Crop mitochondria editing site prediction method and system and storage medium

Through a deep learning-based method of convolutional neural network, the genomic sequences and RNA editing site information of multiple species are used to predict the C-to-U RNA editing site of crop mitochondria, solving the low prediction accuracy and species differences in the prior art. Unconsidered issues, higher precision and reliable RNA editing site predictions are achieved.

CN119068978AActive Publication Date: 2024-12-03HUNAN AGRI UNIV
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202310631135.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-31
Publication Date
2024-12-03
Estimated Expiration
2043-05-31

AI Technical Summary

Technical Problem

The prior art has low prediction accuracy in RNA editing site prediction and fails to effectively consider differences between species. Traditional machine learning methods rely on artificially set feature extraction, resulting in insufficient accuracy.

Method used

Based on deep learning method, the training set is constructed using mitochondrial genome sequences and C-to-U RNA editing site annotation information of multiple species. Through independent learning characteristics, the mitochondrial C-to-U crops are predicted. RNA editing sites, considering species closeness to improve prediction accuracy.

Benefits of technology

It improves the accuracy of RNA editing site prediction, can better fit the potential coding mechanism of C-to-U RNA editing in different species, and enhances the reliability and accuracy of the prediction results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068978A_ABST
    Figure CN119068978A_ABST
Patent Text Reader

Abstract

The invention discloses a crop mitochondrial editing site prediction method, a crop mitochondrial editing site prediction system and a storage medium. The method comprises the following steps: acquiring mitochondrial genome sequences of multiple species and corresponding C-to-U RNA editing site annotation information; taking the editing site as a center, and taking fragments with set lengths as positive samples at the upstream and downstream of the mitochondrial genome sequence and the corresponding C-to-U RNA editing site annotation information sequence; taking any site without editing site annotation as a center, and taking fragments with set lengths as negative samples at the upstream and downstream of the mitochondrial genome sequence and the corresponding C-to-U RNA editing site annotation information; and constructing a training set of each species by using a part of positive samples and negative samples corresponding to each species, and taking the training set of each species as input of a convolutional neural network to obtain an optimal prediction model corresponding to each species. The method is high in prediction precision, and can better fit potential coding mechanisms of C-to-U RNA editing of different species.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to editing site prediction technology, in particular to a crop mitochondrial editing site prediction method, system and storage medium. Background Art

[0002] RNA editing refers to a predetermined modification of codons caused by changes in nucleotides at the RNA level. It is achieved by replacing, inserting or deleting nucleotides. It is a necessary processing step for mitochondrial genes to produce functional proteins. It is also one of the important ways for the cell nucleus to regulate mitochondrial gene expression. Since it was first discovered in the 20th century, research on the mechanism of RNA editing has always received great attention.

[0003] C-to-U RNA editing is the most common type in crops. Efficiently, quickly and economically identifying C-to-U RNA editing sites remains a huge challenge. Traditional RNA editing site identification experiments are complex, multi-step and costly. Although the results are relatively accurate, they can often only identify a single gene in the coding region. In fact, RNA editing in non-coding regions can also regulate gene expression. Secondly, there are many potential interferences in the entire experimental process, such as PCR errors, DNA editing interference from SNPs and mutants, and the effects of gene expression levels in different types of cells on transcripts. During identification, a series of technical means are needed to reduce the impact of erroneous results.

[0004] Thanks to the development of machine learning, new research ideas have been provided for the high-precision prediction of editing sites. Many scholars have made a lot of attempts to predict RNA editing sites through traditional machine learning methods, such as decision trees, random forests, support vector machines, etc., but the prediction accuracy is generally low, and the differences between species are rarely considered. It was not until 2022 that Qin et al. speculated that the C-to-U RNA editing mechanisms of closely related species may have greater similarities through a variety of feature extraction modeling methods, which provided a new research idea for the prediction of C-to-U RNA editing cross-editing sites.

[0005] The artificially set feature extraction method has a great impact on the accuracy of traditional machine learning models. Thanks to the deep learning autonomous feature learning mode, although the learned features are too abstract and bring trouble to feature interpretation, they can avoid the limitations of human thinking, restore the occurrence law of things to the greatest extent, and improve the fitting ability of the model. In 1986, Rumelhart et al. proposed a network (Back Propagation Network, BPN) (Learning representations by back-propagating errors), and deep learning has gradually been taken seriously by researchers. As a deep learning model, the Convolutional Neural Network (CNN) has the characteristics of local connection, weight sharing, pooling operation and multi-layer structure. Its powerful automatic feature learning ability is most widely used in the field of image and text recognition.

[0006] Genomic data is essentially a text sequence composed of bases. Existing technologies have not explored the interspecies relationship of the C-to-U RNA editing mechanism, resulting in limited prediction accuracy. Summary of the invention

[0007] The technical problem to be solved by the present invention is to provide a crop mitochondrial editing site prediction method, system and storage medium in view of the deficiencies in the existing technology, so as to improve the editing site prediction accuracy.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for predicting crop mitochondrial editing sites, comprising the following steps:

[0009] S1. Obtain mitochondrial genome sequences and corresponding C-to-U RNA editing site annotation information of multiple species; take the editing site as the center, and take fragments of set lengths upstream and downstream of the mitochondrial genome sequence and the corresponding C-to-U RNA editing site annotation information sequence as positive samples; take any site without editing site annotation as the center, and take fragments of set lengths upstream and downstream of the mitochondrial genome sequence and the corresponding C-to-U RNA editing site annotation information as negative samples;

[0010] S2. Using some positive samples and negative samples corresponding to each species to construct a training set for each species, using the training set for each species as the input of the convolutional neural network, using the probability that the central site of each sample is an editing site as the output of the convolutional neural network, training the convolutional neural network, and obtaining the optimal prediction model corresponding to each species;

[0011] The convolutional neural network includes an input layer, an embedding layer, a first convolutional layer, a second convolutional layer, a downsampling layer and an output layer which are connected in sequence.

[0012] This paper uses the genome sequence to predict crop mitochondrial C-to-U RNA editing events for the first time through deep learning methods. Compared with traditional machine learning modeling prediction, this model has the advantage of autonomous learning features, avoids the limitations of artificially designed feature extraction, can restore the laws of C-to-U RNA editing phenomena in biology to the greatest extent, and improves the accuracy of editing site prediction.

[0013] After interspecies analysis of the model's prediction results, it was found that the C-to-U RNA editing mechanisms of closely related species were more similar. This suggests that when predicting new species, this finding should be used as a condition for screening models to make the prediction results more reliable.

[0014] In the present invention, during the training process of the convolutional neural network, the learning rate is set to 0.001.

[0015] In the present invention, the sizes of the first convolutional layer and the second convolutional layer are set to 16-30.

[0016] In the present invention, the number of the positive samples and the number of the negative samples are equal, so as to minimize the risk of sub-optimization (i.e., poor optimization) of parameters and evaluation distortion.

[0017] In order to further improve the prediction reliability, in step S1 of the present invention:

[0018] 1) determining whether the genus of the multiple species is the same, fusing the positive samples and negative samples of all species of the same genus to obtain a first fused sample;

[0019] 2) Determine whether the families of the remaining species of different genera are the same, and fuse the positive samples and negative samples of all species of the same family to obtain a second fused sample;

[0020] 3) Determine whether the orders of the remaining species of different families are the same, and fuse the positive samples and negative samples of all species of the same order to obtain a third fused sample;

[0021] 4) Determine whether the classes of the remaining species of different orders are the same, and fuse the positive samples and negative samples of all species of the same class to obtain the fourth fused sample;

[0022] Then the training set acquisition process in step S2 corresponds to:

[0023] A training set is constructed using the first fusion sample / the second fusion sample / the third fusion sample / the fourth fusion sample.

[0024] The method of the present invention further comprises:

[0025] S3. For newly acquired species, the C-to-U RNA editing sites are predicted using the optimal prediction model corresponding to the species.

[0026] In order to further improve the reliability of the prediction results, in the above step S3, in step S3, for the species to be predicted, the optimal prediction model that is closest to the species to be predicted is obtained according to the genus, family, order and class of the species to be predicted, and the C-to-U RNA editing site of the predicted species is predicted using the optimal prediction model.

[0027] As an inventive concept, the present invention also provides a crop mitochondrial editing site prediction system, which comprises:

[0028] one or more processors;

[0029] A memory having one or more programs stored thereon, which, when executed by the one or more processors, enables the one or more processors to implement the steps of the above method of the present invention.

[0030] As an inventive concept, the present invention further provides a computer-readable storage medium storing a computer program, which implements the steps of the above method of the present invention when executed by a processor.

[0031] Compared with the prior art, the present invention has the following beneficial effects: the present invention has high prediction accuracy and can better fit the potential coding mechanism of C-to-U RNA editing in different species. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is the hyperparameter range of the model in the embodiment of the present invention;

[0033] Figure 2 The model structure of the embodiment of the present invention;

[0034] Figure 3 The interspecific relationship of the Euclidean distance model of 13 crops in the embodiment of the present invention;

[0035] Figure 4 The interspecific relationship of the Euclidean distance model of dicotyledonous plants in the embodiment of the present invention;

[0036] Figure 5 The interspecific prediction and interspecific ANI relationship of 13 crops in the embodiment of the present invention;

[0037] Figure 6 The predicted sites of the mitochondrial full editing sites of C. pepo in the present invention are compared with the NCBI annotated sites; Figure 6 In the figure, the blue dots are C-to-U RNA editing sites annotated by NCBI, and the green dots are the prediction results of DEEPReditor-CMG. The darker the color, the more sites there are near this region. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] Example 1

[0040] This embodiment provides a method for predicting crop mitochondrial editing sites.

[0041] In this example, the complete mitochondrial genomes of 13 crops (A. thaliana, Arabidopsis thaliana ecotype Col-0, A. thaliana ecotype Col-0, B. napus, Capsella bursa-pastoris, C. bursa-pastoris, Citrullus lanatus, C. lanatus, Cucurbita pepo, C. pepo, Lactuca sativa var. Capitala, L. sativa var. Capitala, Nelumbo nucifera, N. nucifera, O. sativa, Raphanus sativus, R. sativus, Solanum tuberosum, S. tuberosum, Triticum aestivum, T. aestivum and Zea mays, Z. mays) and their annotation information were obtained from the National Center for Biotechnology Information. Information (NCBI) (Table 1). Since the mitochondrial reference genome NC_036024.1 of T. aestivum has no editing site annotation information, but 11 mitochondrial genome fragments, including X57965, X57968, Y13920, AB022060, AJ295996, X69720, X75036, X79609, X69205, AF082025 and AJ535507, have editing site annotation information, the samples of T. aestivum are derived from the above 11 fragments, and the samples of other species are all derived from the whole mitochondrial genome (with editing site annotation).

[0042] Table 1 Classification information and sample numbers of thirteen crops

[0043]

[0044] In this example, according to the mitochondrial genome sequences of 13 species and the corresponding C-to-U RNA editing site annotation information, the positive samples were all centered on the editing site C, and 250 bp were taken upstream and downstream (Du P, He T, Li Y. Prediction of C-to-U RNA editing sites in higher plant mitochondria using only nucleotide sequence features. Biochem Biophys Res Commun. 2007, 358: 336-341. and Qin S, Fan Y, Hu S, et al. iPReditor-CMG: Improving a predictive RNAeditor for crop mitochondrial genomes using genomic sequence features and an optimal support vector machine.Phytochemistry.2022,200:113222.); the negative sample is centered on C with no editing site annotation, and 250bp are taken upstream and downstream; the sequence of G annotated as a C-to-U editing site is complemented (according to the central dogma, sequence complementation is achieved through python programming: A→T, C→G, T→A, G→C). The number of negative samples is equal to the number of positive samples and the extraction process is random.

[0045] The training set, validation set and test set are composed of equal amounts of positive samples and negative samples, and the ratio of training set to independent test set and validation set to training set are both 8:2. The training set and independent test set are randomly divided five times using different random factors. The training data set of this embodiment is balanced (data balance: the ratio of positive samples to negative samples in the training set is 1:1, and the ratio of positive samples to negative samples in the test set is also 1:1), which can minimize the risk of poor optimization of parameters and evaluation distortion (Krawczyk, B. Learning from imbalanced data: open challenges and future directions. Prog Artif Intell. 2016, 5: 221-232.).

[0046] In this embodiment, based on Tensorflow (Goldiegadde.TensorFlow 2.0.0. https: / / github.com / tensorflow / tensorflow / releases / tag / v2.0.0.) and Keras (Keras. https: / / github.com / keras-team / keras .) Build a CNN model. The structure of the model is input layer + embedding layer + two convolutional layers and pooling layer + regularization layer (30%) + output layer, batch_size = 50, optimizer is Adam, loss function is binary first-order cross entropy function (structure see Figure 2 ), the evaluation index is ACC (Formula 1).

[0047] The range hyperparameter optimization process of this embodiment is mainly range search. Some preliminary ranges can be given by random search first, and then the hyperparameters can be optimized by grid search. The advantage of this is that the hyperparameter range can be quickly narrowed, the computing power consumption is greatly reduced, time and cost are saved, and it is more suitable for fast training models on portable hosts.

[0048] In this embodiment, the effectiveness of this solution is verified by the following interspecies relationship indicators. The interspecies relationship indicators are as follows:

[0049] (1) Interspecies ANI values

[0050] "Average nucleotide identity (ANI)" is to replace DDH by imitating the experimental process of DNA-DNA hybridization (DDH) at the nucleic acid level, thereby calculating the similarity of all orthologous protein-coding genes between two genomes (Goris J, Konstantinidis KT, Klappenbach JA, et al. DNA-DNA hybridization values ​​and their relationship to whole-genome sequence similarities. Int J Syst Evol Microbiol. 2007, 57: 81-91.). "Basic local alignment search tool (Basic local alignment search tool, Blast)" is an algorithm for calculating ANI (Altschul SF, Gish W, Miller W, et al. Basic local alignment search tool. J Mol Biol. 1990, 215: 403-410.), which is widely used in sequence alignment research based on its advantages of fast alignment speed and high alignment accuracy. ANI analysis was performed on the mitochondrial genomes of 13 species (T. aestivum was the mitochondrial reference genome, ID: NC_036024.1). The Blast method was used to compare the 13 species in pairs, and a total of 78 groups of comparison results were obtained. These values ​​are referred to as "inter-species ANI values" in this embodiment. When the "inter-species ANI value" of two species is larger, it means that the two species are more closely related, and vice versa, the relationship is more distant.

[0051] (2) Interspecific prediction value

[0052] In the model building part, after selecting the optimal CNN model for each of the 13 species, interspecies prediction was performed on the 13 species. For example, when the model of species A predicted species B, it was recorded as "interspecies prediction value +", and vice versa, when the model of species B predicted species A, it was recorded as "interspecies prediction value -". All samples of the species were used for prediction, and the evaluation index was ACC. A total of 2*78 groups of interspecies prediction results were obtained. When the "interspecies prediction value" of two species is larger, it means that the two species are more closely related, and vice versa, the relationship is more distant.

[0053] (3) Model Euclidean distance

[0054] In the step of designing interspecific prediction values, the last sigmoid layer of the optimal CNN model of each species is removed, and then the penultimate layer is used as the output layer. Assuming that the output of each model will be a vector of length 256, then the vector can maximize the difference between the models; then the positive samples and negative samples of the 13 species are used as the input layer of each model, and finally 13 positive sample vectors and 13 negative sample vectors are obtained (here, considering the stability of the vector and reducing the error, the input layer is all samples of each species, and then the vector matrix is ​​averaged vertically); finally, the Euclidean distance between the 13 vectors of the positive and negative samples is calculated (Formula 2), and a total of 2*78 groups of Euclidean distances can be obtained. When the "model Euclidean distance" of two species is smaller, it means that the two species are more closely related, and vice versa, the relationship is more distant.

[0055]

[0056]

[0057] TP: true positive, FN: false negative, TN: true negative, FP: false positive, d: Euclidean distance, a i : vector A i-th position, b i : The i-th bit of vector B.

[0058] Taking the species C.pepo (randomly selected) as an example, a single-layer convolutional network model was built, and the learning rates were selected as 0.1, 0.001, and 0.0001. The convolution kernel size in the convolution layer was 10-26, with a step size of 2; the embedding layer and the convolution kernel size changed synchronously; the filter was 32-256, increasing by 2 times; the size of the fully connected layer was 32-256, increasing by 2 times; in this way, 135 single-layer convolution models were set for each learning rate.

[0059] When the learning rate is 0.01, 65 out of 135 models cannot fit (val_ACC=0.5), while when the learning rate is 0.001 and 0.0001, the model fitting is stronger and more stable, and when the learning rate is 0.001, it is significantly better than when the learning rate is 0.0001 ( Figure 1 (a) in the figure); the influence of convolution kernel size on the validation set accuracy is not obvious, but it is roughly positively correlated. When it is less than 14, the model result is poor, and when it is 26, the validation set accuracy is optimal ( Figure 1 (b) in the figure); the larger the filter, the more accurate the validation set will be ( Figure 1 (c) in the figure); the size of the fully connected layer has little effect on the accuracy of the validation set, and the result is better when it is 256 ( Figure 1 (d) in the figure.

[0060] According to the results of the single-layer convolutional neural network, the learning rate is set to 0.001, and the convolution kernel size of each layer is set to 10-26 with a step size of 2; the first layer "filter" is 256, the second layer is 128 and 256, and the third layer is 64 and 256; the size of the fully connected layer is 256; there are three types of convolutional neural network models, each model is configured with 18 different parameters ( Figure 1 (e) in the figure). There are 2 and 7 models that cannot be fitted in the two-layer and three-layer convolutional networks respectively, and the result is that the verification set accuracy is 0.5; the two-layer convolutional structure has higher verification set accuracy than the single-layer convolutional structure and better stability than the three-layer convolutional structure.

[0061] Finally, the two-layer convolutional network structure was selected as the final model structure ( Figure 2 ), the convolution kernel size of each layer ranges from 16 to 30 with a step size of 2, the filters are 128 and 256, the "size of the fully connected layer" is 256, the learning rate is 0.001, the optimizer is Adam, the loss function is the binary first-order cross entropy function, the evaluation index is ACC, and the number of iterations changes dynamically. In this embodiment, the CNN model after strict range adjustment is called DEEPReditor-CMG.

[0062] In the reference species: A. thaliana, B. napus and O. sativa, the independent prediction results of DEEPReditor-CMG in this embodiment are better than those of previous methods. Based on the PREP-Mt method, test_ACC is increased by 13.41%, 6.90% and 14.46% respectively; based on the Du-SVM method, test_ACC is increased by 9.41%, 9.41% and 14.46% respectively; based on the iPReditor-CMG method, test_ACC is increased by 2.20%, 10.71% and 17.28% respectively (Table 2). Compared with the iPReditor-CMG method, which uses 10% of the data set as an independent test set, the new method DEEPReditor-CMG in this embodiment uses 20%, and the training set and the independent test set are randomly divided five times using different random factors, which makes the results more convincing. Moreover, in the other 13 species, compared with the independent prediction results of Super sampler+BPN (SSBPN) and empirical hyperparameters (Chen J, Qian X, He Y, et al. An artificial neural network model based on DNA damage response genes to predict outcomes of lower-grade glioma patients. Brief Bioinform. 2021, 22 (6): bbab190.) two-layer convolutional CNN (Empirical hyperparameters CNN, EHCNN), the strict range parameter adjustment model DEEPReditor-CMG of this embodiment also performed better (Table 3), which also shows that the range hyperparameter optimization of this embodiment is necessary and effective. Therefore, the model of this embodiment can better fit the potential coding mechanism of C-to-U RNA editing in different species.

[0063] Table 2 Comparison of ACC between DEEPReditor-CMG and reference methods

[0064]

[0065] Note: In DEEPReditor-CMG, the training set and independent test set are randomly divided five times using different random factors.

[0066] Table 3 ACC results of the four methods for different species

[0067]

[0068] Note: In DEEPReditor-CMG, the training set and independent test set are randomly divided five times using different random factors.

[0069] Among 13 species (10 dicots and 3 monocots), the "interspecific ANI value" and "model Euclidean distance" were extremely significantly negatively correlated (edited: R = -0.31, p = 0.006; unedited: R = -0.35, p = 0.002; Figure 3 ); moreover, this negative correlation was upgraded to a highly significant strong negative correlation among the 10 dicotyledonous plants (edited: R = -0.66, p = 9.68e-7; unedited: R = -0.64, p = 2.74e-6; Figure 4 ); Since there are only three monocotyledons, although their "interspecific ANI value" is strongly negatively correlated with the "model Euclidean distance", the credibility is low; the final results show that: when the two species are more closely related, the smaller the "model Euclidean distance", the more similar the C-to-U RNA editing coding mechanisms of the two species are; on the contrary, when the two species are not closely related, the difference in the C-to-U RNA editing coding mechanisms of the two species is greater.

[0070] Since the "interspecies prediction value" value has directionality, but essentially reflects the similarity of the RNA editing mechanism between two species, the maximum value is selected for the "interspecies prediction value" value between two species in this embodiment. Among the 13 species: "interspecies ANI value" and "interspecies prediction value" are extremely significantly positively correlated (R = 0.45, p = 4e-5; Figure 5 ); This conclusion is consistent with the "model Euclidean distance", that is, the more closely related the two species are, the more similar the RNA editing coding mechanisms are.

[0071] The "inter-species ANI value" can reflect the close relationship between different species at the genome level, while the "model Euclidean distance" and "inter-species prediction value" are based on high-precision DL models, reflecting the similarity of the editing event rules and mechanisms between different species at the C-to-U RNA editing level. Between different species: "Inter-species ANI value" is negatively correlated with "model Euclidean distance", and "inter-species ANI value" is positively correlated with "inter-species prediction value" (all detailed evaluation values ​​are shown in Appendix 1). It is explained from multiple angles that the closer the two species are, the more similar their C-to-U RNA editing mechanisms are, which further supports the speculation of the SVM part. Therefore, when modeling and predicting, we cannot simply merge different species data sets, but should take into account the close relationship of species. This important conclusion provides a theoretical basis for the subsequent application part.

[0072] During the application process, the 13 species were first classified according to the class, order, family and genus (Table 1). At the genus level, if there were no species of the same genus, the trained model would continue to be used (for example, if there were three species of the same genus, at the genus level, the three species could be fused to form a model; if one of the 13 species had only one species in one genus, the model at the genus level would continue to use the model in Table 3. Finally, we added six fused models, covering all situations at the class, order, family and genus levels for the 13 species, that is, as long as the species to be predicted was a dicot or monocot, a most suitable model could be found for prediction). The species datasets of the same genus, family, order and class were fused in turn, and the model was constructed and trained using the DEEPReditor-CMG method (Table 4).

[0073] Table 4 Test_ACC results of the amplification model

[0074]

[0075] DEEPReditor-CMG of this embodiment can match the model with the closest relationship to the species to be predicted step by step according to the genus, family, order and class of the species to be predicted, so as to maximize the credibility of the model prediction results and enhance the personalized prediction function. At the same time, in order to facilitate the use of developers, DEEPReditor-CMG of this embodiment is suitable for predicting user-specified samples and predicting the whole genome of species.

[0076] Taking the species C. pepo as an example, although this species has appeared in the model training set, only a very small part of the whole genome data is used, so it is still a challenge to predict the whole genome. The prediction results of DEEPReditor-CMG in this embodiment are as follows: Figure 6 , this example predicted 18 of the 22 editing regions (as indicated by the red boxes), and also found that the 0.7×1e6-0.8×1e6 bp interval in the whole genome is very likely to be a high editing region, which provides guiding suggestions for experimental determination.

[0077] Example 2

[0078] Embodiment 2 of the present invention provides a terminal device corresponding to the above-mentioned embodiment 1. The terminal device may be a processing device for a client, such as a mobile phone, a laptop computer, a tablet computer, a desktop computer, etc., to execute the method of the above-mentioned embodiment.

[0079] The terminal device of this embodiment includes a memory, a processor, and a computer program stored in the memory; the processor executes the computer program in the memory to implement the steps of the method in the above-mentioned embodiment 1.

[0080] In some implementations, the memory may be a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory.

[0081] In some other implementations, the processor may be a central processing unit (CPU), a digital signal processor (DSP), or other general-purpose processors of various types, which are not limited herein.

[0082] Example 3

[0083] Embodiment 3 of the present invention provides a computer-readable storage medium corresponding to the above embodiment 1, on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the steps of the method of the above embodiment 1 are implemented.

[0084] Computer readable storage media can be tangible devices that hold and store instructions used by instruction execution devices. Computer readable storage media can be, for example, but not limited to, electronic storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any combination thereof.

[0085] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0086] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0087] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0088] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0089] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for predicting crop mitochondrial editing sites, characterized in that: The following steps are involved: S1. Obtain mitochondrial genome sequences and corresponding C-to-U RNA editing site annotation information of multiple species; centering on the editing site, take fragments of set lengths upstream and downstream of the mitochondrial genome sequence and the corresponding C-to-U RNA editing site annotation information sequence as positive samples; Taking any site without editing site annotation as the center, we took fragments of set lengths upstream and downstream of the mitochondrial genome sequence and the corresponding C-to-U RNA editing site annotation information as negative samples; S2. Using some positive samples and negative samples corresponding to each species to construct a training set for each species, using the training set for each species as the input of the convolutional neural network, using the probability that the central site of each sample is an editing site as the output of the convolutional neural network, training the convolutional neural network, and obtaining the optimal prediction model corresponding to each species; The convolutional neural network includes an input layer, an embedding layer, a first convolutional layer, a second convolutional layer, a downsampling layer and an output layer connected in sequence.

2. The crop mitochondrial editing site prediction method according to claim 1, characterized in that: During the convolutional neural network training process, the learning rate is set to 0.

001.

3. The crop mitochondrial editing site prediction method according to claim 1, characterized in that: The sizes of the first convolutional layer and the second convolutional layer are set to 16-30.

4. The crop mitochondrial editing site prediction method according to claim 1, characterized in that: The number of the positive samples is equal to the number of the negative samples.

5. The crop mitochondrial editing site prediction method according to any one of claims 1 to 4, characterized in that: In step S1: 1) determining whether the genus of the multiple species is the same, fusing the positive samples and negative samples of all species of the same genus to obtain a first fused sample; 2) Determine whether the families of the remaining species of different genera are the same, and fuse the positive samples and negative samples of all species of the same family to obtain a second fused sample; 3) Determine whether the orders of the remaining species of different families are the same, and fuse the positive samples and negative samples of all species of the same order to obtain a third fused sample; 4) Determine whether the classes of the remaining species of different orders are the same, and fuse the positive samples and negative samples of all species of the same class to obtain the fourth fused sample; Then the training set acquisition process in step S2 corresponds to: A training set is constructed using the first fusion sample / the second fusion sample / the third fusion sample / the fourth fusion sample.

6. The method for predicting crop mitochondrial editing sites according to claim 5, characterized in that: Also includes: S3. For the species to be predicted, the C-to-U RNA editing site is predicted using the optimal prediction model corresponding to the species to be predicted.

7. The method for predicting crop mitochondrial editing sites according to claim 6, characterized in that: In step S3, for the species to be predicted, the optimal prediction model that is closest to the species to be predicted is obtained according to the genus, family, order and class of the species to be predicted, and the C-to-U RNA editing site of the predicted species is predicted using the optimal prediction model.

8. A crop mitochondrial editing site prediction system, characterized in that: include: one or more processors; A memory having one or more programs stored thereon, which, when the one or more programs are executed by the one or more processors, enables the one or more processors to implement the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that: It stores a computer program, which, when executed by a processor, implements the steps of the method according to any one of claims 1 to 7.

Citation Information

Patent Citations

  • High-precision prediction method for RNA editing sites of tobacco mitochondria

    CN108595914A

  • Method and system for combined DNA-RNA sequencing analysis to enhance variant-calling performance and characterize variant expression status

    WO2021094175A1

  • Machine-learning based design of engineered guide systems for adenosine deaminase acting on RNA editing

    WO2023086902A1