrated network. It extracts the representation vectors of each node in the first integrated network and the gene-disease relationship network to obtain a first representation vector and a relationship node representation vector. Based on a preset first weight, it performs a weighted summation of the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vectors to obtain a first target representation vector. It then integrates the second gene regulation network and the disease similarity network to obtain a second integrated network. It extracts the representation vectors of each node in the second integrated network to obtain a second representation vector. Based on a preset second weight, it performs a weighted summation of the representation vectors representing similar genes and disease relationships in the second representation vector and the relationship node representation vectors to obtain a second target representation vector. Finally, based on a preset third weight, it performs a weighted summation of the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector to obtain a target representation vector.

[0134] The correlation probability calculation module 103 is used to perform linear calculations on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed.

[0135] like Figure 5 The diagram shown is a schematic representation of the electronic device used in this invention to predict the relationship between diseases and genes.

[0136] The electronic device may include a processor 10, a memory 11, a communication bus 12 and a communication interface 13, and may also include a computer program stored in the memory 11 and capable of running on the processor 10, such as a disease-gene relationship prediction program.

[0137] The memory 11 includes at least one type of readable storage medium, such as flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of an electronic device, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc. Furthermore, the memory 11 can include both internal and external storage units of the electronic device. The memory 11 can be used not only to store application software and various types of data installed on the electronic device, such as code for disease-gene relationship prediction programs, but also to temporarily store data that has been output or will be output.

[0138] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules (such as disease-gene relationship prediction programs) stored in the memory 11, and calls data stored in the memory 11 to perform various functions of the electronic device and process data.

[0139] The communication bus 12 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The communication bus 12 is configured to enable communication between the memory 11 and at least one processor 10, etc. For ease of illustration, only one thick line is used in the figure, but this does not indicate that there is only one bus or one type of bus.

[0140] Figure 5 Only electronic devices with components are shown; it will be understood by those skilled in the art that... Figure 5 The structure shown does not constitute a limitation on the electronic device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0141] For example, although not shown, the electronic device may also include a power supply (such as a battery) to power the various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0142] Optionally, the communication interface 13 may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, a Bluetooth interface, etc.), which is typically used to establish communication connections between the electronic device and other electronic devices.

[0143] Optionally, the communication interface 13 may further include a user interface, which may be a display, an input unit (such as a keyboard), or a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface.

[0144] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0145] The disease-gene relationship prediction program stored in the memory 11 of the electronic device is a combination of multiple computer programs, which, when run in the processor 10, can achieve the following:

[0146] Acquire gene data and disease data to be analyzed, construct a first gene regulatory network based on protein interaction data in the gene data to be analyzed, and construct a second gene regulatory network based on the relationship between transcription factors and messenger RNA in the gene data to be analyzed.

[0147] Calculate the semantic similarity between the disease data to be analyzed, and construct a disease similarity network based on the semantic similarity;

[0148] Extract the relationship between the gene data to be analyzed and the disease data to be analyzed, and construct a gene-disease relationship network based on the relationship;

[0149] The first gene regulation network and the disease similarity network are integrated to obtain a first integrated network. The representation vectors of each node in the first integrated network and the gene-disease relationship network are extracted to obtain a first representation vector and a relationship node representation vector. According to a preset first weight, the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector are weighted and summed to obtain a first target representation vector.

[0150] The second gene regulation network and the disease similarity network are integrated to obtain a second integrated network. The representation vectors of each node in the second integrated network are extracted to obtain a second representation vector. According to the preset second weight, the representation vectors representing similar genes and disease relationships in the second representation vector and the relationship node representation vectors are weighted and summed to obtain a second target representation vector.

[0151] According to the preset third weight, the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector are weighted and summed to obtain the target representation vector;

[0152] Linear calculations are performed on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed.

[0153] Specifically, the processor 10's implementation method of the above-mentioned computer program can be found in [reference needed]. Figure 1 The descriptions of the relevant steps in the corresponding embodiments are not repeated here.

[0154] Furthermore, if the modules / units integrated into the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable medium can be non-volatile or volatile. The computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0155] Embodiments of the present invention may also provide a computer-readable storage medium storing a computer program, which, when executed by a processor of an electronic device, can perform the following:

[0156] Acquire gene data and disease data to be analyzed, construct a first gene regulatory network based on protein interaction data in the gene data to be analyzed, and construct a second gene regulatory network based on the relationship between transcription factors and messenger RNA in the gene data to be analyzed.

[0157] Calculate the semantic similarity between the disease data to be analyzed, and construct a disease similarity network based on the semantic similarity;

[0158] Extract the relationship between the gene data to be analyzed and the disease data to be analyzed, and construct a gene-disease relationship network based on the relationship;

[0159] The first gene regulation network and the disease similarity network are integrated to obtain a first integrated network. The representation vectors of each node in the first integrated network and the gene-disease relationship network are extracted to obtain a first representation vector and a relationship node representation vector. According to a preset first weight, the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector are weighted and summed to obtain a first target representation vector.

[0160] The second gene regulation network and the disease similarity network are integrated to obtain a second integrated network. The representation vectors of each node in the second integrated network are extracted to obtain a second representation vector. According to the preset second weight, the representation vectors representing similar genes and disease relationships in the second representation vector and the relationship node representation vectors are weighted and summed to obtain a second target representation vector.

[0161] According to the preset third weight, the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector are weighted and summed to obtain the target representation vector;

[0162] Linear calculations are performed on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed.

[0163] Furthermore, the computer's usable storage medium may mainly include a program storage area and a data storage area, wherein the program storage area may store the operating system, applications required for at least one function, etc.; and the data storage area may store data created based on the use of blockchain nodes, etc.

[0164] In the several embodiments provided by this invention, it should be understood that the disclosed electronic devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0165] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0166] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0167] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0168] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0169] The blockchain referred to in this invention is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0170] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. The term "second class" is used to indicate names and does not indicate any specific order.

[0171] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for predicting the relationship between diseases and genes, characterized in that, The method includes: Acquire gene data and disease data to be analyzed, construct a first gene regulatory network based on protein interaction data in the gene data to be analyzed, and construct a second gene regulatory network based on the relationship between transcription factors and messenger RNA in the gene data to be analyzed. Calculate the semantic similarity between the disease data to be analyzed, and construct a disease similarity network based on the semantic similarity; Extract the relationship between the gene data to be analyzed and the disease data to be analyzed, and construct a gene-disease relationship network based on the relationship; The first gene regulation network and the disease similarity network are integrated to obtain a first integrated network. The representation vectors of each node in the first integrated network and the gene-disease relationship network are extracted to obtain a first representation vector and a relationship node representation vector. According to a preset first weight, the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector are weighted and summed to obtain a first target representation vector. The second gene regulation network and the disease similarity network are integrated to obtain a second integrated network. The representation vectors of each node in the second integrated network are extracted to obtain a second representation vector. According to the preset second weight, the representation vectors representing similar genes and disease relationships in the second representation vector and the relationship node representation vectors are weighted and summed to obtain a second target representation vector. According to the preset third weight, the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector are weighted and summed to obtain the target representation vector; Linear calculations are performed on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed.

2. The method for predicting the relationship between diseases and genes as described in claim 1, characterized in that, The linear calculation of the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed includes: The target representation vector is linearly transformed using the first preset parameter, the second preset parameter, and the third preset parameter in the fully connected layer of the pre-trained prediction model to obtain the query vector, the key vector, and the numerical vector. The similarity matrix is ​​obtained by multiplying the query vector with the transpose of the key vector. The similarity matrix is ​​normalized to obtain a normalized matrix; The normalization matrix is ​​calculated using a preset activation function to obtain the activation matrix; The activation matrix is ​​multiplied by the numerical vector to obtain the probability that the gene data to be analyzed and the disease data to be analyzed contained in the target representation vector are related.

3. The method for predicting the relationship between diseases and genes as described in claim 1, characterized in that, The calculation of semantic similarity between the disease data to be analyzed includes: Obtain the disease name data from the disease data to be analyzed; Extract the disease feature vector from the disease name data; The semantic similarity between each disease feature vector is calculated using the cosine similarity formula.

4. The method for predicting the relationship between diseases and genes as described in claim 3, characterized in that, The calculation of semantic similarity between each disease feature vector using the cosine similarity formula includes: The semantic similarity between each disease feature vector is calculated using the following formula. : in, Represents any one of the disease feature vectors, Indicates except Any disease feature vector other than, Indicates except The transpose of any disease feature vector other than .

5. The method for predicting the relationship between diseases and genes as described in claim 1, characterized in that, The step of extracting the representation vectors of each node in the first integrated network and the gene-disease relationship network to obtain the first representation vector and the relationship node representation vector includes: The feature vectors of the first integrated network and the gene-disease relationship network are extracted using the pre-trained prediction model to obtain the first feature vector and the gene-disease feature vector. The pre-trained prediction model is used to perform position index encoding on the first feature vector and the gene disease feature vector respectively to obtain the first position encoding vector and the gene disease position encoding vector. The first position encoding vector and the gene disease position encoding vector are combined with the corresponding first feature vector and the gene disease feature vector to obtain the first point encoding vector and the gene disease encoding vector, respectively. The convolutional layers in the pre-trained prediction model are used to extract features from the first point encoding vector and the gene disease encoding vector to obtain the first representation vector and the relationship node representation vector.

6. The method for predicting the relationship between diseases and genes as described in claim 5, characterized in that, Before extracting feature vectors from the first integrated network and the gene-disease relationship network using the pre-trained prediction model to obtain the first feature vector and the gene-disease feature vector, the method further includes: Acquire historical gene data, historical disease data, and the true relationship between historical diseases and genes; extract the relationship between the historical gene data and the historical disease data; and construct a historical gene-disease relationship network based on the relationship. A historical gene regulatory network was constructed based on the aforementioned historical gene data; Calculate the historical semantic similarity between the historical disease data, and construct a historical disease similarity network based on the historical semantic similarity; A preliminary prediction model is constructed, integrating the historical gene regulation network and the historical disease similarity network to obtain a historical integrated network. The representation vectors of each node in the historical integrated network and the historical gene-disease relationship network are extracted to obtain the historical node representation vector and the historical relationship node representation vector. Based on the preset historical weights, the representation vectors representing similar genes and disease relationships in the historical node representation vector and the historical relationship node representation vector are weighted and summed to obtain the historical target node representation vector. The pooling layer in the preliminary prediction model is used to perform feature dimensionality reduction on the historical target node representation vector to obtain the historical dimensionality-reduced node representation vector. Linear calculation is performed on the historical dimensionality reduction node representation vector to obtain the relationship probability of each node in the historical gene-disease relationship network. Based on the true relationship between the historical disease and the gene, the loss value of the relationship probability is calculated using a preset loss function. The parameters of the preliminary prediction model are adjusted according to the loss value, and the steps of using the pooling layer in the preliminary prediction model to perform feature dimensionality reduction on the historical target node representation vector to obtain the historical dimensionality-reduced node representation vector are returned. The preliminary prediction model is iteratively updated until the loss value is less than a preset threshold, and the pre-trained prediction model is obtained.

7. The method for predicting the relationship between diseases and genes as described in claim 1, characterized in that, The step of weighted summation of the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector according to a preset first weight, to obtain the first target representation vector, includes: The similarity between the first representation vector and the relationship node representation vector is calculated using cosine similarity, and the representation vectors with similarity higher than a preset threshold between the first representation vector and the relationship node representation vector are used as the representation vectors of similar genes and disease relationships. Based on the representation vectors of the similar genes and disease relationships and the preset first weights, the first target representation vector is calculated using the following formula. : in, This indicates the weight percentage of the first representation vector in the preset first weight. This represents the representation vector in the first representation vector that represents the relationship between similar genes and diseases. This indicates the weight percentage of the relation node representation vector in the preset first weight. This represents the representation vector in the relationship node representation vector that represents the relationship between similar genes and diseases in the first representation vector.

8. A device for predicting the relationship between diseases and genes, characterized in that, The device includes: A network construction module is used to acquire gene data and disease data to be analyzed, construct a first gene regulation network based on protein interaction data in the gene data to be analyzed, construct a second gene regulation network based on the relationship between transcription factors and messenger RNA in the gene data to be analyzed, calculate the semantic similarity between the disease data to be analyzed, construct a disease similarity network based on the semantic similarity, extract the relationship between the gene data to be analyzed and the disease data to be analyzed, and construct a gene-disease relationship network based on the relationship. The representation vector calculation module is used to integrate the first gene regulation network and the disease similarity network to obtain a first integrated network, extract the representation vectors of each node in the first integrated network and the gene-disease relationship network to obtain a first representation vector and a relationship node representation vector, and perform a weighted summation of the representation vectors representing similar genes and disease relationships in the first representation vector and the relationship node representation vector according to a preset first weight to obtain a first target representation vector. The module then integrates the second gene regulation network and the disease similarity network to obtain a second integrated network, extracts the representation vectors of each node in the second integrated network to obtain a second representation vector, and performs a weighted summation of the representation vectors representing similar genes and disease relationships in the second representation vector and the relationship node representation vector according to a preset second weight to obtain a second target representation vector. Finally, it performs a weighted summation of the representation vectors representing similar genes and disease relationships in the first target representation vector and the second target representation vector according to a preset third weight to obtain a target representation vector. The relevant probability calculation module is used to perform linear calculations on the target representation vector to obtain the probability that the gene data to be analyzed contained in the target representation vector is related to the disease data to be analyzed.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores computer program instructions that can be executed by the at least one processor to enable the at least one processor to perform the disease-gene relationship prediction method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the disease-gene relationship prediction method as described in any one of claims 1 to 7.

Images