Method, device and equipment for excavating corn pleiotropic genes

Abstract

The invention relates to the technical field of corn gene prediction, in particular to a method, device and equipment for mining corn pleiotropic genes. According to the embodiment of the invention, the full-connection neural network model is trained through a sample composed of gene expression attributes and integrated omics network attributes, and the multi-effect gene capable of balancing corn growth and development and responding to environmental stress resistance can be mined from unknown genes. According to the method, the device and the equipment, the misjudgment risk of irrelevant genes can be effectively reduced, the screened candidate genes are ensured to have relatively high functional correlation, and a more efficient verification basis is provided.

Description

Technical Field

[0001] The present application relates to the technical field of corn gene prediction, and in particular to methods, devices and equipment for mining pleiotropic genes in corn. Background Art

[0002] As one of the most important food crops in the world, the yield and quality of corn have a significant impact on global food security. However, corn faces many challenges during its growth, such as disease invasion, adverse environmental conditions (such as drought, salinity, etc.) and other growth and development disorders. These factors will not only affect the growth and development of corn, but also significantly reduce its stress resistance, ultimately affecting yield and quality. At present, the research methods on the genetic improvement of pleiotropic genes mainly focus on studying the effect of a certain gene on the shape through genetic engineering technology or the replacement, addition or substitution of related protein amino acid residues, and lack the ability to predict unknown pleiotropic genes. Summary of the invention

[0003] The inventors of the present application have developed a method, apparatus, equipment and medium for mining pleiotropic genes in corn, and introduced a fully connected neural network model to automatically identify potential pleiotropic genes from large-scale biological data, and further enhance the prediction capability by using the multi-dimensional characteristics of network topology properties and gene expression properties, thereby enabling rapid mining of pleiotropic genes.

[0004] This application can accurately select the appropriate probability threshold to balance the prediction accuracy and false positive rate by analyzing the probability density curve distribution of the test samples and unknown gene samples. This application can effectively reduce the risk of misjudgment of irrelevant genes, ensure that the selected candidate genes have high functional relevance, and provide a more efficient verification basis.

[0005] This application not only screens candidate genes by constructing a gene regulatory network and parsing network subgraphs, but also further confirms the actual effect of the gene through functional verification (such as inoculating corn leaf blight bacteria and constructing mutants). This systematic verification process ensures the effectiveness of candidate genes in practical applications. This system verification method greatly improves the reliability of the prediction probability density curve and provides direct experimental data support for subsequent breeding work.

[0006] This application applies machine learning technology to the screening and verification of corn pleiotropic genes, which is innovative in the current technical field. Compared with the existing single gene research methods, the corn pleiotropic gene mining method, device, equipment and medium provided by this application can more comprehensively reveal the regulatory mechanism between genes, providing a new technical path for corn high-yield breeding.

[0007] To this end, the embodiments of the present application disclose at least the following technical solutions:

[0008] In the first aspect, the embodiment discloses a method for mining corn pleiotropic genes. The method includes: obtaining at least one test sample, the test sample comprising a matrix composed of gene expression attributes of M genes and integrated omics network attributes, M being a positive integer; outputting a predicted probability density curve according to a trained fully connected neural network, the fully connected neural network comprising an input layer, a hidden layer and an output layer, the input layer being used to input a matrix, the hidden layer being used to extract multidimensional features in gene expression attributes and / or multidimensional features in integrated omics network attributes from the matrix, and the output layer being used to compress the multidimensional features to output a predicted probability density curve; obtaining a first probability threshold and N genes regulating corn growth and development above the first probability threshold according to the predicted probability density curve, obtaining a second probability threshold and P genes regulating corn stress resistance above the second probability threshold, wherein N and P are both positive integers; constructing a gene regulatory network balancing corn growth and development performance and responding to environmental stress resistance performance according to the N genes and the P genes; extracting a network subgraph from the gene regulatory network; mining pleiotropic genes balancing corn growth and development performance and responding to environmental stress resistance performance according to the network subgraph.

[0009] In the second aspect, the embodiment discloses a device for predicting pleiotropic genes in corn. The device includes a first acquisition unit, a first processing unit, a second processing unit, a third processing unit, and a fourth processing unit. The acquisition unit is used to acquire at least one test sample, wherein the test sample comprises a matrix composed of gene expression properties and integrated omics network properties of M genes, where M is a positive integer. The first processing unit is used to output a predicted probability density curve according to a trained fully connected neural network, wherein the fully connected neural network comprises an input layer, a hidden layer, and an output layer, wherein the input layer is used to input a matrix, the hidden layer is used to extract multidimensional features in gene expression properties and / or multidimensional features in integrated omics network properties from the matrix, and the output layer is used to compress the multidimensional features to output a predicted probability density curve. The second processing unit is used to acquire a first probability threshold and N genes that regulate corn growth and development above the first probability threshold according to the predicted probability density curve, and acquire a second probability threshold and P genes that regulate corn stress resistance above the second probability threshold, wherein N and P are both positive integers. The third processing unit is used to construct a gene regulatory network that balances corn growth and development performance and responds to environmental stress resistance based on the N genes and the P genes. The fourth processing unit is used to extract a network subgraph from the gene regulatory network. The fifth processing unit is used to mine pleiotropic genes that balance the growth and development performance of corn and the performance of responding to environmental stress according to the network subgraph.

[0010] In a third aspect, an embodiment discloses an electronic device, comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, a code set or an instruction set, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the method described in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic flow chart of a method for mining pleiotropic genes in maize provided in an embodiment.

[0012] Figure 2 A schematic diagram of the process of constructing an integrated omics network provided in an embodiment.

[0013] Figure 3 A schematic diagram of the structure of a fully connected neural network provided in an embodiment.

[0014] Figure 4 A schematic diagram of the training process of the first fully connected neural network provided in an embodiment.

[0015] Figure 5 A schematic diagram of the training process of the second fully connected neural network provided in the embodiment.

[0016] Figure 6 This is a flow chart of step S30 provided in the embodiment.

[0017] Figure 7 The embodiment provides a first fully connected neural network that is trained to obtain a predicted probability density curve and a true probability density curve.

[0018] Figure 8 The embodiment provides a second fully connected neural network that is trained to obtain a predicted probability density curve and a true probability density curve.

[0019] Fig. 9 A network subgraph provided for an embodiment. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application is further described in detail below in conjunction with the examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. The reagents not described separately in detail in the present application are all conventional reagents and can be obtained from commercial channels; the methods not described in detail are all conventional experimental methods and can be obtained from the prior art.

[0021] It should be understood that the references to "one embodiment" or "some embodiments" throughout the specification mean that specific features, structures, or characteristics associated with the embodiment are included in at least one embodiment of the present application. Therefore, the references to "in one embodiment" or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0022] In the various embodiments of the present application, it should be understood that the size of the serial numbers of the following processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0023] By introducing machine learning models, this application can automatically identify potential pleiotropic genes in large-scale biological data. Compared with traditional single-gene research methods, data-driven methods can systematically analyze the entire genome and quickly and accurately predict a group of genes that may have important regulatory effects on growth and development and response to adversity.

[0024] Below, in conjunction with the accompanying drawings, the method, device, equipment and medium for mining pleiotropic genes of corn provided in the embodiments of the present application are described in detail through specific embodiments and their application scenarios. In the present application, "pleiotropic genes" are genes that affect the growth and development and stress resistance of corn. For example, the "pleiotropic genes" mined by the method provided in the embodiment are genes that simultaneously regulate the angle of corn leaves and resist corn leaf blight. Among them, the "growth and development" performance includes corn leaf area, plant height, and leaf angle. "Stress resistance" performance includes salt tolerance, drought resistance and disease resistance genes (rust, leaf blight, etc.).

[0025] See also Figure 1 As shown, the embodiment discloses a method for mining corn pleiotropic genes. The method comprises:

[0026] S10, obtaining at least one test sample, wherein the test sample comprises a matrix consisting of gene expression attributes and integrated omics network attributes of M genes, where M is a positive integer;

[0027] S20, outputting a predicted probability density curve according to the trained fully connected neural network, the fully connected neural network comprising an input layer, a hidden layer and an output layer, the input layer is used to input a matrix, the hidden layer is used to extract multidimensional features in gene expression attributes and / or multidimensional features in integrated omics network attributes from the matrix, and the output layer is used to compress the multidimensional features to output a predicted probability density curve;

[0028] S30, according to the predicted probability density curve, obtain a first probability threshold and N genes that regulate corn growth and development above the first probability threshold, and obtain a second probability threshold and P genes that regulate corn stress resistance above the second probability threshold, wherein N and P are both positive integers;

[0029] S40, constructing a gene regulatory network that balances corn growth and development performance and responds to environmental stress resistance according to the N genes and the P genes;

[0030] S50, extracting a network subgraph from the gene regulatory network;

[0031] S60. Mining pleiotropic genes that balance maize growth and development performance and respond to environmental stress resistance based on network subgraphs.

[0032] In this application, a "sample" is a data object that includes information records of gene expression properties and integrated omics network properties of M genes. For example, a "sample" records the above matrix. Among them, the M genes include genes that regulate corn growth and development and also include genes related to environmental stress.

[0033] In some embodiments, the matrix includes attribute information and numerical information of M genes. The attribute information includes gene expression attributes and integrated omics network attributes of the M genes. The numerical information includes the gene expression amount of the gene expression attributes of the M genes, the shortest distance, eccentricity, intimacy, degree, and feature centrality of the M genes in the integrated omics network attributes.

[0034] Among them, the gene expression attributes include attribute objects of M gene expression levels in tissues of corn at different developmental stages, as shown in Table 1, where "trans" represents the transcriptome, "ribo" represents the proteome, "N" means that the material was collected at 22:00, and "L" means that the material was collected at 9:00.

[0035] Table 1 Gene expression attribute objects

[0036]

[0037]

[0038] Among them, Figure 2 As shown in Figure 2, the construction process of the integrated omics network includes:

[0039] S101, constructing a co-expression network, where the co-expression network is a weighted gene co-expression network constructed based on RNA sequencing data, where the RNA sequencing data includes strand-specific RNA sequences (strand-specific RNA-seq), circular RNA sequences (CircRNA-seq), and small RNA sequences (sRNA-seq);

[0040] S102, constructing a co-translation network, where the co-translation network is a weighted gene co-expression network constructed based on Ribo-seq sequencing data;

[0041] S103, constructing a co-interaction network, wherein the co-interaction network is based on the processed data generated by the PacBio Sequel II sequencer, and a network constructed by screening 56,244 protein pairs with high confidence protein interactions;

[0042] S104. Construct an integrated omics network based on the weighted sum of the co-expression network, co-translation network, and co-interaction network.

[0043] In step S101 provided in some embodiments, a transcriptome co-expression network is constructed using an expression matrix of 37,180 elements (31,294 genes, 5,427 lncRNAs, 12 fusion RNAs, 289 circRNAs and 158 miRNAs) detected in more than 10 different tissues or stages. For example, WGCNA (v.1.70-3) is used to construct a co-expression network with default parameters.

[0044] In step S102 provided in some embodiments, Ribo-seq can capture mRNA fragments being translated, which is directly related to translation activity. After Ribo-seq quality control, 920,427,741 reads were obtained, and sequence alignment was performed using Hisat2. The translatome.tpm file generated after processing contained the expression levels of the translation group of 39,497 genes, in units of reads per million (TPM).

[0045] In step S103 provided in some embodiments, the raw data generated by the PacBio Sequel II sequencer is processed by SMRTLink and seqkit tools, and then fine sequence alignment and gene pair identification are performed to screen out 56,244 pairs of high-confidence protein-protein interactions.

[0046] In steps S101 to S104 provided in some embodiments, a co-expression network is constructed using RNA-seq data of 31 tissues or stages, a co-translation network is constructed using Ribo-seq data of 21 tissues or stages, and a co-interaction network is constructed using yeast two-hybrid screening data. An integrated omics network is constructed based on the sum of the weights of the co-expression network, the co-translation network, and the co-interaction network. The weight of the integrated omics map is equal to the sum of the weights at different levels.

[0047] Among them, the integrative omics network attributes include the shortest distance attribute object, eccentricity attribute object, intimacy attribute object, degree attribute object, and feature centrality attribute object of M genes in different developmental stages or different tissues of corn in the integrative omics network.

[0048] The numerical information of the shortest distance attribute object is the Euclidean distance value of M genes in different developmental stages or tissues of corn in the integrated omics network. The numerical information of the eccentricity attribute object is the eccentricity value of M genes in different developmental stages or tissues of corn in the integrated omics network. The numerical information of the intimacy attribute object is the intimacy value of M genes in different developmental stages or tissues of corn in the integrated omics network. The numerical information of the degree attribute object is the level value of M genes in different developmental stages or tissues of corn in the integrated omics network. The numerical information of the characteristic centrality attribute object is the characteristic centrality value of M genes in different developmental stages or tissues of corn in the integrated omics network.

[0049] Among them, the numerical information of gene expression under transcriptomics is the reading or abundance of mRNA, which is obtained through technologies such as high-throughput sequencing (RNA-Seq). The abundance of RNA can indirectly characterize the activity of genes in cells. Among them, the numerical information of expression under proteomics is the number of ribosome occupancy reads of each gene in the ribosome rotation sequencing (Ribo-seq) data. The number of ribosome occupancy reads of each gene is proportional to the translation activity of the gene. The more ribosome occupancy, the faster the mRNA of the gene is translated and the more protein is generated.

[0050] The numerical information of the eccentricity attribute object is the maximum value of the shortest path from a vertex in the integrative omics network to all other vertices in the graph, which is the length of the shortest path from the vertex to the other vertices farthest from it. The eccentricity reflects the "distance" of a vertex, that is, the degree of its deviation from the "center" position in the graph.

[0051] The numerical information of the closeness attribute object is the average reciprocal of the shortest path from a vertex to all other vertices in the integrative omics network, reflecting the average distance between a vertex and other vertices. Vertices with higher closeness are closer to other vertices.

[0052] The numerical information of the degree attribute object is the number of adjacent vertices of a vertex in the integrative omics network, which reflects the degree of connection of the vertex in the network.

[0053] The numerical information of the feature centrality attribute object is the feature centrality value of the feature vector of the adjacency matrix in the integrated omics network.

[0054] Among them, the numerical information of the shortest path distance attribute object is the shortest path length between two vertices in the integrated omics network, that is, the minimum number of edges passed from one vertex to another.

[0055] In some embodiments of the electronic device executing S10, the electronic device can directly receive a matrix including gene expression attributes of multiple genes and integrative omics network attributes as a test sample. In other embodiments, the electronic device can receive any one or more information in the matrix and automatically generate a matrix as a test sample based on the information.

[0056] In some embodiments of the electronic device executing S20, the fully connected neural network is stored in the electronic device. The fully connected neural network model is a typical multi-input fully connected neural network, which outputs a binary classification result through layer-by-layer feature extraction and fusion.

[0057] like Figure 3 As shown, in some embodiments, the fully connected neural network includes an input layer (Lay1), three hidden layers (Lay21, Lay22, Lay23) and an output layer (Lay3).

[0058] Among them, the input layer Lay1 includes multiple neurons, for example, these neurons are used to input the gene expression properties of multiple genes and the integrated omics network properties. For example, among them, 26 neurons are used to receive a 26-dimensional input, and the 26-dimensional input is the TPM value of different tissues or stages under 26 transcriptomics, and the input format is (m, 26), where m represents the number of samples and 26 is the dimension of the feature. For example, among them, 16 neurons are used to receive a 16-dimensional input, and the 16-dimensional input is the TPM value of different tissues or stages under 16 proteomics, and the input format is (m, 16), where m represents the number of samples and 16 is the dimension of the feature.

[0059] like Figure 3 As shown, the hidden layer Lay21 has multiple neurons for receiving the output of the input layer Lay1 and performing feature extraction.

[0060] For example, the 100 neurons in the hidden layer Lay21 receive the 100 output data of the input layer Lay1, add them up, and then use an activation function ReLU to get their output. The output dimension is (m, 100). The Dropout layer randomly discards some neurons with a probability of 0.1 without changing the output dimension to prevent overfitting.

[0061] For example, the 50 neurons in the hidden layer Lay21 receive the 50 output data of the input layer Lay1, and after weighted summation, they are activated through an activation function ReLU to get their output. The output dimension is (m, 50). The Dropout layer randomly discards some neurons with a probability of 0.1 without changing the output dimension to prevent overfitting.

[0062] For example, the 50 neurons in the hidden layer Lay21 receive the 50 output data of the input layer Lay1, add them up, and then use an activation function ReLU to get their output. The output dimension is (m, 50), and the Dropout layer dimension remains (m, 50).

[0063] For example, hidden layer Lay22 is a fully connected layer that concatenates the output of hidden layer Lay21 to form a comprehensive feature representation. That is, features with dimensions (m, 100), (m, 50), and (m, 50) are concatenated to form a feature vector of (m, 200).

[0064] For example, the hidden layer Lay23 has 50 neurons, and the feature vector (m, 200) output by the hidden layer Lay22 is compressed into 50 dimensions to obtain a 50-dimensional feature vector.

[0065] For example, the output layer Lay3 is a fully connected layer with only a single neuron. Using the Sigmoid activation function, the 50-dimensional feature vector output by the hidden layer Lay23 is compressed into a one-dimensional feature vector, which is a one-dimensional prediction probability vector, (m, 1), which is represented as a prediction probability density curve.

[0066] In some embodiments, the fully connected neural network is trained using an Adam optimizer with an adaptive learning rate, binary_crossentropy is used as a loss function to continuously train and optimize the model, and accuracy is used to measure the accuracy of the fully connected neural network.

[0067] In some embodiments, the fully connected neural network is trained using the fit function, and the ModelCheckpoint callback function is used to save the model with the smallest verification loss in each epoch.

[0068] In some embodiments, Figure 4 As shown, the electronic device executes step S20 to obtain a trained first fully connected neural network, and the first fully connected neural network is used to mine pleiotropic genes that regulate growth and development. Based on this, the training steps of the first fully connected neural network include:

[0069] S201, obtaining a training set, the training set comprising a plurality of positive samples and a plurality of negative samples. The positive sample comprises at least one gene expression attribute and an integrated omics network attribute regulating growth and development. The negative sample comprises at least one gene expression attribute and an integrated omics network attribute expressed in response to adversity.

[0070] S202, obtaining a first fully connected neural network including an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer;

[0071] S203, the input layer obtains each positive sample and each negative sample from the training set;

[0072] S204, the first hidden layer extracts the gene expression attributes and integrated omics network attributes that regulate growth and development from the positive samples and extracts the gene expression attributes and integrated omics network attributes that respond to adversity expression from the negative samples, normalizes the gene expression attributes and integrated omics network attributes that regulate growth and development, and normalizes the gene expression attributes and integrated omics network attributes that respond to adversity expression, and outputs them as a plurality of first multidimensional feature vectors;

[0073] S205, the second hidden layer concatenates all features in the plurality of first multidimensional feature vectors to obtain a second multidimensional feature vector;

[0074] S206, the third hidden layer reduces the dimension of the features in the second multidimensional feature vector to obtain a third multidimensional feature vector;

[0075] S207, the output layer compresses the multi-dimensional features of the third feature vector into a fourth one-dimensional feature vector, which is a prediction probability vector;

[0076] S208. If it is detected that the loss functions in steps S203 to S207 are all minimum, the training of the first fully connected neural network is completed.

[0077] In some embodiments, the positive sample includes 31 data objects of gene expression attributes and integrated omics network attributes related to corn growth and development, and the negative sample includes 54 data objects of gene expression attributes and integrated omics network attributes related to corn environmental stress.

[0078] Among them, the names of these 31 genes are Zm00001d011721, Zm00001d020874, Zm00001d053453, Zm00001d043841, Zm00001d031451, Zm00001d031871, Zm00001d032461, Zm00001d033180, Zm00001d033303, Zm00001d033422, Zm00001d033680, Zm00001d033859, Zm00001d033905, Zm00001d034204, Zm00001d034383, m00001d034898, Zm00001d002349, Zm00001d004889, Zm00001d007842, Zm00001d033174, Zm00001d007234, Zm00001d032385, Zm00001d005737, Zm00001d040611, Zm00001d013895, Zm00001d017391, Zm00001d035256, Zm00001d018742, Zm00001d020384, Zm00001d010948, Zm00001d045945.Among them, the names of these 54 genes are Zm00001d028752, Zm00001d047993, Zm00001d022227, Zm00001d028160, Zm00001d013428, Zm00001d037332, Zm00001d020620, Zm00001d028560, Zm00001d045206, Zm00001d023317, Zm00001d007745, Zm00001d012441, and Zm00001d03357 2. Zm00001d041711, Zm00001d018698, Zm00001d031861, Zm00001d038762, Zm00001d033104, Zm00001d046967, Zm00001d049 244. Zm00001d013111, Zm00001d039000, Zm00001d042880, Zm00001d012398, Zm00001d028630, Zm00001d045036, Zm00001d0 48688, Zm00001d021532, Zm00001d043019, Zm00001d037894, Zm00001d050222, Zm00001d014919, Zm00001d003470, Zm0000 1d032828, Zm00001d038576, Zm00001d012447, Zm00001d024268, Zm00001d008721, Zm00001d024687, Zm00001d047802, Zm00 001d017212, Zm00001d039104, Zm00001d037043, Zm00001d048373, Zm00001d028818, Zm00001d021395, Zm00001d029968, Zm00001d034254, Zm00001d006555, Zm00001d045323, Zm00001d004936, Zm00001d048095, Zm00001d004997, Zm00001d050056. The name identifier is selected from "Zm00001d000001-999999", wherein "Zm" represents corn.

[0079] In some embodiments, Figure 5 As shown, the electronic device executes step S20 to obtain a trained second fully connected neural network, and the second fully connected neural network is used to mine pleiotropic genes that respond to stress resistance. Based on this, the training steps of the second fully connected neural network include:

[0080] S201, obtaining a training set, the training set comprising a plurality of positive samples and a plurality of negative samples. The positive sample comprises at least one gene expression attribute and an integrated omics network attribute that responds to adversity expression. The negative sample comprises at least one gene expression attribute and an integrated omics network attribute that regulates growth and development.

[0081] S202, obtaining a second fully connected neural network including an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer;

[0082] S203, the input layer obtains each positive sample and each negative sample from the training set;

[0083] S204, the first hidden layer extracts the gene expression attributes and integrated omics network attributes expressed in response to adversity from the positive samples and extracts the gene expression attributes and integrated omics network attributes regulating growth and development from the negative samples, normalizes the gene expression attributes and integrated omics network attributes expressed in response to adversity, and normalizes the gene expression attributes and integrated omics network attributes regulating growth and development, and outputs them as a plurality of first multidimensional feature vectors;

[0084] S205, the second hidden layer concatenates all features in the plurality of first multidimensional feature vectors to obtain a second multidimensional feature vector;

[0085] S206, the third hidden layer reduces the dimension of the features in the second multidimensional feature vector to obtain a third multidimensional feature vector;

[0086] S207, the output layer compresses the multi-dimensional features of the third feature vector into a fourth one-dimensional feature vector, which is a prediction probability vector;

[0087] S208. If it is detected that the loss functions in steps S203 to S207 are all minimum, the training of the second fully connected neural network is completed.

[0088] In some embodiments, the positive sample includes the 54 data objects of gene expression attributes and integrated omics network attributes related to corn environmental stress, and the negative sample includes 31 data objects of gene expression attributes and integrated omics network attributes related to corn growth and development. The 35 genes are the same as the above embodiment. The 54 genes are the same as the above embodiment.

[0089] In some embodiments, Figure 6 As shown, step S30 includes:

[0090] S301, obtaining an overlapping portion of a predicted probability density curve and a true probability density curve;

[0091] S302, obtaining a probability threshold in the overlapping part;

[0092] S303, obtaining N genes above the probability threshold.

[0093] In some embodiments, the MCODE module based on Cytoscape parses a network subgraph from the gene regulatory network with default parameters, and then mines pleiotropic genes that regulate corn growth and development or pleiotropic genes that respond to environmental stress based on the network subgraph.

[0094] Figure 7 The gene prediction probability curve and the true probability curve (ie, the probability curve composed of positive samples of 31 genes) for mining pleiotropic genes regulating corn growth and development are shown. Figure 8 The predicted probability curve and true probability curve of the gene for mining the multi-effect genes regulating maize response to environmental stress (i.e., the probability curve composed of positive samples of 54 genes) are shown. In the figure, the horizontal axis represents the probability of a gene being predicted as a certain trait category, the area represents the number of genes predicted under this probability, and the horizontal axis corresponding to the dotted line is the selected probability threshold.

[0095] Will Figure 7 The predicted probability curves of multiple genes above the probability threshold of 0.9 and Figure 8 The gene regulatory network was constructed by combining multiple genes in the predicted probability curve with a probability threshold of 0.87. The following results were obtained based on the MCODE module of Cytoscape with default parameters: Fig. 9 In the network subgraph shown, it was found that the MCODE::Score of Zm00001d047757 (APX1) in the subgraph was 23.88, indicating that the gene occupies a certain important position in the subgraph. It is inferred that the gene also plays a certain role in growth and development and stress resistance, and it is used as a candidate gene for corresponding verification.

[0096] It should be noted that the method for mining pleiotropic genes of maize provided in the embodiment of the present application can be performed by a device for mining pleiotropic genes of maize, or a control unit in the device for mining pleiotropic genes of maize for executing the method for mining pleiotropic genes of maize. In the embodiment of the present application, the method for mining pleiotropic genes of maize is performed by a device for mining pleiotropic genes of maize as an example to illustrate the device for mining pleiotropic genes of maize provided in the embodiment of the present application.

[0097] The device for mining corn pleiotropic genes may include a first acquisition unit, a first processing unit, a second processing unit, a third processing unit and a fourth processing unit. The acquisition unit is used to acquire at least one test sample, and the test sample comprises a matrix composed of multiple genes. The first processing unit is used to output a predicted probability density curve according to a trained fully connected neural network, and the fully connected neural network comprises an input layer, a hidden layer and an output layer, the input layer is used to input the test sample, the hidden layer is used to extract multidimensional features from the input layer, and the output layer is used to compress the multidimensional features to output the predicted probability density curve. The second processing unit is used to acquire multiple probability thresholds of the multiple genes according to the predicted probability density curve. The third processing unit is used to construct a functional subnetwork of the corn pleiotropic genes according to the multiple probability thresholds. The fourth processing unit is used to mine the corn pleiotropic genes according to the functional subnetwork.

[0098] The device for mining corn pleiotropic genes in the embodiment of the present application can be a device, or a component, integrated circuit, or chip in a terminal. The device can be a mobile electronic device or a non-mobile electronic device. Exemplarily, the mobile electronic device can be a mobile phone, a tablet computer, a laptop computer, a palmtop computer, an in-vehicle electronic device, a wearable device, an ultra-mobile personal computer (ultra-mobile personal computer, UMPC), a netbook or a personal digital assistant (personal digital assistant, PDA), etc., and the non-mobile electronic device can be a server, a network attached storage (Network Attached Storage, NAS), a personal computer (personal computer, PC), a television (television, TV), a teller machine or a self-service machine, etc., which is not specifically limited in the embodiment of the present application.

[0099] The device for mining maize pleiotropic genes in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0100] The device for mining pleiotropic genes of maize provided in the embodiment of the present application can realize each process realized by the device for mining pleiotropic genes of maize in the above-mentioned method embodiment for mining pleiotropic genes of maize, and will not be described again here to avoid repetition.

[0101] Optionally, an embodiment of the present application also provides an electronic device, including a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, each process of the above-mentioned method embodiment for mining pleiotropic genes in corn is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0102] It should be noted that the electronic devices in the embodiments of the present application include the above-mentioned mobile electronic devices and non-mobile electronic devices.

[0103] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0104] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0105] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the relevant technology, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0106] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed in the present application should be covered within the protection scope of the present application.

Claims

1. Methods for mining pleiotropic genes in maize, including: Obtain at least one test sample, wherein the test sample comprises a matrix consisting of gene expression properties of M genes and integrated omics network properties, where M is a positive integer Outputting a predicted probability density curve according to a trained fully connected neural network, the fully connected neural network comprising an input layer, a hidden layer and an output layer, the input layer being used to input the matrix, the hidden layer being used to extract multidimensional features in gene expression attributes and / or multidimensional features in integrated omics network attributes from the matrix, and the output layer being used to compress the multidimensional features to output a predicted probability density curve; According to the predicted probability density curve, a first probability threshold and N genes that regulate corn growth and development above the first probability threshold are obtained, and a second probability threshold and P genes that regulate corn stress resistance above the second probability threshold are obtained, wherein N and P are both positive integers; Constructing a gene regulatory network that balances corn growth and development performance and responds to environmental stress resistance performance according to the N genes and the P genes; Extract network subgraphs from gene regulatory networks; Based on the network subgraph, multi-functional genes that balance the growth and development performance of corn and respond to environmental stress resistance are mined.

2. The method as claimed in claim 1, wherein the matrix includes attribute information and numerical information of the M genes, the attribute information includes gene expression attributes and integrated omics network attributes of the M genes, and the numerical information includes gene expression amounts of the gene expression attributes of the M genes, the shortest distance, eccentricity, intimacy, degree and feature centrality of the M genes in the integrated omics network.

3. The method of claim 1, wherein the gene expression attributes include gene expression attribute objects of M gene expression amounts in different developmental stages or tissues of corn and numerical information of the gene expression attribute objects, and the gene expression attribute objects include trans-Root (6DAS), ribo-Root (6DAS), trans-PrimaryRoot (V1), ribo-PrimaryRoot (V1), trans-Root (14DAS), ribo-Root (14DAS), trans-Seedling (6DAS), ribo-Seedling (6DAS), trans-Seedling (14DAS), ribo-Seedling(14DAS), trans-TopmostLeaf-L(V3), ribo-TopmostLeaf-L(V3), trans-FirstLeaf-N(V3), ribo-FirstLeaf-N(V3), trans -ThirdLeaf(V3), ribo-ThirdLeaf(V3), trans-Leaf(14DAS), ribo-Leaf(14DAS), trans-TipofStage2Leaf-N(V5), ribo-TipofStage2Leaf-N(V5), trans-TipofStage2leaf-L(V5), ribo-TipofStage2leaf-L(V5), trans-BaseofStage2Leaf-N(V5), ribo-BaseofStage2Leaf-N(V5), trans-Baseof Stage2leaf-L(V5), ribo-BaseofStage2leaf-L(V5), trans-Ear-leaf, ribo-Ear-leaf, trans-Stem(14DAS), ribo-Stem(14DAS), trans-Stem(V6), ribo-Stem(V6), trans-Internode(V6), ribo-Internode(V6), trans-5mmear, ribo-5mmear, trans-Ear(V12), ribo-Ear(V12), trans-Inflorescen ce, ribo-Inflorescence, trans-1cmtassel, ribo-1cmtassel, trans-Tassel(V12), ribo-Tassel(V12), trans-Kernel(6DAP), ribo-Kernel(6DAP),trans-Embryo(15DAP)、ribo-Embryo(15DAP)、trans-Embryo(20DAP)、ribo-Embryo(20DAP)、trans-Endosperm(15DAP)、ribo-Endosperm(15DAP)、trans-Endosperm(20DAP)、ribo-Endosperm(15DAP)、trans-TopmostLeaf-N(V3)、ribo-TopmostLeaf-N(V3)、trans-2mmear、ribo-2mmear、trans-1cmear、ribo-1cmear、trans-2mmtassel、ribo-2mmtassel、trans-SAM(V1)、ribo-SAM(V1)、trans-SAM(V3)、ribo-SAM(V3);、 The numerical information of the gene expression attribute object includes the gene expression amount of the root tissue of corn 6 days after sowing in transcriptomics, the gene expression amount of the root tissue of corn 6 days after sowing in proteomics, the gene expression amount of the main root of corn developed to the first leaf stage (V1) in transcriptomics, the gene expression amount of the main root of corn developed to the first leaf stage (V1) in proteomics, the gene expression amount of the root tissue of corn 14 days after sowing in transcriptomics, the gene expression amount of the root tissue of corn 14 days after sowing in proteomics, the gene expression amount of the seedling tissue of corn 6 days after sowing in transcriptomics, the gene expression amount of the seedling tissue of corn 6 days after sowing in proteomics, and the gene expression amount of the seedling tissue of corn 14 days after sowing in transcriptomics. The gene expression levels in omics, the gene expression levels in proteomics of corn seedling tissues 14 days after sowing, the gene expression levels in transcriptomics of the leaf heart (L) of corn in the third leaf stage (V3 growth stage), the gene expression levels in proteomics of the leaf heart (L) of corn in the third leaf stage (V3 growth stage), the gene expression levels in transcriptomics of the first leaf (N) of corn in the third leaf stage (V3 growth stage), the gene expression levels in proteomics of the first leaf (N) of corn in the third leaf stage (V3 growth stage), the gene expression levels in transcriptomics of the third leaf of corn in the third leaf stage (V3 growth stage), Gene expression levels in proteomics of the third leaf in the leaf period (V3 growth stage), gene expression levels in transcriptomics of leaf tissue 14 days after corn sowing, gene expression levels in proteomics of leaf tissue 14 days after corn sowing, gene expression levels in transcriptomics of the tip of the leaf (N) in the second developmental period of leaves in the V5 stage of corn, gene expression levels in proteomics of the tip of the leaf (N) in the second developmental period of leaves in the V5 stage of corn, gene expression levels in transcriptomics of the tip of the leaf (L) in the second developmental period of leaves in the V5 stage of corn, gene expression levels in proteomics of the tip of the leaf (L) in the second developmental period of leaves in the V5 stage of corn, The gene expression level of the base of leaves (N) in transcriptomics, the gene expression level of the base of leaves in the second developmental period of corn leaves at stage V5 (N) in proteomics, the gene expression level of the base of leaves in the second developmental period of corn leaves at stage V5 (L) in transcriptomics, the gene expression level of the base of leaves in the second developmental period of corn leaves at stage V5 (L) in proteomics, the gene expression level of the middle 3 cm part of the first ear leaf of unpollinated corn in transcriptomics, the gene expression level of the middle 3 cm part of the first ear leaf of unpollinated corn in proteomics, the gene expression level of stem tissue 14 days after corn sowing in transcriptomics, the gene expression level of stem tissue 14 days after corn sowing in proteomics,Gene expression levels of the 6th internode corresponding to corn V6 stage in transcriptomics, gene expression levels of the 6th internode corresponding to corn V6 stage in proteomics, gene expression levels of the 6th internode of corn V6 stage in transcriptomics, gene expression levels of the 6th internode of corn V6 stage in proteomics, gene expression levels of corn 5 mm long ear (ear in early development stage) in transcriptomics, gene expression levels of corn 5 mm long ear (ear in early development stage) in proteomics, gene expression levels of corn ear in corn V12 stage in transcriptomics, gene expression levels of corn ear in corn V12 stage in proteomics, transcriptomics of outer layer of 2.5-3 cm female ear The gene expression levels in the outer layer of the 2.5-3 cm female ear of corn in proteomics, the gene expression levels in the transcriptomics of the male ear of corn about 1 cm long, the gene expression levels in the proteomics of the male ear of corn about 1 cm long, the gene expression levels in the transcriptomics of the male ear of corn at the V12 stage, the gene expression levels in the proteomics of the male ear of corn at the V12 stage, the gene expression levels in the transcriptomics of the corn kernels developed 6 days after pollination, the gene expression levels in the proteomics of the corn kernels developed 6 days after pollination, the gene expression levels in the transcriptomics of the embryos developed 15 days after pollination, the gene expression levels in the proteomics of the embryos developed 15 days after pollination, the gene expression levels in the proteomics of the corn pollination Gene expression levels in transcriptomics of embryos developing 20 days after pollination, gene expression levels in proteomics of embryos developing 20 days after pollination of corn, gene expression levels in transcriptomics of endosperm developing 15 days after pollination of corn, gene expression levels in proteomics of endosperm developing 15 days after pollination of corn, gene expression levels in transcriptomics of endosperm developing 20 days after pollination of corn, gene expression levels in proteomics of endosperm developing 20 days after pollination of corn, gene expression levels in transcriptomics of leaf heart (N) at the V3 stage of corn, gene expression levels in proteomics of the top leaf heart (N) at the V3 stage of corn, gene expression levels in transcriptomics of corn ears about 2 mm long, and gene expression levels in proteomics of corn ears about 2 mm long The gene expression levels in proteomics, the gene expression levels in transcriptomics of corn ears about 1 cm long, the gene expression levels in proteomics of corn ears about 1 cm long, the gene expression levels in transcriptomics of corn tassels about 2 mm long, the gene expression levels in proteomics of corn tassels about 2 mm long, the gene expression levels in transcriptomics of the stem apical meristem of corn at the first leaf expansion stage (V1 stage), the gene expression levels in proteomics of the stem apical meristem of corn at the first leaf expansion stage (V1 stage), the gene expression levels in transcriptomics of the stem apical meristem of corn at the third leaf expansion stage (V3 stage), and the gene expression levels in proteomics of the stem apical meristem of corn at the third leaf expansion stage (V3 stage).

4. The method according to claim 2, wherein the process of constructing the integrated omics network comprises: constructing a co-expression network, which is a weighted gene co-expression network constructed based on RNA sequencing data, wherein the RNA sequencing data includes strand-specific RNA sequences, circular RNA sequences, and small RNA sequences; Constructing a co-translation network, which is a weighted gene co-expression network based on Ribo-seq sequencing data; Constructing a co-interaction network based on the processed data generated by the PacBio SequelII sequencer, screening out 56,244 protein pairs with high confidence protein interactions; An integrated omics network was constructed based on the weighted sum of co-expression network, co-translation network, and co-interaction network.

5. The method of claim 1, comprising obtaining a trained first fully connected neural network, wherein the training step of the first fully connected neural network comprises: Obtaining a training set, the training set comprising a plurality of positive samples and a plurality of negative samples, the positive samples comprising at least one gene expression attribute and an integrated omics network attribute regulating growth and development, and the negative samples comprising at least one gene expression attribute and an integrated omics network attribute expressed in response to adversity; Obtaining a first fully connected neural network including an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; The input layer takes every positive sample and every negative sample from the training set; The first hidden layer extracts the gene expression attributes and integrated omics network attributes regulating growth and development from the positive samples and extracts the gene expression attributes and integrated omics network attributes expressing in response to adversity from the negative samples, normalizes the gene expression attributes and integrated omics network attributes regulating growth and development, and normalizes the gene expression attributes and integrated omics network attributes expressing in response to adversity, and outputs them as a plurality of first multidimensional feature vectors; The second hidden layer concatenates all features in the plurality of first multidimensional feature vectors to obtain a second multidimensional feature vector; The third hidden layer reduces the dimension of the features in the second multi-dimensional feature vector to obtain a third multi-dimensional feature vector; The output layer compresses the multi-dimensional features of the third feature vector into a fourth one-dimensional feature vector, which is the prediction probability vector; If it is detected that the loss functions in the steps of obtaining each positive sample and each negative sample from the training set and in the step of outputting the fourth one-dimensional feature vector by the output layer are all minimized, the training of the first fully connected neural network is completed.

6. The method of claim 5, wherein the positive sample comprises 31 data objects of gene expression attributes and integrated omics network attributes related to corn growth and development, and the negative sample comprises 54 data objects of gene expression attributes and integrated omics network attributes related to corn environmental stress, The names of the 31 genes are Zm00001d011721, Zm00001d020874, Zm00001d053453, Zm00001d043841, Zm00001d031451, Zm00001d031871, Zm00001d032461, Zm00001d033180, Zm00001d033303, Zm00001d033422, Zm00001d033680, Zm00001d033859, Zm00001d033905, Zm00001d034204, Zm00001d034383, and Zm 00001d034898, Zm00001d002349, Zm00001d004889, Zm00001d007842, Z m00001d033174, Zm00001d007234, Zm00001d032385, Zm00001d005737, Z m00001d040611, Zm00001d013895, Zm00001d017391, Zm00001d035256, Zm00001d018742, Zm00001d020384, Zm00001d010948, Zm00001d045945; The 54 genes are named as Zm00001d028752, Zm00001d047993, Zm00001d022227, Zm00001d028160, Zm00001d013428, Zm00001d037332, Zm00001d020620, Zm00001d028560, Zm00001d045206, Zm00001d023317, Zm00001d007745, Zm00001d012441, Zm00001d033572, Zm00001d041711, Zm00001d018698, Zm00001d031861, Zm00001d038762, Zm00001d033104, Zm00001d046967, Zm00001d0492 44. Zm00001d013111, Zm00001d039000, Zm00001d042880, Zm00001d012398, Zm00001d028630, Zm00001d045036, Zm00001d04 8688, Zm00001d021532, Zm00001d043019, Zm00001d037894, Zm00001d050222, Zm00001d014919, Zm00001d003470, Zm00001 d032828、Zm00001d038576、Zm00001d012447、Zm00001d024268、Zm00001d008721、Zm00001d024687、Zm00001d047802、Zm000 01d017212、Zm00001d039104、Zm00001d037043、Zm00001d048373、Zm00001d028818、Zm00001d021395、Zm00001d029968、Zm 00001d034254, Zm00001d006555, Zm00001d045323, Zm00001d004936, Zm00001d048095, Zm00001d004997, Zm00001d050056.

7. The method of claim 1, comprising obtaining a trained second fully connected neural network, wherein the training step of the second fully connected neural network comprises: Obtaining a training set, the training set comprising a plurality of positive samples and a plurality of negative samples, the positive samples comprising at least one gene expression attribute and an integrated omics network attribute expressed in response to adversity, and the negative samples comprising at least one gene expression attribute and an integrated omics network attribute regulating growth and development; Obtaining a second fully connected neural network including an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer; The input layer takes every positive sample and every negative sample from the training set; The first hidden layer extracts the gene expression attributes and integrated omics network attributes expressed in response to adversity from the positive samples and extracts the gene expression attributes and integrated omics network attributes regulating growth and development from the negative samples, normalizes the gene expression attributes and integrated omics network attributes expressed in response to adversity, and normalizes the gene expression attributes and integrated omics network attributes regulating growth and development, and outputs them as a plurality of first multidimensional feature vectors; The second hidden layer concatenates all features in the plurality of first multidimensional feature vectors to obtain a second multidimensional feature vector; The third hidden layer reduces the dimension of the features in the second multi-dimensional feature vector to obtain a third multi-dimensional feature vector; The output layer compresses the multi-dimensional features of the third feature vector into a fourth one-dimensional feature vector, which is the prediction probability vector; If it is detected that the loss functions in the steps of obtaining each positive sample and each negative sample from the training set and in the step of outputting the fourth one-dimensional feature vector by the output layer are all minimized, the training of the first fully connected neural network is completed.

8. The method of claim 7, wherein the positive sample comprises the 54 data objects of gene expression attributes and integrated omics network attributes related to corn environmental stress, and the negative sample comprises the 31 data objects of gene expression attributes and integrated omics network attributes related to corn growth and development; The names of the 31 genes are Zm00001d011721, Zm00001d020874, Zm00001d053453, Zm00001d043841, Zm00001d031451, Zm00001d031871, Zm00001d032461, Zm00001d033180, Zm00001d033303, Zm00001d033422, Zm00001d033680, Zm00001d033859, Zm00001d033905, Zm00001d034204, Zm00001d034383, and Zm 00001d034898, Zm00001d002349, Zm00001d004889, Zm00001d007842, Z m00001d033174, Zm00001d007234, Zm00001d032385, Zm00001d005737, Z m00001d040611, Zm00001d013895, Zm00001d017391, Zm00001d035256, Zm00001d018742, Zm00001d020384, Zm00001d010948, Zm00001d045945; The 54 genes are named as Zm00001d028752, Zm00001d047993, Zm00001d022227, Zm00001d028160, Zm00001d013428, Zm00001d037332, Zm00001d020620, Zm00001d028560, Zm00001d045206, Zm00001d023317, Zm00001d007745, Zm00001d012441, Zm00001d033572, Zm00001d041711, Zm00001d018698, Zm00001d031861, Zm00001d038762, Zm00001d033104, Zm00001d046967, Zm00001d0492 44. Zm00001d013111, Zm00001d039000, Zm00001d042880, Zm00001d012398, Zm00001d028630, Zm00001d045036, Zm00001d04 8688, Zm00001d021532, Zm00001d043019, Zm00001d037894, Zm00001d050222, Zm00001d014919, Zm00001d003470, Zm00001 d032828、Zm00001d038576、Zm00001d012447、Zm00001d024268、Zm00001d008721、Zm00001d024687、Zm00001d047802、Zm000 01d017212、Zm00001d039104、Zm00001d037043、Zm00001d048373、Zm00001d028818、Zm00001d021395、Zm00001d029968、Zm 00001d034254, Zm00001d006555, Zm00001d045323, Zm00001d004936, Zm00001d048095, Zm00001d004997, Zm00001d050056.

9. A device for mining pleiotropic genes of corn, comprising: An acquisition unit is configured to acquire at least one test sample, wherein the test sample comprises a matrix consisting of gene expression properties of M genes and integrative omics network properties, where M is a positive integer; A first processing unit is used to output a prediction probability density curve according to a trained fully connected neural network, wherein the fully connected neural network includes an input layer, a hidden layer, and an output layer, wherein the input layer is used to input a matrix, the hidden layer is used to extract multidimensional features in gene expression attributes and / or multidimensional features in integrated omics network attributes from the matrix, and the output layer is used to compress the multidimensional features to output a prediction probability density curve; The second processing unit is used to obtain a first probability threshold and N genes that regulate corn growth and development above the first probability threshold according to the predicted probability density curve, and obtain a second probability threshold and P genes that regulate corn stress resistance above the second probability threshold, wherein N and P are both positive integers; A third processing unit is used to construct a gene regulatory network that balances the growth and development performance of corn and responds to environmental stress resistance according to the N genes and the P genes; A fourth processing unit, configured to extract a network subgraph from the gene regulatory network; and The fifth processing unit is used to mine pleiotropic genes that balance the growth and development performance of corn and respond to environmental stress resistance based on the network subgraph.

10. A device comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction implements the method according to any one of claims 1 to 8 when executed by the processor.

Images