Crop phenotype prediction method based on hybrid expert algorithm and related device

Abstract

The invention discloses a crop phenotype prediction method based on a hybrid expert algorithm and a related device, and relates to the field of crop phenotype intelligent prediction, and the method comprises the steps: obtaining genotype data of a to-be-detected sample of a target crop; taking genotype data of a target crop sample to be detected as input, and outputting corresponding phenotype prediction data of the target crop sample to be detected by using a phenotype prediction model; the phenotype prediction model is a model trained by applying a hybrid expert algorithm. According to the invention, the phenotype prediction model is obtained through training based on the hybrid expert algorithm, so that accurate prediction of different agronomic trait expressions of the target crop is realized.

Description

Technical Field

[0001] The present application relates to the field of intelligent prediction of crop phenotypes, and in particular to a crop phenotype prediction method and related devices based on a mixture of experts (MOE) algorithm. Background Art

[0002] Crop breeding originated in the late 19th century and is a scientific technology for humans to purposefully transform crops. Selection is the key foundation of crop breeding, and human improvement of crops has gone through a process from phenotypic selection to molecular selection. Phenotypic selection is based on the phenotype of observable or investigated agronomic traits, and decisions are made through the knowledge and experience of breeders. It has the characteristics of long cycles and strong empiricism. Molecular selection is based on genomic variation information and makes decisions through the association between alleles and agronomic traits. It has the advantages of being fast, accurate and efficient. Whole genome selection technology is a selection technology developed on the basis of molecular marker-assisted selection breeding. Since whole genome selection predicts the phenotypic values ​​of unknown samples based on known population data, whole genome selection is also called whole genome prediction. Early whole genome predictions constructed whole genome prediction models through hybrid linear regression methods. With the rapid development of artificial intelligence, in recent years, many units have reported the use of intelligent computing methods in information science for genotype-to-phenotype prediction modeling. Machine learning methods such as LASSO regression, support vector machine, LightGBM, and deep learning methods such as convolutional neural networks and long short-term memory networks have been used to establish intelligent prediction models for crop phenotypes. The research results show great application potential and demonstrate the powerful advantages of AI in enabling the rapid development of traditional industries.

[0003] At present, whole genome prediction research based on intelligent algorithms is still in the early stages of technology development, and its development speed lags behind artificial intelligence technology. Some deep learning methods have average performance on small-scale crop population data, and the relevant algorithm framework needs to be further optimized. Therefore, developing new whole genome prediction technologies, improving prediction model performance, and improving the accuracy of phenotypic prediction are of great significance to the development of intelligent design breeding disciplines. Summary of the invention

[0004] The purpose of this application is to provide a crop phenotype prediction method and related devices based on a hybrid expert algorithm, which can improve the accuracy of crop phenotype prediction.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a crop phenotype prediction method based on a hybrid expert algorithm, comprising:

[0007] Obtain genotype data of target crop samples to be tested;

[0008] The genotype data of the target crop sample to be tested is used as input, and the phenotype prediction model is used to output the corresponding phenotype prediction data of the target crop sample to be tested; the phenotype prediction model is a model trained by using a hybrid expert algorithm.

[0009] In a second aspect, the present application provides a crop phenotype prediction device based on a hybrid expert algorithm, comprising:

[0010] A gene data acquisition module is used to obtain genotype data of target crop samples to be tested;

[0011] The phenotype prediction module is used to take the genotype data of the target crop sample to be tested as input and output the corresponding phenotype prediction data of the target crop sample to be tested using a phenotype prediction model; the phenotype prediction model is a model trained using a hybrid expert algorithm.

[0012] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned crop phenotype prediction method based on the hybrid expert algorithm.

[0013] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned crop phenotype prediction method based on a hybrid expert algorithm.

[0014] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned crop phenotype prediction method based on a hybrid expert algorithm.

[0015] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0016] The present application provides a crop phenotype prediction method and related device based on a hybrid expert algorithm, which obtains the genotype data of a target crop sample to be tested; takes the genotype data of the target crop sample to be tested as input, and uses a phenotype prediction model to output the corresponding phenotype prediction data of the target crop sample to be tested; the phenotype prediction model is a model trained by a hybrid expert algorithm. In the present invention, a phenotype prediction model is obtained based on hybrid expert algorithm training, thereby achieving accurate prediction of the performance of different agronomic traits of the target crop. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 This is an application environment diagram of a crop phenotype prediction method based on a hybrid expert algorithm in an embodiment of the present application;

[0019] Figure 2 A schematic diagram of a process flow of a crop phenotype prediction method based on a hybrid expert algorithm provided in an embodiment of the present application;

[0020] Figure 3 A schematic diagram of a framework of a hybrid expert model provided in one embodiment of the present application;

[0021] Figure 4 A schematic diagram of evaluating the Pearson correlation coefficient index of a model provided in an embodiment of the present application;

[0022] Figure 5 A schematic diagram of evaluating the mean absolute error index of a model provided in one embodiment of the present application;

[0023] Figure 6 A schematic diagram of functional modules of a crop phenotype prediction device based on a hybrid expert algorithm provided in one embodiment of the present application;

[0024] Figure 7 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0026] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0027] The crop phenotype prediction method based on the hybrid expert algorithm provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal communicates with the server through the network. The data storage system can store the data that the server needs to process. The data storage system can be set up separately, integrated on the server, or placed on the cloud or other servers. The terminal can send the genotype data of the target crop sample to be tested to the server. After the server receives the genotype data of the target crop sample to be tested, the server uses the genotype data of the target crop sample to be tested as input, and uses the phenotype prediction model to output the corresponding phenotype prediction data of the target crop sample to be tested; the phenotype prediction model is a model trained by the hybrid expert algorithm. The server can feed back the obtained phenotype prediction data to the terminal. In addition, in some embodiments, the crop phenotype prediction method based on the hybrid expert algorithm can also be implemented by a server or a terminal alone, such as the terminal can directly perform crop phenotype prediction processing on the genotype data of the target crop sample to be tested, or the server can obtain the genotype data of the target crop sample to be tested from the data storage system and perform crop phenotype prediction processing.

[0028] The terminals may be, but are not limited to, various desktop computers, laptops, smart phones, tablet computers, IoT devices and portable wearable devices. IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server may be implemented as an independent server or a server cluster consisting of multiple servers, or a cloud server.

[0029] In an exemplary embodiment, Figure 2 As shown, a crop phenotype prediction method based on a hybrid expert algorithm is provided. The method is executed by a computer device, and specifically can be executed by a computer device such as a terminal or a server alone, or can be executed by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server in the example is used to illustrate, including the following steps 101 to 102. Among them:

[0030] Step 101, obtaining genotype data of a target crop sample to be tested.

[0031] Among them, target crops include food crops, oil crops and cash crops.

[0032] Genotype data includes single base nucleotide variation, deletion / insertion variation and other molecular marker variation. Genotype data specifically refers to genetic variation site data.

[0033] Step 102, taking the genotype data of the target crop sample to be tested as input, using a phenotype prediction model to output the corresponding phenotype prediction data of the target crop sample to be tested; the phenotype prediction model is a model trained using a hybrid expert algorithm.

[0034] Among them, phenotypes are agronomic traits that researchers are interested in, such as growth period, plant type, yield, stress resistance, disease and insect resistance, as well as physiological indicators such as chlorophyll content, amino acid content, and starch content. Phenotypic data are numerical values ​​or other coded features that describe differences in agronomic traits.

[0035] The second-generation high-throughput sequencing technology was used to resequence the rice germplasm resources to be tested, and the sequencing depth was greater than 10×. After obtaining the original sequencing data, conventional sequencing data analysis methods were used to detect variations, obtain the genetic variation of the whole genome of the tested germplasm resources, and extract genetic variation information.

[0036] Implement the above steps 101 to 102 to obtain the genotype data of the target crop sample to be tested; use the genotype data of the target crop sample to be tested as input, and use the phenotype prediction model to output the corresponding phenotype prediction data of the target crop sample to be tested; the phenotype prediction model is a model trained by the hybrid expert algorithm. In the present invention, on the one hand, a phenotype prediction model is obtained based on the hybrid expert algorithm training, so as to achieve accurate prediction of the performance of different agronomic traits of the target crop; on the other hand, by performing genome prediction on the unknown phenotype sample to be tested, the sample phenotype value is quickly obtained based on the genotype data, which has the advantages of being economical and efficient.

[0037] In another exemplary embodiment of the present application, before the genotype data of the target crop samples to be tested are input into the phenotype prediction model, the genotype data of the target crop samples to be tested need to be preprocessed, including missing value filling and one-hot encoding of genetic variation data.

[0038] In another exemplary embodiment of the present application, before executing step 102 "using the genotype data as input and using the phenotype prediction model to output the corresponding phenotype prediction data of the target crop sample to be tested", the crop phenotype prediction method based on the hybrid expert algorithm further includes: constructing a phenotype prediction model, specifically including:

[0039] (1) Determine different numerical combinations of hyperparameters of the hybrid expert model; the hyperparameters include learning rate, neuron structure (number of neurons) of the expert network (MLP), dropout, and number of experts. Each hyperparameter corresponds to a preset indicator value range, and takes a value within the value range.

[0040] (2) The hybrid expert model corresponding to each hyperparameter value combination is trained, and the model performance of the trained hybrid expert model corresponding to each hyperparameter value combination is compared to obtain the hyperparameter value combination with the best model performance. Figure 3 Schematic diagram of the structure of the hybrid expert model. Figure 3 In the example, the Gate network includes a gate weight network and a noise weight network. Expert1 to Expertn refer to n experts. Phenotype refers to phenotypic data.

[0041] (3) Retraining the hybrid expert model corresponding to the optimal hyperparameter value combination to obtain the phenotype prediction model. The training set here can be the training set of the training application in step (2) or a re-set training set.

[0042] In another exemplary embodiment of the present application, in step (2), the hybrid expert model corresponding to each hyperparameter value combination is trained, and the model performance of the trained hybrid expert model corresponding to each hyperparameter value combination is compared to obtain the hyperparameter value combination with the best model performance, which specifically includes:

[0043] (2-1) For each combination of hyperparameter values, the hybrid expert model is trained and validated using the K-fold cross-validation method to obtain the model evaluation index after K-fold validation;

[0044] (2-2) The model evaluation indicators of K validations are averaged to obtain the final model evaluation indicators of the hybrid expert model corresponding to each hyperparameter value combination;

[0045] (2-3) Compare the final model evaluation indicators of the hybrid expert model corresponding to each hyperparameter value combination to obtain the hyperparameter value combination with the best model evaluation indicator.

[0046] The optimal hyperparameter value combination is screened out by comparing the actual value and predicted value of the validation set.

[0047] In another exemplary embodiment of the present application, in step (2-1), the hybrid expert model corresponding to each hyperparameter value combination is trained and validated using a K-fold cross-validation method to obtain a model evaluation index validated K times, specifically including:

[0048] (a) For each hyperparameter value combination corresponding to the hybrid expert model, obtain the training and validation sample set of the target crop.

[0049] First, we need to build a research population of the target crop. We can obtain the target crop research population by collecting germplasm resources, selected varieties, intermediate breeding materials, mutants and other biological samples with different genetic components of the target crop, and obtaining their seeds or other propagules. Then, we collect the genotype data of the samples in the research population, using methods such as gene chips and resequencing to obtain the original nucleotide base sequences of the samples.

[0050] Specifically, this example uses rice and wheat datasets from public literature. The rice dataset rice469 contains 5291 SNP markers distributed throughout the genome and 8 main agronomic trait phenotypic data (tiller number / TN, tiller angle / TA, plant height / PH, sword leaf length / SLL, sword leaf width / SLW, sword leaf length-to-width ratio / SLAR, number of spikes / SN, spike length / SL), and the wheat dataset (wheat599) contains 1279 DArT markers and 4 yield trait data (yield1, yield2, yield3, yield4).

[0051] Both the genotype data and phenotypic data in the dataset need to be preprocessed. The genotype preprocessing method is to perform quality control and alignment on the obtained raw sequence data, and further detect genetic variations, and perform missing value processing and one-hot encoding on genetic variations; genetic variations include single-base nucleotide variations, deletion / insertion variations, and other molecular marker variations. The preprocessing methods for phenotypic data are outlier and missing value processing, character data recoding, data normalization, etc. Specifically, for the rice genome dataset, one-hot encoding is used to encode the bases into numerical features; for the wheat dataset, the DArT marker itself is already in numerical form and no encoding processing is performed. For the phenotypic datasets of rice and wheat, the missing value samples are first removed, and then the data is normalized so that the model maintains the same dimensional attributes for different traits. All datasets are randomly divided into training and validation sets (376 rice samples and 480 wheat samples) and test sets (93 rice samples and 119 wheat datasets).

[0052] (b) Divide the training validation sample set into K parts of data.

[0053] (c) Each of the K pieces of data is used as a validation sample set, and the remaining K-1 pieces are used as training sample sets.

[0054] (d) For each training sample set, the genotype data in the training sample set is used as input, and the phenotype data in the training sample set is used as label to train the mixed expert model. The mini-batch stochastic gradient descent method is used for model training.

[0055] (e) Calculate the loss error (using the MSE loss function) based on the training phenotype prediction results and the corresponding actual phenotype data in the training sample set.

[0056] The training phenotype prediction result is obtained by summing and merging the prediction results of the two experts with the highest weight in the mixed expert model; the weight in the mixed expert model refers to the weight value obtained by summing the gating weight and the noise weight in the mixed expert model. Specifically, first initialize the gating weight network and the noise weight network in the mixed expert model. In order to make the experts initially assigned in the mixed expert model as balanced as possible, the weight networks are initialized to 0. Multiply each genotype feature with the gating weight network to obtain a gating weight matrix, multiply each feature with the noise weight network to obtain a noise weight matrix, and then use the Softplus activation function to nonlinearly activate the noise weight matrix, add Gaussian noise to obtain the final noise matrix, sum the gating weight matrix and the noise weight matrix and retain the two experts with the highest weight through softmax. Take the results of the predictions of the two experts with the highest weights and sum them to obtain the result output.

[0057] (f) Return to step (a) "obtaining a training and verification sample set of the target crop" until the loss error converges, and obtain K trained hybrid expert models.

[0058] (g) Use K validation sample sets to validate the corresponding trained hybrid expert models respectively, and obtain the model evaluation index of the K-times validation of the hybrid expert model corresponding to each hyperparameter numerical combination.

[0059] In another exemplary embodiment of the present application, after executing step (3) "retraining the hybrid expert model corresponding to the optimal hyperparameter numerical combination to obtain the phenotype prediction model", the crop phenotype prediction method based on the hybrid expert algorithm further includes:

[0060] The phenotype prediction model is tested and evaluated using the genotype data and phenotype data in the test sample set of the target crop.

[0061] The constructed phenotype prediction model is used to predict the test sample set to obtain the predicted phenotype value, and the Pearson correlation coefficient and mean absolute error index between the predicted value and the true value are further calculated. Figure 4 The Pearson correlation coefficient index of the phenotypic prediction model constructed using the rice469 and wheat599 datasets on the test set is displayed, indicating that the prediction accuracy of all phenotypic traits of the present invention is higher than that of the existing rrBLUP method. Figure 5The mean absolute error index of the phenotypic prediction model constructed using the rice469 and wheat599 datasets on the test set is shown, indicating that the MAE of the present invention for all phenotypic traits is smaller than that of the rrBLUP method.

[0062] In another exemplary embodiment of the present application, the genotype data of the target crop sample to be tested is used as input, and the phenotype prediction model is used to output the corresponding phenotype prediction data of the target crop sample to be tested, which specifically includes:

[0063] (i) Input the genotype data of the target crop samples to be tested into the phenotypic prediction model.

[0064] (ii) Identify the two experts with the highest weights in the phenotype prediction model.

[0065] (iii) summing the prediction results of the two experts with the highest weights in the phenotypic prediction model to obtain the phenotypic prediction data of the target crop sample to be tested.

[0066] When the phenotypic prediction model is used for prediction, the model outputs the weighted sum of the predictions of the two experts. The phenotypic prediction model includes multiple experts, and the number of experts is the value in the optimal hyperparameter combination. When different input data are input into the phenotypic prediction model, the gating network will distribute the data to two experts based on the data characteristics, and the weighted sum of the two experts is the predicted value of the current data. The two selected experts are networks that are more professional and accurate in prediction compared to other experts.

[0067] In this embodiment, a multilayer perceptron can be used for the experts in the hybrid expert algorithm, and the weights of each expert can be allocated through the gating mechanism in the hybrid expert algorithm, and the optimal hyperparameters can be determined using an automatic parameter search method. By constructing a prediction model for the desired target phenotypic trait, genome prediction can be performed on unknown phenotypic samples, thereby quickly obtaining sample phenotypic values, which has the advantages of being economical and efficient.

[0068] The present application also provides an application scenario, which applies the above-mentioned crop phenotype prediction method based on a hybrid expert algorithm. Specifically: the crop phenotype prediction method based on a hybrid expert algorithm provided in this embodiment can be applied in a crop breeding scenario. The scenario includes a data collection link, a crop phenotype prediction link and a crop breeding link. The data collection link is used to collect genotype data of the target crop. The crop phenotype prediction link is used to predict the crop phenotype based on the genotype data. The crop breeding link is used to screen out the required phenotype data based on the predicted phenotype data for subsequent breeding experiments. The crop phenotype prediction method based on a hybrid expert algorithm provided in this embodiment belongs to the crop phenotype prediction link.

[0069] Based on the same inventive concept, the embodiment of the present application also provides a device for implementing the above-mentioned method for crop phenotype prediction based on hybrid expert algorithm. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above-mentioned method, so the specific limitations in one or more embodiments of the crop phenotype prediction device based on hybrid expert algorithm provided below can refer to the limitations of the crop phenotype prediction method based on hybrid expert algorithm above, and will not be repeated here.

[0070] In an exemplary embodiment, Figure 6 As shown, a crop phenotype prediction device based on a hybrid expert algorithm is provided, comprising:

[0071] Gene data acquisition module M1 is used to obtain the genotype data of the target crop sample to be tested.

[0072] The phenotype prediction module M2 is used to take the genotype data of the target crop sample to be tested as input and output the corresponding phenotype prediction data of the target crop sample to be tested using a phenotype prediction model; the phenotype prediction model is a model trained using a hybrid expert algorithm.

[0073] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store phenotypic prediction data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a crop phenotype prediction method based on a hybrid expert algorithm is implemented.

[0074] Those skilled in the art will understand that Figure 7The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0075] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0076] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0077] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0078] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0079] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0080] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0081] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A crop phenotype prediction method based on a hybrid expert algorithm, characterized in that: The crop phenotype prediction method based on the hybrid expert algorithm includes: Obtain genotype data of target crop samples to be tested; The genotype data of the target crop sample to be tested is used as input, and the phenotype prediction model is used to output the corresponding phenotype prediction data of the target crop sample to be tested; the phenotype prediction model is a model trained by using a hybrid expert algorithm.

2. The crop phenotype prediction method based on hybrid expert algorithm according to claim 1, characterized in that: Before executing the step of "taking the genotype data as input and using the phenotype prediction model to output the corresponding phenotype prediction data of the target crop sample to be tested", the crop phenotype prediction method based on the hybrid expert algorithm further includes: constructing a phenotype prediction model, specifically including: Determine different numerical combinations of hyperparameters of the hybrid expert model; the hyperparameters include the number of experts, the learning rate, and the number of neurons in the expert network; Train the hybrid expert model corresponding to each hyperparameter value combination, and compare the model performance of the trained hybrid expert model corresponding to each hyperparameter value combination to obtain the hyperparameter value combination with the best model performance; The hybrid expert model corresponding to the optimal hyperparameter numerical combination is retrained to obtain the phenotype prediction model.

3. The crop phenotype prediction method based on hybrid expert algorithm according to claim 1, characterized in that: The hybrid expert model corresponding to each hyperparameter value combination is trained, and the model performance of the trained hybrid expert model corresponding to each hyperparameter value combination is compared to obtain the hyperparameter value combination with the best model performance, including: For each combination of hyperparameter values, the hybrid expert model is trained and validated using the K-fold cross-validation method to obtain the model evaluation index of K-fold validation. The model evaluation indicators of K validations are averaged to obtain the final model evaluation indicators of the hybrid expert model corresponding to each hyperparameter value combination; Compare the final model evaluation indicators of the hybrid expert model corresponding to each hyperparameter value combination to determine the optimal hyperparameter value combination.

4. The crop phenotype prediction method based on hybrid expert algorithm according to claim 3, characterized in that: For each combination of hyperparameter values, the hybrid expert model is trained and validated using the K-fold cross-validation method to obtain the model evaluation indicators of K-fold validation, including: For each hybrid expert model corresponding to each hyperparameter value combination, obtain a training and verification sample set of the target crop; Divide the training validation sample set into K parts of data; Each of the K pieces of data is used as a validation sample set, and the remaining K-1 pieces are used as training sample sets; For each training sample set, the genotype data in the training sample set is used as input, and the phenotype data in the training sample set is used as label to train the mixed expert model; The loss error is calculated based on the training phenotype prediction result and the real phenotype data in the corresponding training sample set; the training phenotype prediction result is obtained by summing and fusing the prediction results of the two experts with the highest weight in the hybrid expert model; the weight in the hybrid expert model refers to the weight value obtained by summing the gating weight and the noise weight in the hybrid expert model; Return to step "obtaining a training and verification sample set of the target crop" until the loss error reaches convergence, and obtain K trained hybrid expert models; K validation sample sets are used to validate the corresponding trained hybrid expert models respectively, and the model evaluation index of the K-times validation of the hybrid expert model corresponding to each hyperparameter value combination is obtained.

5. The crop phenotype prediction method based on hybrid expert algorithm according to claim 1, characterized in that: After executing the step of "retraining the hybrid expert model corresponding to the optimal hyperparameter numerical combination to obtain the phenotype prediction model", the crop phenotype prediction method based on the hybrid expert algorithm further includes: The phenotype prediction model is tested and evaluated using the genotype data and phenotype data in the test sample set of the target crop.

6. The crop phenotype prediction method based on hybrid expert algorithm according to claim 1, characterized in that: Taking the genotype data of the target crop sample to be tested as input, the phenotype prediction model is used to output the corresponding phenotype prediction data of the target crop sample to be tested, specifically including: Inputting the genotype data of the target crop samples to be tested into the phenotypic prediction model; Identify the two experts with the highest weights in the phenotypic prediction model; The prediction results of the two experts with the highest weights in the phenotypic prediction model are summed to obtain the phenotypic prediction data of the target crop sample to be tested.

7. A crop phenotype prediction device based on a hybrid expert algorithm, characterized in that: The crop phenotype prediction device based on the hybrid expert algorithm comprises: A gene data acquisition module is used to obtain genotype data of target crop samples to be tested; The phenotype prediction module is used to take the genotype data of the target crop sample to be tested as input and output the corresponding phenotype prediction data of the target crop sample to be tested using a phenotype prediction model; the phenotype prediction model is a model trained using a hybrid expert algorithm.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the crop phenotype prediction method based on a hybrid expert algorithm as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the crop phenotype prediction method based on a hybrid expert algorithm described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the crop phenotype prediction method based on a hybrid expert algorithm described in any one of claims 1 to 6 is implemented.

Images