Intelligent matching method, system and equipment for sugarcane hybrid combination and medium

Abstract

The invention relates to an intelligent matching method, system and equipment for sugarcane hybrid combinations and a medium. The method comprises the following steps: acquiring a plurality of sugarcane parents; any two of the sugarcane parents are used as a hybrid combination, and the phenotype corresponding to each hybrid combination is predicted according to the performance character of each sugarcane parent; evaluating the value of each hybrid combination according to the phenotype corresponding to each hybrid combination and a preset value evaluation rule; and determining a target hybrid combination according to the value of each hybrid combination. According to the method disclosed by the invention, the condition that the target hybrid combination is selected and matched after the progeny of each hybrid combination is really cultivated is not needed, so that the workload is reduced, the breeding period is shortened, and the selection and matching efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of bioinformatics and neural network technology, and in particular to a method, system, equipment and medium for intelligent selection of sugarcane hybrid combinations. Background Art

[0002] In the prior art, the selection of hybrid combinations in the hybrid breeding process is usually based on the performance of the offspring cultivated after hybrid breeding to reversely evaluate the corresponding hybrid combinations. This method is not only time-consuming and costly, but also inefficient. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide a method, system, device and medium for intelligent matching of sugarcane hybrid combinations, aiming to solve at least one of the above technical problems.

[0004] In the first aspect, the technical solution of the present invention to solve the above technical problem is as follows: a method for intelligent selection of sugarcane hybrid combinations, the method comprising:

[0005] Obtain multiple sugarcane parents;

[0006] Taking any two sugarcane parents from the plurality of sugarcane parents as a hybrid combination, and predicting the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents;

[0007] Evaluate the value of each of the hybrid combinations according to the phenotypes corresponding to each of the hybrid combinations and preset value evaluation rules;

[0008] According to the value of each of the hybrid combinations, the target hybrid combination is determined.

[0009] The beneficial effects of the present invention are as follows: in the process of selecting sugarcane hybrid combinations, the phenotype of each possible hybrid combination is first predicted in combination with the performance traits of the hybrid combination, and then the value of each hybrid combination is evaluated based on the phenotype corresponding to each hybrid combination and a preset value evaluation rule, and then the target hybrid combination is selected based on the value of each hybrid combination, without waiting until each hybrid combination actually breeds offspring before selecting the target hybrid combination, thereby reducing the workload, saving breeding years, and improving the selection efficiency.

[0010] Based on the above technical solution, the present invention can also be improved as follows.

[0011] Furthermore, for each of the hybrid combinations, if the parents in the hybrid combination are homozygous parents, the value of the hybrid combination is higher than the value of the hybrid combination whose parents are heterozygous parents.

[0012] Furthermore, the performance traits of each of the sugarcane parents are expressed by SNP gene data, and any two of the plurality of sugarcane parents are used as a hybrid combination, including:

[0013] Performing phenotypic identification on the SNP gene data of each of the sugarcane parents according to different phenotypic types;

[0014] According to the phenotypic identifiers corresponding to the sugarcane parents, any two sugarcane parents corresponding to the same phenotypic type are regarded as a hybrid combination, and any two sugarcane parents corresponding to different phenotypic types are regarded as a hybrid combination.

[0015] Furthermore, the prediction of the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents is determined based on a phenotype prediction model, and the phenotype prediction model is trained based on the following method:

[0016] S1, obtaining SNP gene data and phenotypic data of multiple sugarcane sample parents, and preprocessing each of the SNP gene data and phenotypic data;

[0017] S2, for each of the SNP gene data, performing numerical mapping on the preprocessed SNP gene data to obtain a gene sequence after numerical mapping;

[0018] S3, for each of the numerically mapped gene sequences, using a fixed window to perform discrete Fourier transform on the numerically mapped gene sequence, calculating a power spectrum curve of each window sequence after the transform, determining whether each window is a protein coding region, and performing feature enhancement on the protein coding region according to the determination result to obtain a feature-enhanced gene sequence;

[0019] S4, for each gene sequence after feature enhancement, the gene sequence after feature enhancement is subjected to wavelet transform to obtain high-frequency features and low-frequency features, the high-frequency features are denoised, and the low-frequency features are subjected to inverse wavelet transform. The initial network is trained using the low-frequency features, the denoised high-frequency features, the low-frequency features after inverse wavelet transform, and the preprocessed phenotypic data as labels to obtain a phenotypic prediction model.

[0020] Furthermore, the size of the gene sequence after feature enhancement is 4*N. For each gene sequence after feature enhancement, the high-frequency features are denoised, the low-frequency features are inversely transformed by wavelet transform, and the initial network is trained using the low-frequency features, the denoised high-frequency features, the low-frequency features after inverse wavelet transform, and the preprocessed phenotypic data as labels to obtain a phenotypic prediction model, including:

[0021] Perform wavelet transform on the N-dimensional sequence of each channel to obtain high-frequency features and low-frequency features, and denoise the high-frequency features;

[0022] Perform inverse wavelet transform on the transformed low-frequency features of each channel;

[0023] The low-frequency features, high-frequency features after denoising, low-frequency features after inverse wavelet transform, and preprocessed phenotypic data as labels are input into the initial network for phenotypic prediction training, and the network model parameters are optimized by the loss functions Focal_L1 and Triplet_Contrast to obtain the phenotypic prediction model.

[0024] Furthermore, the method further comprises:

[0025] Obtain transcriptomes and metabolomes of multiple sugarcane sample parents;

[0026] For each of the sugarcane sample parents, the SNP gene data, transcriptome and metabolome of the sugarcane sample parent are used as training data;

[0027] The same phenotype in all preprocessed phenotypic data is used as the target phenotype, the first similarity between the training data corresponding to each sugarcane sample parent corresponding to the target phenotype is calculated, and the graph structure data of each sugarcane sample parent is constructed using the first similarity;

[0028] The method uses low-frequency features, high-frequency features after denoising, low-frequency features after wavelet inverse transformation, and pre-processed phenotypic data as labels to train the initial network to obtain a phenotypic prediction model, including:

[0029] The initial network is trained for the first time using the low-frequency features, the high-frequency features after denoising, the low-frequency features after wavelet inverse transformation, and the preprocessed phenotypic data as labels to obtain the first predicted phenotype;

[0030] Using the graph structure data and pre-processed phenotypic data of each of the sugarcane sample parents, the initial network is trained for a second time to obtain a second predicted phenotype;

[0031] Calculating a first loss function value of the initial network according to each of the first predicted phenotypes and each of the preprocessed phenotype data;

[0032] Calculating a second loss function value of the initial network according to each of the second predicted phenotypes and each of the preprocessed phenotype data;

[0033] The initial network is trained according to the first loss function value and the second loss function value to obtain the phenotype prediction model.

[0034] Further, the value of each hybrid combination is evaluated according to the phenotype corresponding to each hybrid combination and the preset value evaluation rules, including:

[0035] For each of the hybrid combinations, the value of the hybrid combination is evaluated according to a first correspondence between a phenotype corresponding to the hybrid combination and the value evaluation rule, wherein the first correspondence is a correspondence between different hybrid combinations and the values ​​of each of the hybrid combinations.

[0036] In a second aspect, in order to solve the above technical problems, the present invention also provides a sugarcane hybrid combination intelligent matching system, the system comprising:

[0037] An acquisition module, used for acquiring multiple sugarcane parents;

[0038] A phenotype prediction module, for taking any two sugarcane parents from the plurality of sugarcane parents as a hybrid combination, and predicting the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents;

[0039] A value assessment module, used to assess the value of each of the hybrid combinations according to the phenotypes corresponding to each of the hybrid combinations and preset value assessment rules;

[0040] The hybrid combination determination module is used to determine the target hybrid combination according to the value of each hybrid combination.

[0041] In a third aspect, in order to solve the above-mentioned technical problem, the present invention also provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the method for intelligent selection of sugarcane hybrid combinations of the present application is implemented.

[0042] In a fourth aspect, in order to solve the above-mentioned technical problem, the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the intelligent matching method for sugarcane hybrid combinations of the present application is implemented.

[0043] Additional aspects and advantages of the present application will be partially given in the following description, which will become apparent from the following description, or will be understood through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for describing the embodiments of the present invention are briefly introduced below.

[0045] Figure 1 A schematic diagram of a process flow of a method for intelligent selection of sugarcane hybrid combinations provided by one embodiment of the present invention;

[0046] Figure 2 A schematic diagram of the structure of a sugarcane hybrid combination intelligent matching system provided by one embodiment of the present invention;

[0047] Figure 3 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] The principles and features of the present invention are described below. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0049] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0050] The solution provided by the embodiment of the present invention can be applied to any application scenario where intelligent matching of sugarcane hybrid combinations is required. The solution provided by the embodiment of the present invention can be executed by any electronic device, for example, it can be a user's terminal device, including at least one of the following: a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, a smart TV, and a smart car device.

[0051] The embodiment of the present invention provides a possible implementation method, such as Figure 1 As shown, a flowchart of a method for intelligent selection of sugarcane hybrid combinations is provided. The scheme can be executed by any electronic device, for example, a terminal device, or a terminal device and a server. For the convenience of description, the method provided by the embodiment of the present invention will be described below by taking the terminal device as the execution subject as an example. Figure 1 The method may include the following steps:

[0052] Step S10, obtaining a plurality of sugarcane parents;

[0053] Step S20, taking any two sugarcane parents from the plurality of sugarcane parents as a hybrid combination, and predicting the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents;

[0054] Step S30, evaluating the value of each of the hybrid combinations according to the phenotypes corresponding to the hybrid combinations and preset value evaluation rules;

[0055] Step S40, determining a target hybrid combination according to the value of each hybrid combination.

[0056] By the method of the present invention, in the process of selecting sugarcane hybrid combinations, the phenotype of each possible hybrid combination is first predicted in combination with the performance traits of the hybrid combination, and then the value of each hybrid combination is evaluated based on the phenotype corresponding to each hybrid combination and a preset value evaluation rule, and then the target hybrid combination is selected based on the value of each hybrid combination, without waiting until each hybrid combination actually breeds offspring before selecting the target hybrid combination, thereby reducing the workload, saving breeding years, and improving the selection efficiency.

[0057] The scheme of the present invention is further described below in conjunction with the following specific embodiments. In this embodiment, the method for intelligent matching of sugarcane hybrid combinations may include the following steps:

[0058] Step S10, obtaining a plurality of sugarcane parents;

[0059] Among them, multiple sugarcane parents refer to the data basis for selecting the target hybrid combination, that is, selecting the sugarcane parents that can form the target hybrid combination from multiple sugarcane parents. The multiple sugarcane parents may include homozygous parents and heterozygous parents. The multiple sugarcane parents may include parents with the same performance traits or parents with different performance traits. The performance traits of each sugarcane parent represent the characteristics of the corresponding sugarcane parent and can be used as a reference for the subsequent target hybrid combination.

[0060] Among them, multiple sugarcane parents can be provided by the MS ACCESS platform.

[0061] Step S20, taking any two sugarcane parents from the plurality of sugarcane parents as a hybrid combination, and predicting the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents;

[0062] For each hybrid combination, the hybrid combination may include parents with the same performance traits, or parents with different performance traits. In this solution, as many hybrid combinations formed by various performance traits as possible are covered.

[0063] Among them, for each of the hybrid combinations, if the parents in the hybrid combination are homozygous parents, the value of the hybrid combination is higher than the value of the hybrid combination whose parents are heterozygous parents.

[0064] Wherein, the performance trait of each sugarcane parent is expressed by SNP gene data, then in the above S20, any two sugarcane parents among the plurality of sugarcane parents are taken as a hybrid combination, including:

[0065] S201, performing phenotypic identification on the SNP gene data of each of the sugarcane parents according to different phenotypic types;

[0066] Among them, SNP genotype data can be identified according to different phenotypic types. As an example, SNP gene data of different phenotypic types can be identified based on 0, 1, and 2, wherein 0 represents SNP gene data with a high homozygous frequency, 1 represents SNP gene data corresponding to a heterozygote, and 2 represents SNP gene data with a low homozygous frequency.

[0067] Among them, SNP gene data can be determined based on genome-BLUP analysis

[0068] S202, according to the phenotypic identifiers corresponding to the sugarcane parents, any two sugarcane parents corresponding to the same phenotypic type are taken as a hybrid combination, and any two sugarcane parents corresponding to different phenotypic types are taken as a hybrid combination.

[0069] Optionally, in the above S20, according to the performance traits of each of the sugarcane parents, the prediction of the phenotype corresponding to each of the hybrid combinations is determined based on a phenotype prediction model, and the phenotype prediction model is trained based on the following method:

[0070] S1, obtaining SNP gene data and phenotypic data of multiple sugarcane sample parents, and preprocessing each of the SNP gene data and phenotypic data;

[0071] Among them, since the gene sequence of each sugarcane sample parent may have quality problems due to sequencing technology, the SNP gene data and phenotypic data are preprocessed: the gene loci are filtered according to the gene locus missing rate and the minimum allele frequency; the phenotypic data are filtered for unknown values ​​to remove data with unknown phenotypic values.

[0072] S2, for each of the SNP gene data, performing numerical mapping on the preprocessed SNP gene data to obtain a gene sequence after numerical mapping;

[0073] Among them, the gene length in the SNP gene data of each sugarcane sample parent is Mbp, and the SNP gene data of each sugarcane sample parent is numerically mapped. Let I = {A, T, C, G}, the gene sequence corresponding to the SNP gene data of each sugarcane sample parent can be expressed as: S = {S(n)|S(n)∈I, n = 0, 1, 2…N-1}, and each gene sequence is numerically mapped in the following way:

[0074]

[0075] Where b is A, G, C, and T respectively; S(n) represents the genotype at position n of the gene sequence, and t b [n] is the result of numerical mapping of the genotype at position n of gene sequence b. Thus, through the above encoding, each sugarcane sample parent will produce 4 numerical mapping sequences.

[0076] S3, for each of the numerically mapped gene sequences, using a fixed window to perform discrete Fourier transform on the numerically mapped gene sequence, calculating a power spectrum curve of each window sequence after the transform, determining whether each window is a protein coding region, and performing feature enhancement on the protein coding region according to the determination result to obtain a feature-enhanced gene sequence;

[0077] Among them, one possible implementation method of the above S3 is:

[0078] The gene sequence encoding after numerical mapping is subjected to discrete Fourier transform using a sliding window with a window size of W and a step size of L, and the power spectrum curve of each window sequence after the transformation is calculated:

[0079] For the power spectrum curve of each window, if the spectrum peak at W / 3 and the spectrum mean of the entire window are higher than the threshold, the window is considered to be a protein coding region, otherwise it is a non-coding region, and the starting position of the protein coding region is recorded;

[0080] The obtained protein coding region was enhanced by the following methods:

[0081]

[0082] Among them, P(W / 3) is the spectrum value of the coding region at W / 3. The sliding window with window W and step size L is used to merge the codes of each window after feature enhancement, that is, for the same gene position, the weighted mean of the code at that position is used as the final code of that position. Among them, the non-coding region still maintains the numerical mapping of S2.

[0083] Among them, the value range of W is 200bp to 1000bp, and the value range of L is 100bp to 300bp. The optimal value is found in the two intervals.

[0084] S4, for each gene sequence after feature enhancement, the gene sequence after feature enhancement is subjected to wavelet transform to obtain high-frequency features and low-frequency features, the high-frequency features are denoised, and the low-frequency features are subjected to inverse wavelet transform. The initial network is trained using the low-frequency features, the denoised high-frequency features, the low-frequency features after inverse wavelet transform, and the preprocessed phenotypic data as labels to obtain a phenotypic prediction model.

[0085] Optionally, the initial network can be a three-stream network, each branch of which includes three convolutional layers, two LSTM layers and one fully connected layer, and each convolutional layer is processed by a batch normalization layer and a ReLU layer.

[0086] Optionally, the size of the gene sequence after feature enhancement is 4*N. For each gene sequence after feature enhancement, in S4, high-frequency features are denoised, low-frequency features are inversely transformed by wavelet transform, and the initial network is trained using low-frequency features, denoised high-frequency features, low-frequency features after inverse wavelet transform, and preprocessed phenotypic data as labels to obtain a phenotypic prediction model, including:

[0087] S41, performing wavelet transform on the N-dimensional sequence of each channel to obtain high-frequency features and low-frequency features F1, and denoising the high-frequency features to obtain denoised high-frequency features F2;

[0088] S42, performing inverse wavelet transform on the transformed low-frequency features of each channel to obtain low-frequency features F3 after inverse wavelet transform;

[0089] S43, the low-frequency feature F1, the denoised high-frequency feature F2, the low-frequency feature F3 after wavelet inverse transformation, and the preprocessed phenotypic data as labels are input into the initial network for phenotypic prediction training, and the network model parameters are optimized through the loss functions Focal_L1 and Triplet_Contrast to obtain the phenotypic prediction model.

[0090] Optionally, the method further includes:

[0091] Obtain transcriptomes and metabolomes of multiple sugarcane sample parents;

[0092] For each of the sugarcane sample parents, the SNP gene data, transcriptome and metabolome of the sugarcane sample parent are used as training data, and each training data includes the SNP gene data, transcriptome and metabolome;

[0093] The same phenotype in all preprocessed phenotypic data is used as the target phenotype, the first similarity between the training data corresponding to each sugarcane sample parent corresponding to the target phenotype is calculated, and the graph structure data of each sugarcane sample parent is constructed using the first similarity;

[0094] Among them, in the graph structure data of each sugarcane sample parent, the nodes of the graph represent different omics (SNP gene data, transcriptome or metabolome), and the edges of the graph represent the first similarity between different omics.

[0095] It should be understood that the weight of the edge reflects the degree of association between different groups, that is, the greater the first similarity, the higher the degree of association between different groups.

[0096] In the above S4, the initial network is trained using low-frequency features, high-frequency features after denoising, low-frequency features after wavelet inverse transformation, and pre-processed phenotypic data as labels to obtain a phenotypic prediction model, including:

[0097] The initial network is trained for the first time using the low-frequency features, the high-frequency features after denoising, the low-frequency features after wavelet inverse transformation, and the preprocessed phenotypic data as labels to obtain the first predicted phenotype;

[0098] Using the graph structure data and pre-processed phenotypic data of each of the sugarcane sample parents, the initial network is trained for a second time to obtain a second predicted phenotype;

[0099] Calculating a first loss function value of the initial network according to each of the first predicted phenotypes and each of the preprocessed phenotype data;

[0100] Calculating a second loss function value of the initial network according to each of the second predicted phenotypes and each of the preprocessed phenotype data;

[0101] The initial network is trained according to the first loss function value and the second loss function value to obtain the phenotype prediction model.

[0102] Step S30, evaluating the value of each of the hybrid combinations according to the phenotypes corresponding to the hybrid combinations and preset value evaluation rules;

[0103] Among them, the above-mentioned value assessment rules can be pre-configured based on actual needs, and the value can be expressed based on numerical values. The larger the value, the larger the corresponding numerical value.

[0104] Optionally, in the above S30, the value of each hybrid combination is evaluated according to the phenotype corresponding to each hybrid combination and a preset value evaluation rule, including:

[0105] For each of the hybrid combinations, the value of the hybrid combination is evaluated according to a first correspondence between a phenotype corresponding to the hybrid combination and the value evaluation rule, wherein the first correspondence is a correspondence between different hybrid combinations and the values ​​of each of the hybrid combinations.

[0106] Among them, the above-mentioned first corresponding relationship is pre-set, which describes the corresponding relationship between different hybrid combinations and the values ​​of different hybrid combinations, that is, hybrid combinations composed of different parents may correspond to different values. The above-mentioned first corresponding relationship can be determined based on historical hybridization data, that is, from historical hybrid combinations, the corresponding values ​​of different hybrid combinations are determined.

[0107] Step S40, determining a target hybrid combination according to the value of each hybrid combination.

[0108] Among them, one possible implementation method of the above S40 is:

[0109] From the values ​​of various hybrid combinations, the hybrid combination with the highest value is selected as the target hybrid combination.

[0110] Optionally, in the present application, if the initial network is a three-stream network, the three-stream network includes three branches, each of which includes three convolutional layers, two LSTM layers and one fully connected layer, and each convolutional layer is processed by a batch normalization layer and a ReLU layer.

[0111] Optionally, one possible implementation of the method of using the low-frequency features, the high-frequency features after denoising, the low-frequency features after wavelet inverse transformation, and the preprocessed phenotypic data as labels to perform a first training on the initial network to obtain the first predicted phenotype is:

[0112] The low-frequency features, the denoised high-frequency features, the low-frequency features after wavelet inverse transformation, and the preprocessed phenotypic data as labels are input into the trained three-stream network according to the set mode, and the first predicted phenotype is output.

[0113] Among them, different modes may include:

[0114] Mode 1: Only one of the low-frequency features, high-frequency features after denoising, and low-frequency features after inverse wavelet transformation is selected and input into the corresponding branch network to obtain the first predicted phenotype;

[0115] Mode 2: The low-frequency features and the denoised high-frequency features are respectively input into the corresponding branch networks, and the two phenotypic values ​​are averaged to obtain the first predicted phenotype;

[0116] Mode 3: The high-frequency features after denoising and the low-frequency features after wavelet inverse transformation are input into the corresponding branch networks respectively, and the two phenotypic values ​​are averaged to obtain the first predicted phenotype;

[0117] Mode 4: The low-frequency features, the high-frequency features after denoising, and the low-frequency features after wavelet inverse transformation are all input into the corresponding branch network, and the three predicted phenotypic values ​​are averaged to obtain the first predicted phenotype.

[0118] Optionally, the method further includes:

[0119] Divide multiple sugarcane sample parents into test samples and training samples;

[0120] Selecting multiple training models, the training models at least include a GBLUP model and a RRBLUP model; the input of the training model is the training data of multiple sugarcane sample parents, and the output is the actual phenotypic value;

[0121] Training the training model using the training data of all the training samples and the corresponding actual phenotypic values ​​to obtain multiple optimized models;

[0122] Inputting the training data of all the test samples into each of the optimization models respectively, obtaining a third predicted phenotype corresponding to each of the optimization models and corresponding to the first characteristic parameter, wherein the first characteristic parameter is a characteristic parameter corresponding to one of the same sugarcane sample parents among all the sugarcane sample parents;

[0123] Calculate the correlation value corresponding to each optimization model according to the third predicted phenotype of all the test samples corresponding to each optimization model and the actual phenotype value;

[0124] Determining a third loss function value according to each of the correlation values;

[0125] The step of training the initial network according to the first loss function value and the second loss function value to obtain the phenotype prediction model comprises:

[0126] The initial network is trained according to the first loss function value, the second loss function value and the third loss function value to obtain the phenotype prediction model.

[0127] During the training process, the goal is to minimize the correlation value corresponding to the third loss function value.

[0128] In this scheme, genetic methods such as G-BLUP analysis, genome-BLUP analysis and MiXBLUP are used to predict the phenotype and evaluate the value of common sugarcane breeding parents and combinations, and to achieve the field performance prediction of various unconfigured hybrid combinations.

[0129] In this scheme, the true phenotype of the parents can also be encoded on the barcode, that is, one parent corresponds to one barcode.

[0130] In this scheme, based on the phenotypic prediction and value evaluation of common sugarcane breeding parents and hybrid combinations, a sugarcane hybrid combination matching database management system is established to achieve automatic matching of high-value combinations according to the hybridization plan.

[0131] In this scheme, based on the hybrid combination selection database, a suitable machine learning model (phenotype prediction model) was constructed, and the Python programming language was used to develop a fully automatic phenotype prediction and value assessment process for parents and hybrid offspring.

[0132] For the combination configured by the intelligent selection software for sugarcane hybrid combinations, combined with the previous phenotypic data, the mixed linear model is used to obtain the phenotypic prediction value of the trait. For the combination with good prediction effect, the phenotypic prediction is directly used to judge the performance of the offspring, thereby reducing the workload and saving breeding years.

[0133] Based on Figure 1Based on the same principle as the method shown in , the embodiment of the present invention also provides a sugarcane hybrid combination intelligent matching system 20, such as Figure 2 As shown in , the sugarcane hybrid combination intelligent matching system 20 may include an acquisition module 210, a phenotype prediction module 220, a value assessment module 230 and a hybrid combination determination module 240, wherein:

[0134] An acquisition module 210 is used to acquire a plurality of sugarcane parents;

[0135] The phenotype prediction module 220 is used to take any two sugarcane parents from the plurality of sugarcane parents as a hybrid combination, and predict the phenotype corresponding to each of the hybrid combinations according to the performance characteristics of each of the sugarcane parents;

[0136] A value evaluation module 230, used to evaluate the value of each of the hybrid combinations according to the phenotypes corresponding to each of the hybrid combinations and preset value evaluation rules;

[0137] The hybrid combination determination module 240 is used to determine a target hybrid combination according to the value of each hybrid combination.

[0138] Optionally, for each of the hybrid combinations, if the parents in the hybrid combination are homozygous parents, the value of the hybrid combination is higher than the value of a hybrid combination whose parents are heterozygous parents.

[0139] Optionally, the performance trait of each sugarcane parent is represented by SNP gene data, and when any two of the plurality of sugarcane parents are used as a hybrid combination, the phenotype prediction module 220 is specifically used to:

[0140] Performing phenotypic identification on the SNP gene data of each of the sugarcane parents according to different phenotypic types;

[0141] According to the phenotypic identifiers corresponding to the sugarcane parents, any two sugarcane parents corresponding to the same phenotypic type are regarded as a hybrid combination, and any two sugarcane parents corresponding to different phenotypic types are regarded as a hybrid combination.

[0142] Optionally, the prediction of the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents is determined based on a phenotype prediction model, and the phenotype prediction model is obtained by training based on the following model training module, and the model training module is used to:

[0143] Acquiring SNP gene data and phenotypic data of multiple sugarcane sample parents, and preprocessing each of the SNP gene data and phenotypic data;

[0144] For each of the SNP gene data, numerical mapping is performed on the preprocessed SNP gene data to obtain a gene sequence after numerical mapping;

[0145] For each of the numerically mapped gene sequences, a fixed window is used to perform discrete Fourier transform on the numerically mapped gene sequence, a power spectrum curve of each window sequence after the transform is calculated, whether each window is a protein coding region is determined, and according to the determination result, feature enhancement is performed on the protein coding region to obtain a feature-enhanced gene sequence;

[0146] For each gene sequence after feature enhancement, the gene sequence after feature enhancement is subjected to wavelet transform to obtain high-frequency features and low-frequency features, the high-frequency features are denoised, and the low-frequency features are subjected to inverse wavelet transform. The initial network is trained using the low-frequency features, the denoised high-frequency features, the low-frequency features after inverse wavelet transform, and the preprocessed phenotypic data as labels to obtain a phenotypic prediction model.

[0147] Optionally, the size of the gene sequence after feature enhancement is 4*N. For each gene sequence after feature enhancement, the model training module denoises the high-frequency features, performs inverse wavelet transform on the low-frequency features, uses the low-frequency features, the denoised high-frequency features, the low-frequency features after inverse wavelet transform, and the preprocessed phenotypic data as labels to train the initial network to obtain the phenotypic prediction model, which is specifically used for:

[0148] Perform wavelet transform on the N-dimensional sequence of each channel to obtain high-frequency features and low-frequency features, and denoise the high-frequency features;

[0149] Perform inverse wavelet transform on the transformed low-frequency features of each channel;

[0150] The low-frequency features, high-frequency features after denoising, low-frequency features after inverse wavelet transform, and preprocessed phenotypic data as labels are input into the initial network for phenotypic prediction training, and the network model parameters are optimized by the loss functions Focal_L1 and Triplet_Contrast to obtain the phenotypic prediction model.

[0151] Optionally, the system further includes:

[0152] The graph structure data determination module is used to:

[0153] Obtain transcriptomes and metabolomes of multiple sugarcane sample parents;

[0154] For each of the sugarcane sample parents, the SNP gene data, transcriptome and metabolome of the sugarcane sample parent are used as training data;

[0155] The same phenotype in all preprocessed phenotypic data is used as the target phenotype, the first similarity between the training data corresponding to each sugarcane sample parent corresponding to the target phenotype is calculated, and the graph structure data of each sugarcane sample parent is constructed using the first similarity;

[0156] The above-mentioned model training module uses low-frequency features, high-frequency features after denoising, low-frequency features after wavelet inverse transformation, and pre-processed phenotypic data as labels to train the initial network and obtain a phenotypic prediction model, which is specifically used for:

[0157] The initial network is trained for the first time using the low-frequency features, the high-frequency features after denoising, the low-frequency features after wavelet inverse transformation, and the preprocessed phenotypic data as labels to obtain the first predicted phenotype;

[0158] Using the graph structure data and pre-processed phenotypic data of each of the sugarcane sample parents, the initial network is trained for a second time to obtain a second predicted phenotype;

[0159] Calculating a first loss function value of the initial network according to each of the first predicted phenotypes and each of the preprocessed phenotype data;

[0160] Calculating a second loss function value of the initial network according to each of the second predicted phenotypes and each of the preprocessed phenotype data;

[0161] The initial network is trained according to the first loss function value and the second loss function value to obtain the phenotype prediction model.

[0162] Optionally, when evaluating the value of each hybrid combination according to the phenotype corresponding to each hybrid combination and a preset value evaluation rule, the value evaluation module 230 is specifically used to:

[0163] For each of the hybrid combinations, the value of the hybrid combination is evaluated according to a first correspondence between a phenotype corresponding to the hybrid combination and the value evaluation rule, wherein the first correspondence is a correspondence between different hybrid combinations and the values ​​of each of the hybrid combinations.

[0164] The sugarcane hybrid combination intelligent matching system of the embodiment of the present invention can execute the sugarcane hybrid combination intelligent matching method provided by the embodiment of the present invention, and the implementation principle is similar. The actions executed by each module and unit in the sugarcane hybrid combination intelligent matching system in each embodiment of the present invention correspond to the steps in the sugarcane hybrid combination intelligent matching method in each embodiment of the present invention. For the detailed functional description of each module of the sugarcane hybrid combination intelligent matching system, please refer to the description of the corresponding sugarcane hybrid combination intelligent matching method shown in the above text, which will not be repeated here.

[0165] Among them, the above-mentioned sugarcane hybrid combination intelligent matching system can be a computer program (including program code) running in a computer device. For example, the sugarcane hybrid combination intelligent matching system is an application software; the system can be used to execute the corresponding steps in the method provided in the embodiment of the present invention.

[0166] In some embodiments, the sugarcane hybrid combination intelligent matching system provided by the embodiments of the present invention can be implemented in a combination of software and hardware. As an example, the sugarcane hybrid combination intelligent matching system provided by the embodiments of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the sugarcane hybrid combination intelligent matching method provided by the embodiments of the present invention. For example, the processor in the form of a hardware decoding processor can adopt one or more application specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field programmable gate arrays (FPGAs) or other electronic components.

[0167] In other embodiments, the sugarcane hybrid combination intelligent matching system provided by the embodiments of the present invention can be implemented in a software manner. Figure 2 The invention shows a sugarcane hybrid combination intelligent matching system stored in a memory, which can be software in the form of a program and a plug-in, and includes a series of modules, including an acquisition module 210, a phenotype prediction module 220, a value assessment module 230 and a hybrid combination determination module 240, for implementing the sugarcane hybrid combination intelligent matching method provided in an embodiment of the present invention.

[0168] The modules involved in the embodiments of the present invention may be implemented by software or hardware, wherein the name of a module does not limit the module itself in some cases.

[0169] Based on the same principle as the method shown in the embodiments of the present invention, an electronic device is also provided in the embodiments of the present invention, which may include but is not limited to: a processor and a memory; the memory is used to store a computer program; the processor is used to execute the method shown in any embodiment of the present invention by calling the computer program.

[0170] In an alternative embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3The electronic device 4000 shown includes: a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, such as through a bus 4002. Optionally, the electronic device 4000 may also include a transceiver 4004, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0171] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of the present invention. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0172] The bus 4002 may include a path to transmit information between the above components. The bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 4002 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0173] The memory 4003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to this.

[0174] The memory 4003 is used to store the application code (computer program) for executing the solution of the present invention, and the execution is controlled by the processor 4001. The processor 4001 is used to execute the application code stored in the memory 4003 to implement the content shown in the above method embodiment.

[0175] The electronic device may also be a terminal device. Figure 3 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0176] An embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding contents of the aforementioned method embodiment.

[0177] According to another aspect of the present invention, a computer program product or a computer program is also provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in the implementation modes of the above-mentioned various embodiments.

[0178] Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0179] It should be understood that the flow charts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the module, the program segment, or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0180] The computer-readable storage medium provided by the embodiment of the present invention may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, device, or device.

[0181] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device executes the method shown in the above embodiment.

[0182] The above description is only a preferred embodiment of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the disclosure scope involved in the present invention is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in the present invention (but not limited to) by each other.

Claims

1. A method for intelligent selection of sugarcane hybrid combinations, characterized in that: The following steps are involved: Obtain multiple sugarcane parents; Taking any two sugarcane parents from the plurality of sugarcane parents as a hybrid combination, and predicting the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents; Evaluate the value of each of the hybrid combinations according to the phenotypes corresponding to each of the hybrid combinations and preset value evaluation rules; According to the value of each hybrid combination, the target hybrid combination is determined.

2. The method according to claim 1, characterized in that For each of the hybrid combinations, if the parents in the hybrid combination are homozygous parents, the value of the hybrid combination is higher than the value of the hybrid combination whose parents are heterozygous parents.

3. The method according to claim 1, characterized in that The performance traits of each sugarcane parent are expressed by SNP gene data, and any two sugarcane parents among the plurality of sugarcane parents are used as a hybrid combination, including: Performing phenotypic identification on the SNP gene data of each of the sugarcane parents according to different phenotypic types; According to the phenotypic identifiers corresponding to the sugarcane parents, any two sugarcane parents corresponding to the same phenotypic type are regarded as a hybrid combination, and any two sugarcane parents corresponding to different phenotypic types are regarded as a hybrid combination.

4. The method according to any one of claims 1 to 3, characterized in that The prediction of the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents is determined based on a phenotype prediction model, and the phenotype prediction model is trained based on the following method: S1, obtaining SNP gene data and phenotypic data of multiple sugarcane sample parents, and preprocessing each of the SNP gene data and phenotypic data; S2, for each of the SNP gene data, performing numerical mapping on the preprocessed SNP gene data to obtain a gene sequence after numerical mapping; S3, for each of the numerically mapped gene sequences, using a fixed window to perform discrete Fourier transform on the numerically mapped gene sequence, calculating a power spectrum curve of each window sequence after the transform, determining whether each window is a protein coding region, and performing feature enhancement on the protein coding region according to the determination result to obtain a feature-enhanced gene sequence; S4, for each gene sequence after feature enhancement, the gene sequence after feature enhancement is subjected to wavelet transform to obtain high-frequency features and low-frequency features, the high-frequency features are denoised, and the low-frequency features are subjected to inverse wavelet transform. The initial network is trained using the low-frequency features, the denoised high-frequency features, the low-frequency features after inverse wavelet transform, and the preprocessed phenotypic data as labels to obtain a phenotypic prediction model.

5. The method according to claim 4, characterized in that The size of the gene sequence after feature enhancement is 4*N. For each gene sequence after feature enhancement, the high-frequency features are denoised, the low-frequency features are inversely transformed by wavelet transform, and the initial network is trained using the low-frequency features, the denoised high-frequency features, the low-frequency features after inverse wavelet transform, and the pre-processed phenotypic data as labels to obtain a phenotypic prediction model, including: Perform wavelet transform on the N-dimensional sequence of each channel to obtain high-frequency features and low-frequency features, and denoise the high-frequency features; Perform inverse wavelet transform on the transformed low-frequency features of each channel; The low-frequency features, high-frequency features after denoising, low-frequency features after inverse wavelet transform, and preprocessed phenotypic data as labels were input into the initial network for phenotypic prediction training, and the network model parameters were optimized by the loss functions Focal_L1 and Triplet_Contrast to obtain the phenotypic prediction model.

6. The method according to claim 4, characterized in that The method further comprises: Obtain transcriptomes and metabolomes of multiple sugarcane sample parents; For each of the sugarcane sample parents, the SNP gene data, transcriptome and metabolome of the sugarcane sample parent are used as training data; The same phenotype in all preprocessed phenotypic data is used as the target phenotype, the first similarity between the training data corresponding to each sugarcane sample parent corresponding to the target phenotype is calculated, and the graph structure data of each sugarcane sample parent is constructed using the first similarity; The method uses low-frequency features, high-frequency features after denoising, low-frequency features after wavelet inverse transformation, and pre-processed phenotypic data as labels to train the initial network to obtain a phenotypic prediction model, including: The initial network is trained for the first time using the low-frequency features, the high-frequency features after denoising, the low-frequency features after wavelet inverse transformation, and the preprocessed phenotypic data as labels to obtain the first predicted phenotype; Using the graph structure data and pre-processed phenotypic data of each of the sugarcane sample parents, the initial network is trained for a second time to obtain a second predicted phenotype; Calculating a first loss function value of the initial network according to each of the first predicted phenotypes and each of the preprocessed phenotype data; Calculating a second loss function value of the initial network according to each of the second predicted phenotypes and each of the preprocessed phenotype data; The initial network is trained according to the first loss function value and the second loss function value to obtain the phenotype prediction model.

7. The method according to any one of claims 1 to 3, characterized in that The step of evaluating the value of each hybrid combination according to the phenotype corresponding to each hybrid combination and a preset value evaluation rule includes: For each of the hybrid combinations, the value of the hybrid combination is evaluated according to a first correspondence between a phenotype corresponding to the hybrid combination and the value evaluation rule, wherein the first correspondence is a correspondence between different hybrid combinations and the values ​​of each of the hybrid combinations.

8. A sugarcane hybrid combination intelligent selection system, characterized in that: include: An acquisition module, used for acquiring multiple sugarcane parents; A phenotype prediction module, for taking any two sugarcane parents from the plurality of sugarcane parents as a hybrid combination, and predicting the phenotype corresponding to each of the hybrid combinations according to the performance traits of each of the sugarcane parents; A value assessment module, used to assess the value of each of the hybrid combinations according to the phenotypes corresponding to each of the hybrid combinations and preset value assessment rules; The hybrid combination determination module is used to determine the target hybrid combination according to the value of each hybrid combination.

9. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images