A processing method and device for an average ribosome load prediction system

Abstract

The embodiments of the present invention relate to a processing method and device for an average ribosome load prediction system, the method comprising: constructing an average ribosome load prediction system; constructing a data set; and training the prediction system based on the data set; after the system training is completed, receiving a system application mode and corresponding system application data input by a user; when the system application mode is a single sequence prediction mode, predicting the average ribosome load of a target RNA sequence based on the prediction system; when the system application mode is a multiple sequence sorting mode, sorting multiple target RNA sequences based on the prediction system; and when the system application mode is a single sequence optimization mode, optimizing the target mRNA sequence based on the prediction system. The present invention can improve the prediction efficiency and quality of the average ribosome load prediction task.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a processing method and device for an average ribosome load prediction system. Background Art

[0002] Mean Ribosome Load (MRL) is a key biological parameter that reflects the average number of ribosomes on an mRNA molecule at a given time. The mean ribosome load is related to parameters such as the translation efficiency and translation rate of mRNA, as well as the frequency of ribosome translation stalling (also known as ribosome retention frequency). Analyzing and optimizing the mean ribosome load metric can help improve mRNA translation efficiency and protein expression levels. In other words, the prediction of mean ribosome load is of great significance for a deeper understanding of protein synthesis and regulation, as well as for the development of disease treatment strategies. However, traditional experimental methods for predicting mean ribosome load are limited by experimental conditions, making it difficult to guarantee both efficiency and quality. Summary of the Invention

[0003] The purpose of the present invention is to address the defects of the prior art and provide a processing method, device, electronic device and computer-readable storage medium for an average ribosome load prediction system. An average ribosome load prediction system is constructed with an RNA sequence pre-training model and a ribosome load prediction model as the core, and the system is trained based on three types of data sets (all types of RNA sequence data sets, non-coding RNA sequence data sets, and RNA modification sequence data sets) to improve the system's versatility, applicability and prediction accuracy; and three application modes are designed based on the system (single sequence prediction mode, multi-sequence sorting mode and single sequence optimization mode); in the single sequence prediction mode, the average ribosome load prediction of any target RNA sequence is processed based on the system. Task: In the multi-sequence sorting mode, first predict the average ribosome load of multiple target RNA sequences based on the system, then calculate the corresponding evaluation parameters according to each prediction result in combination with the preset evaluation parameter rules (RNA translation efficiency parameter rules, RNA translation rate parameter rules and RNA ribosome retention frequency parameter rules), and then sort all target RNA sequences according to the sorting rules corresponding to the evaluation parameter rules; In the single sequence optimization mode, predict the ribosome load and average ribosome load of any target mRNA sequence based on the system, and determine whether to optimize based on the average ribosome load, and optimize the target mRNA sequence in the direction of maximizing the ribosome load prediction when it is confirmed that optimization is needed. The average ribosome load prediction system provided by the present invention can improve the prediction efficiency and prediction quality of daily average ribosome load prediction tasks, and the three application modes provided by the present invention can provide a convenient system application tool for RNA sequence research, and can help scientific researchers better understand the relationship between RNA sequences and the biological functions they encode.

[0004] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present invention provides a processing method of an average ribosome load prediction system, the method comprising:

[0005] Pre-constructing an average ribosome load prediction system as a corresponding first prediction system; constructing a data set as a corresponding first data set; and training the first prediction system based on the first data set;

[0006] After the system training is completed, a system application mode and corresponding system application data input by the user are received; the system application mode includes a single sequence prediction mode, a multiple sequence sorting mode and a single sequence optimization mode; when the system application mode is the single sequence prediction mode, the system application data is the corresponding first target RNA sequence; when the system application mode is the multiple sequence sorting mode, the system application data is the corresponding multiple second target RNA sequences; when the system application mode is the single sequence optimization mode, the system application data is the corresponding first target mRNA sequence;

[0007] When the system application mode is a single sequence prediction mode, performing average ribosome load prediction processing on the first target RNA sequence based on the first prediction system;

[0008] When the system application mode is a multiple sequence sorting mode, performing target RNA sequence sorting processing on the multiple second target RNA sequences based on the first prediction system;

[0009] When the system application mode is a single sequence optimization mode, sequence optimization processing is performed on the first target mRNA sequence based on the first prediction system.

[0010] Preferably, the first prediction system includes an RNA sequence pre-training unit, an RNA sequence feature dimensionality reduction unit, a ribosome load prediction unit, and an average ribosome load prediction unit;

[0011] The input end of the RNA sequence pre-training unit is the input end of the first prediction system, and the output end is connected to the input end of the RNA sequence feature dimensionality reduction unit; the output end of the RNA sequence feature dimensionality reduction unit is connected to the input end of the ribosome load prediction unit; the output end of the ribosome load prediction unit is connected to the input end of the average ribosome load prediction unit; the output end of the average ribosome load prediction unit is the output end of the first prediction system;

[0012] The RNA sequence pre-training unit is used to perform feature encoding on the input sequence of the first prediction system and output a corresponding feature tensor; the RNA sequence pre-training unit is implemented based on a preset pre-training model; the preset pre-training model includes a pre-training model based on a Transformer structure, a pre-training model based on a CNN structure, a pre-training model based on an RNN structure, a pre-training model based on an autoencoder structure, a pre-training model based on a variational autoencoder structure, and a pre-training model based on an autoregressive model structure;

[0013] The RNA sequence feature dimensionality reduction unit is used to perform feature dimension reduction processing on the feature tensor output by the RNA sequence pre-training unit and output a corresponding reduced dimension tensor; the RNA sequence feature dimensionality reduction unit is implemented based on a type of linear neural network; the type of linear neural network includes a fully connected neural network;

[0014] The ribosome load prediction unit is used to perform ribosome load prediction processing based on the dimensionality reduction tensor output by the RNA sequence feature dimensionality reduction unit to generate a corresponding prediction tensor; the ribosome load prediction unit is implemented based on a preset prediction model; the preset prediction model includes a prediction model based on a multi-layer ResNet structure, a prediction model based on a Transformer structure, a prediction model based on a CNN structure, and a prediction model based on a GNN structure;

[0015] The average ribosome load prediction unit is used to perform average ribosome load estimation processing based on the prediction tensor output by the ribosome load prediction unit and output the obtained estimation data as the average ribosome load prediction result of the first prediction system; the average ribosome load prediction unit is implemented based on a second-class linear neural network; the second-class linear neural network includes a fully connected neural network.

[0016] Preferably, the first data set includes a first-stage data set, a second-stage data set, and a third-stage data set;

[0017] The first-stage data set is an RNA sequence database that integrates the RNAcentral database, MG-RAST database, GenomeWarehouse database, MGnify database, NCBI nucleic acid sequence database and its subsets; the first-stage data set includes multiple first RNA sequences;

[0018] The second-stage dataset includes a plurality of first non-coding RNA sequences;

[0019] The three-stage dataset includes a plurality of first RNA modification sequences.

[0020] Preferably, the training of the first prediction system based on the first data set specifically includes:

[0021] The first-stage data set is used as the corresponding first data set; each of the first RNA sequences in the first data set is used as the corresponding first sequence; the first data set is divided into a training data set and a test data set according to a preset training-test data ratio, where the training-test data ratio = the number of first sequences in the training data set: the number of first sequences in the test data set; and a round of system training is performed on the first prediction system based on the training data set and the test data set obtained this time;

[0022] At the end of this round of system training, the second-stage data set is used as the new first data set; each of the first non-coding RNA sequences in the new first data set is used as the corresponding first sequence; the new first data set is divided into the new training data set and the test data set according to the training-test data ratio; and the first prediction system is trained for one round based on the training data set and the test data set obtained this time;

[0023] At the end of this round of system training, the three-stage dataset is used as the new first dataset; each of the first RNA modification sequences in the new first dataset is used as the corresponding first sequence; the new first dataset is divided into the new training dataset and the test dataset according to the training-test data ratio; and the first prediction system is trained for one round based on the training dataset and the test dataset obtained this time;

[0024] And at the end of this round of system training, it is confirmed that the system training of the first prediction system is completed.

[0025] Furthermore, performing a round of system training on the first prediction system based on the training data set and the test data set obtained this time specifically includes:

[0026] Step 51: Use the training data set as the corresponding current data set; and set the first state to be the training state; the first state includes the training state and the test state;

[0027] Step 52: Select the first of the first sequences from the current data set as the corresponding current sequence;

[0028] Step 53: using the ribosome load of the current sequence as the corresponding current sequence tag; and performing standardization processing on the current sequence tag to generate corresponding first tag data;

[0029] Step 54: input the current sequence into the first prediction system; the RNA sequence pre-training unit performs feature encoding on the current sequence and outputs a corresponding first feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the first feature tensor and outputs a corresponding first reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the first reduced dimensionality tensor to generate a corresponding first prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the first prediction tensor to generate corresponding first estimation data; and the first estimation data is output as the average ribosome load prediction result of the first prediction system;

[0030] Step 55: Substitute the first estimated data and the first label data into a preset error function to calculate a corresponding first prediction error;

[0031] Step 56, identifying the first state; when the first state is a training state, go to step 57; when the first state is a test state, go to step 58;

[0032] Step 57: Identify whether the first prediction error satisfies a preset first reasonable error range; if so, proceed to step 59; if not, perform full model parameter modulation on the RNA sequence pre-training unit and the ribosome load prediction unit of the first prediction system based on a preset system optimization algorithm, and return to step 54 to continue training after parameter modulation is completed; the system optimization algorithm includes a gradient descent method, a genetic optimization algorithm, a simulated annealing algorithm, a particle swarm optimization algorithm, and a differential evolution algorithm; the full model parameter modulation is to modulate all model parameters of the RNA sequence pre-training unit and the ribosome load prediction unit;

[0033] Step 58: Identify whether the first prediction error meets a preset second reasonable error range; if so, proceed to step 59; if not, modulate the specified model parameters of the ribosome load prediction unit of the first prediction system based on the system optimization algorithm and return to step 54 to continue training after the parameter modulation is completed; the specified model parameter modulation is to modulate a portion of the model parameters pre-specified in the ribosome load prediction unit;

[0034] Step 59: Identify whether the current sequence is the last first sequence in the current data set; if so, proceed to step 60; if not, select the next first sequence from the current data set as the new current sequence and return to step 53 to continue training;

[0035] Step 60, identifying the first state; when the first state is the training state, using the test data set as the new current data set, setting the first state to the test state, and returning to step 52 to continue training; when the first state is the test state, stopping training and confirming the end of this round of system training.

[0036] Preferably, performing average ribosome load prediction processing on the first target RNA sequence based on the first prediction system specifically includes:

[0037] The first target RNA sequence is input into the first prediction system; the RNA sequence pre-training unit performs feature encoding on the first target RNA sequence and outputs a corresponding second feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the second feature tensor and outputs a corresponding second reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the second reduced dimensionality tensor to generate a corresponding second prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the second prediction tensor to generate corresponding second estimation data; and the second estimation data is output as the processing result of the average ribosome load prediction processing.

[0038] Preferably, performing target RNA sequence sorting processing on the plurality of second target RNA sequences based on the first prediction system specifically includes:

[0039] Input each second target RNA sequence into the first prediction system respectively; the RNA sequence pre-training unit performs feature encoding on the currently input second target RNA sequence and outputs a corresponding third feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the third feature tensor and outputs a corresponding third reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the third reduced dimensionality tensor to generate a corresponding third prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the third prediction tensor to generate corresponding third estimation data; and the third estimation data is output as the average ribosome load prediction result of the first prediction system;

[0040] According to the preset evaluation parameter rules, the evaluation parameters are calculated according to each of the third estimation data to obtain the corresponding first evaluation parameters; and the first evaluation parameters are sorted according to the sorting rules corresponding to the evaluation parameter rules; and the second target RNA sequences corresponding to each of the first evaluation parameters are sorted based on the sorting order of all the first evaluation parameters to obtain the corresponding second target RNA sequence sorting array; and the obtained second target RNA sequence sorting array is output as the processing result of the target RNA sequence sorting process; the evaluation parameter rules include RNA translation efficiency parameter rules, RNA translation rate parameter rules and RNA ribosome retention frequency parameter rules; the RNA translation efficiency parameter rules are corresponding rules for calculating translation efficiency parameters based on average ribosome load, and the RNA translation The sorting rule corresponding to the translation efficiency parameter rule is to sort all the second target RNA sequences in order of translation efficiency from high to low; the RNA translation rate parameter rule is the corresponding rule for calculating the translation rate parameter based on the average ribosome load, and the sorting rule corresponding to the RNA translation rate parameter rule is to first sort all the second target RNA sequences by level based on a pre-specified multi-level translation rate range, and then sort multiple second target RNA sequences of the same level in order of translation rate from high to low; the RNA ribosome retention frequency parameter rule is the corresponding rule for calculating the ribosome retention frequency parameter based on the average ribosome load, and the sorting rule corresponding to the RNA ribosome retention frequency parameter rule is to sort all the second target RNA sequences in order of ribosome retention frequency from low to high.

[0041] Preferably, performing sequence optimization processing on the first target mRNA sequence based on the first prediction system specifically includes:

[0042] Step 81, using the first target mRNA sequence as the corresponding current mRNA sequence;

[0043] Step 82: Input the current mRNA sequence into the first prediction system; the RNA sequence pre-training unit performs feature encoding on the current mRNA sequence and outputs a corresponding fourth feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the fourth feature tensor and outputs a corresponding fourth reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the fourth reduced dimensionality tensor to generate a corresponding fourth prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the fourth prediction tensor to generate corresponding fourth estimation data; and the fourth estimation data is output as the average ribosome load prediction result of the first prediction system;

[0044] Step 83, identifying whether the fourth estimated data exceeds a preset optimal average ribosome load threshold; if not, proceeding to step 84; if so, proceeding to step 85;

[0045] Step 84, performing sequence optimization processing on the current mRNA sequence based on a preset mRNA sequence optimization algorithm in a direction of maximizing the ribosome load prediction of the ribosome load prediction unit to obtain a corresponding mRNA optimized sequence; and using the mRNA optimized sequence as the new current mRNA sequence and returning to step 82; the mRNA sequence optimization algorithm includes a sequence optimization algorithm implemented based on a simulated annealing algorithm and a sequence optimization algorithm implemented based on a genetic optimization algorithm;

[0046] Step 85: output the latest current mRNA sequence as the processing result of the sequence optimization process.

[0047] A second aspect of an embodiment of the present invention provides an apparatus for implementing the processing method of the average ribosome load prediction system described in the first aspect, the apparatus comprising: a prediction system construction and training module, an application information acquisition module, a first-class application processing module, a second-class application processing module, and a third-class application processing module;

[0048] The prediction system construction and training module is used to construct an average ribosome load prediction system, which is recorded as a corresponding first prediction system; and to construct a data set, which is recorded as a corresponding first data set; and to train the first prediction system based on the first data set;

[0049] The application information acquisition module is used to receive a system application mode and corresponding system application data input by a user after the system training is completed; the system application mode includes a single sequence prediction mode, a multiple sequence sorting mode, and a single sequence optimization mode; when the system application mode is the single sequence prediction mode, the system application data is the corresponding first target RNA sequence; when the system application mode is the multiple sequence sorting mode, the system application data is the corresponding multiple second target RNA sequences; when the system application mode is the single sequence optimization mode, the system application data is the corresponding first target mRNA sequence;

[0050] The first type of application processing module is used for performing average ribosome load prediction processing on the first target RNA sequence based on the first prediction system when the system application mode is a single sequence prediction mode;

[0051] The second type of application processing module is used for performing target RNA sequence sorting processing on the multiple second target RNA sequences based on the first prediction system when the system application mode is a multiple sequence sorting mode;

[0052] The three types of application processing modules are used to perform sequence optimization processing on the first target mRNA sequence based on the first prediction system when the system application mode is a single sequence optimization mode.

[0053] A third aspect of an embodiment of the present invention provides an electronic device, including: a memory, a processor, and a transceiver;

[0054] The processor is configured to be coupled to the memory, read and execute instructions in the memory, so as to implement the method steps described in the first aspect above;

[0055] The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

[0056] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a computer, the computer executes the instructions of the method described in the first aspect above.

[0057] The embodiment of the present invention provides a processing method, device, electronic device and computer-readable storage medium for an average ribosome load prediction system, which constructs an average ribosome load prediction system with an RNA sequence pre-training model and a ribosome load prediction model as the core, and trains the system based on three types of data sets (all types of RNA sequence data sets, non-coding RNA sequence data sets, and RNA modification sequence data sets) to improve the system's versatility, applicability and prediction accuracy; and designs three application modes based on the system (single sequence prediction mode, multi-sequence sorting mode and single sequence optimization mode); in the single sequence prediction mode, the system processes the average ribosome load prediction task of any target RNA sequence; in the multi-sequence sorting ... In the column sorting mode, the average ribosome load of multiple target RNA sequences is first predicted based on the system, and then the corresponding evaluation parameters are calculated according to each prediction result in combination with the preset evaluation parameter rules (RNA translation efficiency parameter rules, RNA translation rate parameter rules and RNA ribosome retention frequency parameter rules), and then all target RNA sequences are sorted according to the sorting rules corresponding to the evaluation parameter rules; in the single sequence optimization mode, the ribosome load and average ribosome load of any target mRNA sequence are predicted based on the system, and it is determined whether to optimize based on the average ribosome load, and when it is confirmed that optimization is needed, the target mRNA sequence is optimized in the direction of maximizing the ribosome load prediction. The average ribosome load prediction system provided by the present invention improves the prediction efficiency and prediction quality of daily average ribosome load prediction tasks, and provides three convenient RNA sequence research tools through the three application modes of the present invention, which can help scientific researchers better understand the relationship between RNA sequences and the biological functions they encode. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of a processing method of an average ribosome load prediction system provided in Example 1 of the present invention;

[0059] Figure 2 This is a module structure diagram of the first prediction system provided in Example 1 of the present invention;

[0060] Figure 3 A module structure diagram of a processing device of an average ribosome load prediction system provided in Example 2 of the present invention;

[0061] Figure 4 This is a structural diagram of an electronic device provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0062] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It should be understood that the embodiments described herein are merely some, rather than all, of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are intended to fall within the scope of protection of the present invention.

[0063] The first embodiment of the present invention provides a processing method of an average ribosome load prediction system, such as Figure 1 The schematic diagram of a processing method of an average ribosome load prediction system provided in Example 1 of the present invention is shown. This method mainly includes the following steps:

[0064] Step 1: pre-construct an average ribosome load prediction system, which is recorded as a corresponding first prediction system; construct a data set, which is recorded as a corresponding first data set; and train the first prediction system based on the first data set;

[0065] Specifically comprising: step 11, pre-building an average ribosome load prediction system as a corresponding first prediction system;

[0066] Among them, such as Figure 2 As shown in the module structure diagram of the first prediction system provided in Example 1 of the present invention, the first prediction system includes an RNA sequence pre-training unit, an RNA sequence feature dimensionality reduction unit, a ribosome load prediction unit, and an average ribosome load prediction unit;

[0067] The input end of the RNA sequence pre-training unit is the input end of the first prediction system, and the output end is connected to the input end of the RNA sequence feature dimensionality reduction unit; the output end of the RNA sequence feature dimensionality reduction unit is connected to the input end of the ribosome load prediction unit; the output end of the ribosome load prediction unit is connected to the input end of the average ribosome load prediction unit; the output end of the average ribosome load prediction unit is the output end of the first prediction system;

[0068] The RNA sequence pre-training unit is used to perform feature encoding on the input sequence of the first prediction system and output the corresponding feature tensor; the RNA sequence pre-training unit is implemented based on a preset pre-training model; the preset pre-training model includes a pre-training model based on a Transformer structure, a pre-training model based on a CNN structure, a pre-training model based on an RNN structure, a pre-training model based on an autoencoder structure, a pre-training model based on a variational autoencoder structure, and a pre-training model based on an autoregressive model structure; here, the preset pre-training model is essentially a model that can extract structural features and encode semantic features based on RNA sequences. In an embodiment of the present invention, a pre-training model based on a Transformer structure is preferably used as the preset pre-training model, and pre-training models based on the Transformer structure include Uni-Fold model, ESM series model, BERT model, GPT model, etc.

[0069] The RNA sequence feature dimensionality reduction unit is used to reduce the feature dimension of the feature tensor output by the RNA sequence pre-training unit and output the corresponding reduced dimension tensor. The RNA sequence feature dimensionality reduction unit is implemented based on a type of linear neural network. A type of linear neural network includes a fully connected neural network. Here, the RNA sequence feature dimensionality reduction unit essentially reduces the channel dimension of the feature tensor output by the pre-training model through a linear neural network.

[0070] The ribosome load prediction unit is used to generate a corresponding prediction tensor based on the dimensionality reduction tensor output by the RNA sequence feature dimensionality reduction unit. The ribosome load prediction unit is implemented based on a preset prediction model. The preset prediction models include prediction models based on a multi-layer ResNet structure, a Transformer structure, a CNN structure, and a GNN structure. The working principle of this prediction model is similar to that of the ribosome load prediction model in the paper "Human 5′UTR design and variant effect prediction from a massively parallel translation assay", both of which predict ribosome load based on the input sequence feature tensor.

[0071] The average ribosome load prediction unit is configured to estimate the average ribosome load based on the prediction tensor output by the ribosome load prediction unit and output the obtained estimated data as the average ribosome load prediction result of the first prediction system. The average ribosome load prediction unit is implemented based on a two-class linear neural network, which includes a fully connected neural network. The estimation principle of the two-class linear neural network of the average ribosome load prediction unit is similar to the method for calculating the average ribosome load based on the predicted ribosome load in the paper "Human 5′UTR design and variant effect prediction from a massively parallel translation assay."

[0072] Step 12, and construct a data set and record it as the corresponding first data set;

[0073] The first data set includes a first-stage data set, a second-stage data set, and a third-stage data set;

[0074] The first-stage dataset is an RNA sequence database that integrates the RNAcentral database, MG-RAST database, Genome Warehouse database, MGnify database, NCBI nucleotide sequence database and its subsets; the first-stage dataset includes multiple first RNA sequences; here, the first-stage dataset is essentially a full-type RNA sequence dataset. Training the system based on the first-stage dataset can ensure the system's versatility;

[0075] The second-stage dataset includes multiple first non-coding RNA sequences. This dataset is essentially a non-coding RNA sequence dataset, similar to the dataset used to train the RNA-FM model in the paper "Interpretable RNA Foundation Model from Unannotated Data for Highly Accurate RNA Structure and Function Predictions." Therefore, the second-stage dataset is also called the RNA-FM dataset. Training the system based on the second-stage dataset can ensure the applicability of the system.

[0076] The three-stage dataset includes multiple first RNA modification sequences. Here, the three-stage dataset is essentially an RNA modification sequence dataset. Each first RNA modification sequence in the dataset is an RNA sequence that has been pre-processed based on a specified RNA modification rule. Training the system based on the three-stage dataset can further improve the system's prediction accuracy.

[0077] Step 13: training the first prediction system based on the first data set;

[0078] Specifically, the method includes: step 131, using the first-stage data set as the corresponding first data set; using each first RNA sequence in the first data set as the corresponding first sequence; dividing the first data set into a training data set and a test data set according to a preset training-test data ratio, where the training-test data ratio = the number of first sequences in the training data set: the number of first sequences in the test data set; and performing a round of system training on the first prediction system based on the training data set and the test data set obtained this time;

[0079] The default training-testing data ratio in this invention is 9:1.

[0080] A round of system training is performed on the first prediction system based on the training data set and test data set obtained this time. Specifically:

[0081] Step A1: Use the training data set as the corresponding current data set; and set the first state to the training state;

[0082] Wherein, the first state includes a training state and a testing state;

[0083] Step A2, selecting the first first sequence from the current data set as the corresponding current sequence;

[0084] Step A3: using the ribosome load of the current sequence as the corresponding current sequence tag; and performing standardization processing on the current sequence tag to generate corresponding first tag data;

[0085] Step A4: Input the current sequence into the first prediction system; the RNA sequence pre-training unit performs feature encoding on the current sequence and outputs a corresponding first feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the first feature tensor and outputs a corresponding first reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the first reduced dimensionality tensor to generate a corresponding first prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the first prediction tensor to generate corresponding first estimation data; and the first estimation data is output as the average ribosome load prediction result of the first prediction system;

[0086] Step A5: Substituting the first estimated data and the first label data into a preset error function to calculate a corresponding first prediction error;

[0087] Here, the preset error function of the embodiment of the present invention defaults to the mean square error MSE function (Mean Squared Error, MSE);

[0088] Step A6, identifying the first state; when the first state is the training state, go to step A7; when the first state is the test state, go to step A8;

[0089] Step A7: Identify whether the first prediction error meets a preset first reasonable error range; if so, proceed to step A9; if not, perform full model parameter modulation on the RNA sequence pre-training unit and the ribosome load prediction unit of the first prediction system based on a preset system optimization algorithm, and return to step A4 to continue training after parameter modulation is completed;

[0090] The system optimization algorithms preset in the embodiment of the present invention include gradient descent method, genetic optimization algorithm, simulated annealing algorithm, particle swarm optimization algorithm and differential evolution algorithm; full model parameter modulation is to modulate all model parameters in the RNA sequence pre-training unit and the ribosome load prediction unit;

[0091] Here, when modulating all model parameters in the RNA sequence pre-training unit and the ribosome load prediction unit, the embodiment of the present invention constructs an objective function based on the determination coefficient R2 function (R-Square, R2), and modulates all model parameters in the RNA sequence pre-training unit and the ribosome load prediction unit in a direction such that the determination coefficient R2 function reaches a maximum value;

[0092] Step A8: Identify whether the first prediction error meets a preset second reasonable error range; if so, proceed to step A9; if not, modulate the specified model parameters of the ribosome load prediction unit of the first prediction system based on the system optimization algorithm and return to step A4 to continue training after the parameter modulation is completed; the specified model parameter modulation is to modulate a portion of the model parameters pre-specified in the ribosome load prediction unit;

[0093] Here, when the embodiment of the present invention modulates the specified model parameters of the ribosome load prediction unit, an objective function is constructed based on the determination coefficient R2 function, and some pre-specified model parameters in the ribosome load prediction unit are modulated in a direction such that the determination coefficient R2 function reaches a maximum value. In general, these pre-specified model parameters are mainly model hyperparameters.

[0094] Step A9: Identify whether the current sequence is the last first sequence in the current data set; if so, go to step A10; if not, select the next first sequence from the current data set as the new current sequence and return to step A3 to continue training;

[0095] Step A10, identifying the first state; when the first state is the training state, using the test data set as the new current data set, setting the first state to the test state, and returning to step A2 to continue training; when the first state is the test state, stopping training and confirming the end of this round of system training;

[0096] At step 132, at the end of this round of system training, the second-stage dataset is used as a new first dataset; each first non-coding RNA sequence in the new first dataset is used as a corresponding first sequence; the new first dataset is divided into a new training dataset and a test dataset according to the training-test data ratio; and a round of system training is performed on the first prediction system based on the training dataset and the test dataset obtained this time;

[0097] Here, the current step 132 is to continue training the first prediction system trained in the previous step 131 based on the second-stage dataset after the training in the previous step 131 is completed. The training implementation method is similar to the previous step 131. In the current step 132, the technical implementation steps of performing a round of system training on the first prediction system based on the training dataset and test dataset obtained this time are consistent with the technical implementation steps of the previous steps A1-A10.

[0098] Step 133: At the end of this round of system training, the three-stage dataset is used as a new first dataset; each first RNA modification sequence in the new first dataset is used as a corresponding first sequence; the new first dataset is divided into a new training dataset and a test dataset according to the training-test data ratio; and the first prediction system is trained for one round based on the training dataset and the test dataset obtained this time.

[0099] Here, the current step 133 is to continue training the first prediction system trained in the previous step 132 based on the three-stage dataset after the training in the previous step 132 is completed. The training implementation method is similar to the previous step 131. In the current step 133, the technical implementation steps of performing a round of system training on the first prediction system based on the training dataset and test dataset obtained this time are consistent with the technical implementation steps of the previous steps A1-A10.

[0100] Step 134 , when this round of system training ends, confirm that the system training of the first prediction system is completed.

[0101] Here, the embodiment of the present invention completes the training of the first prediction system through the above steps 131-134, thereby achieving better system versatility, system applicability and system prediction accuracy.

[0102] It should also be noted that, to further improve the prediction robustness of the system, the embodiment of the present invention may further perform a prediction robustness test on the first prediction system at the end of each round of system training based on the training dataset and test dataset obtained this time, obtain a corresponding test result, and identify whether the test result meets the robustness standard. If so, the next round of system training is performed based on the next stage dataset; if not, the system training is repeated based on the current stage dataset. In this embodiment, when performing the prediction robustness test on the first prediction system, the embodiment of the present invention performs a prediction robustness test using both the eGFP and mCherry CDS methods, and sets the corresponding test result as meeting the robustness standard when the test results of both methods indicate that the robustness standard is met. In this embodiment, the implementation mechanism of the prediction robustness test using the eGFP and mCherry CDS methods in the embodiment of the present invention is similar to the processing mechanism for detecting the correlation between the average ribosome load prediction and the actual protein expression using the eGFP and mCherry CDS methods mentioned in the paper "Human 5′UTR design and variant effect prediction from a massively parallel translation assay";

[0103] Step 2: After the system training is completed, the system application mode and corresponding system application data input by the user are received;

[0104] Among them, the system application modes include single sequence prediction mode, multiple sequence sorting mode and single sequence optimization mode; when the system application mode is single sequence prediction mode, the system application data is the corresponding first target RNA sequence; when the system application mode is multiple sequence sorting mode, the system application data is the corresponding multiple second target RNA sequences; when the system application mode is single sequence optimization mode, the system application data is the corresponding first target mRNA sequence.

[0105] Here, an embodiment of the present invention provides three system application tools for RNA sequence research based on the system application mode; when the system application mode is a single sequence prediction mode, the system application tool provided is an average ribosome load prediction tool for a single RNA sequence, which is specifically implemented by step 3 below; when the system application mode is a multiple sequence sorting mode, the system application tool provided is a ranking prediction tool for multiple RNA sequences, which is specifically implemented by step 4 below; when the system application mode is a single sequence optimization mode, the system application tool provided is a sequence optimization tool for a single mRNA sequence, which is specifically implemented by step 5 below.

[0106] Step 3: When the system application mode is the single sequence prediction mode, an average ribosome load prediction process is performed on the first target RNA sequence based on the first prediction system;

[0107] Specifically, the method includes: inputting a first target RNA sequence into a first prediction system; performing feature encoding on the first target RNA sequence by an RNA sequence pre-training unit and outputting a corresponding second feature tensor; performing feature dimension reduction processing on the second feature tensor by an RNA sequence feature dimensionality reduction unit and outputting a corresponding second reduced dimensionality tensor; performing ribosome load prediction processing on the second reduced dimensionality tensor by a ribosome load prediction unit to generate a corresponding second prediction tensor; performing average ribosome load estimation processing on the second prediction tensor by an average ribosome load prediction unit to generate corresponding second estimation data; and outputting the second estimation data as a processing result of the average ribosome load prediction processing.

[0108] Step 4: When the system application mode is a multiple sequence sorting mode, target RNA sequence sorting processing is performed on the multiple second target RNA sequences based on the first prediction system;

[0109] Specifically, the method includes: step 41, inputting each second target RNA sequence into the first prediction system respectively; performing feature encoding on the currently input second target RNA sequence by the RNA sequence pre-training unit and outputting a corresponding third feature tensor; performing feature dimension reduction processing on the third feature tensor by the RNA sequence feature dimensionality reduction unit and outputting a corresponding third reduced dimensionality tensor; performing ribosome load prediction processing on the third reduced dimensionality tensor to generate a corresponding third prediction tensor by the ribosome load prediction unit; performing average ribosome load estimation processing on the third prediction tensor to generate corresponding third estimated data; and outputting the third estimated data as the average ribosome load prediction result of the first prediction system;

[0110] Step 42: performing evaluation parameter calculation based on each third estimation data according to a preset evaluation parameter rule to obtain a corresponding first evaluation parameter; sorting the first evaluation parameters according to a sorting rule corresponding to the evaluation parameter rule; and sorting the second target RNA sequences corresponding to each first evaluation parameter based on the sorting order of all first evaluation parameters to obtain a corresponding second target RNA sequence sorting array; and outputting the obtained second target RNA sequence sorting array as a processing result of the target RNA sequence sorting process;

[0111] Among them, the evaluation parameter rules include RNA translation efficiency parameter rules, RNA translation rate parameter rules and RNA ribosome retention frequency parameter rules;

[0112] The RNA translation efficiency parameter rule is a corresponding rule for calculating the translation efficiency parameter based on the average ribosome load, and the sorting rule corresponding to the RNA translation efficiency parameter rule is to sort all second target RNA sequences in descending order of translation efficiency;

[0113] The RNA translation rate parameter rule is a corresponding rule for calculating the translation rate parameter based on the average ribosome load. The sorting rule corresponding to the RNA translation rate parameter rule is to first sort all second target RNA sequences by level based on a pre-specified multi-level translation rate range, and then sort multiple second target RNA sequences of the same level in descending order of translation rate; the pre-specified multi-level translation rate range includes multiple level translation rate ranges, and each level translation rate range is a corresponding rate range; it should be noted here that, in the embodiment of the present invention, each level translation rate range does not correspond to a higher level as the faster the rate range, but rather a stable high-speed range is pre-selected as the highest level translation rate range, and then other high-speed or low-speed translation rate ranges that are higher or lower than the highest level translation rate range are marked one by one according to a preset sorting rule;

[0114] The RNA ribosome retention frequency parameter rule is the corresponding rule for calculating the ribosome retention frequency parameter based on the average ribosome load. The sorting rule corresponding to the RNA ribosome retention frequency parameter rule is to sort all second target RNA sequences in order of ribosome retention frequency from low to high. Here, the larger the ribosome retention frequency parameter, the more ribosomes are retained and the greater the translation is blocked. Therefore, the corresponding priority rule, i.e., the sorting rule, should be sorted in order of ribosome retention frequency from low to high.

[0115] Step 5: When the system application mode is the single sequence optimization mode, the first target mRNA sequence is optimized based on the first prediction system;

[0116] Specifically comprising: step 51, taking the first target mRNA sequence as the corresponding current mRNA sequence;

[0117] Step 52: Input the current mRNA sequence into the first prediction system; the RNA sequence pre-training unit performs feature encoding on the current mRNA sequence and outputs a corresponding fourth feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the fourth feature tensor and outputs a corresponding fourth reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the fourth reduced dimensionality tensor to generate a corresponding fourth prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the fourth prediction tensor to generate corresponding fourth estimated data; and the fourth estimated data is output as the average ribosome load prediction result of the first prediction system;

[0118] Step 53, identifying whether the fourth estimated data exceeds a preset optimal average ribosome load threshold; if not, proceeding to step 54; if so, proceeding to step 55;

[0119] Step 54: Optimize the current mRNA sequence based on a preset mRNA sequence optimization algorithm in a direction that maximizes the ribosome load prediction of the ribosome load prediction unit to obtain a corresponding mRNA optimized sequence; and use the mRNA optimized sequence as a new current mRNA sequence and return to step 52;

[0120] Among them, the mRNA sequence optimization algorithm includes a sequence optimization algorithm based on a simulated annealing algorithm and a sequence optimization algorithm based on a genetic optimization algorithm;

[0121] Step 55: Output the latest current mRNA sequence as the processing result of the sequence optimization process.

[0122] Figure 3 This is a module structure diagram of a processing device for an average ribosome load prediction system provided in the second embodiment of the present invention. The device is a terminal device or server that implements the aforementioned method embodiment, or can be a device that enables the aforementioned terminal device or server to implement the aforementioned method embodiment. For example, the device can be a device or chip system of the aforementioned terminal device or server. Figure 3 As shown, the device includes: a prediction system construction and training module 201, an application information acquisition module 202, a first-class application processing module 203, a second-class application processing module 204 and a third-class application processing module 205.

[0123] The prediction system construction and training module 201 is used to construct an average ribosome load prediction system, which is recorded as the corresponding first prediction system; construct a data set, which is recorded as the corresponding first data set; and train the first prediction system based on the first data set.

[0124] The application information acquisition module 202 is used to receive the system application mode and corresponding system application data input by the user after the system training is completed; the system application mode includes a single sequence prediction mode, a multiple sequence sorting mode and a single sequence optimization mode; when the system application mode is a single sequence prediction mode, the system application data is the corresponding first target RNA sequence; when the system application mode is a multiple sequence sorting mode, the system application data is the corresponding multiple second target RNA sequences; when the system application mode is a single sequence optimization mode, the system application data is the corresponding first target mRNA sequence.

[0125] The first type of application processing module 203 is used to perform average ribosome load prediction processing on the first target RNA sequence based on the first prediction system when the system application mode is the single sequence prediction mode.

[0126] The second application processing module 204 is used to perform target RNA sequence sorting processing on multiple second target RNA sequences based on the first prediction system when the system application mode is the multi-sequence sorting mode.

[0127] The third type of application processing module 205 is used to perform sequence optimization processing on the first target mRNA sequence based on the first prediction system when the system application mode is the single sequence optimization mode.

[0128] An embodiment of the present invention provides a processing device for an average ribosome load prediction system, which can execute the method steps in the above method embodiment. Its implementation principles and technical effects are similar and will not be repeated here.

[0129] It should be noted that the division of the modules of the above devices is merely a division of logical functions. In actual implementation, they can be fully or partially integrated into a single physical entity or physically separated. Furthermore, these modules can be implemented entirely in the form of software called by a processing element; or entirely in the form of hardware; or some modules can be implemented in the form of software called by a processing element, and some modules can be implemented in the form of hardware. For example, the prediction system construction and training module can be a separate processing element, or it can be integrated into a chip of the above device. In addition, it can be stored in the form of program code in the memory of the above device, and called by a processing element of the above device to perform the functions of the above-mentioned determination module. The implementation of other modules is similar. In addition, these modules can be fully or partially integrated together, or implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. During implementation, each step of the above method or each of the above modules can be completed by hardware integrated logic circuits in the processor element or by software instructions.

[0130] For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application-specific integrated circuits (ASICs), one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs). For another example, when a module is implemented by scheduling program code through a processing element, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0131] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the above method embodiments are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The above-mentioned computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the above-mentioned computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, Bluetooth, microwave, etc.) means. The above-mentioned computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media. The above-mentioned available medium can be a magnetic medium (such as a floppy disk, hard disk, tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid state disk (SSD)), etc.

[0132] Figure 4 This is a schematic diagram of the structure of an electronic device provided in the third embodiment of the present invention. The electronic device can be a terminal device or server that implements the method of the aforementioned embodiment, or it can be a terminal device or server that implements the method of the aforementioned embodiment connected to the aforementioned terminal device or server. Figure 4As shown, the electronic device may include: a processor 301 (such as a CPU), a memory 302, and a transceiver 303; the transceiver 303 is coupled to the processor 301, and the processor 301 controls the transceiver 303's transceiver actions. Various instructions may be stored in the memory 302 for completing various processing functions and implementing the processing steps described in the aforementioned embodiment method. Preferably, the electronic device involved in the embodiment of the present invention further includes: a power supply 304, a system bus 305, and a communication port 306. The system bus 305 is used to realize communication connections between components. The above-mentioned communication port 306 is used for connecting and communicating between the electronic device and other peripherals.

[0133] exist Figure 4 The system bus 305 mentioned in the figure can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface is used to realize communication between the database access device and other devices (such as clients, read-write libraries, and read-only libraries). The memory may include random access memory (RAM) and may also include non-volatile memory (Non-Volatile Memory), such as at least one disk storage.

[0134] The above-mentioned processors can be general-purpose processors, including central processing units (CPUs), network processors (NPs), graphics processing units (GPUs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0135] It should be noted that an embodiment of the present invention further provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute the methods and processing procedures provided in the above embodiments.

[0136] An embodiment of the present invention further provides a chip for executing instructions, which is used to execute the processing steps described in the above method embodiment.

[0137] The embodiment of the present invention provides a processing method, device, electronic device and computer-readable storage medium for an average ribosome load prediction system, which constructs an average ribosome load prediction system with an RNA sequence pre-training model and a ribosome load prediction model as the core, and trains the system based on three types of data sets (all types of RNA sequence data sets, non-coding RNA sequence data sets, and RNA modification sequence data sets) to improve the system's versatility, applicability and prediction accuracy; and designs three application modes based on the system (single sequence prediction mode, multi-sequence sorting mode and single sequence optimization mode); in the single sequence prediction mode, the system processes the average ribosome load prediction task of any target RNA sequence; in the multi-sequence sorting ... In the column sorting mode, the average ribosome load of multiple target RNA sequences is first predicted based on the system, and then the corresponding evaluation parameters are calculated according to each prediction result in combination with the preset evaluation parameter rules (RNA translation efficiency parameter rules, RNA translation rate parameter rules and RNA ribosome retention frequency parameter rules), and then all target RNA sequences are sorted according to the sorting rules corresponding to the evaluation parameter rules; in the single sequence optimization mode, the ribosome load and average ribosome load of any target mRNA sequence are predicted based on the system, and it is determined whether to optimize based on the average ribosome load, and when it is confirmed that optimization is needed, the target mRNA sequence is optimized in the direction of maximizing the ribosome load prediction. The average ribosome load prediction system provided by the present invention improves the prediction efficiency and prediction quality of daily average ribosome load prediction tasks, and provides three convenient RNA sequence research tools through the three application modes of the present invention, which can help scientific researchers better understand the relationship between RNA sequences and the biological functions they encode.

[0138] Professionals should also be further aware that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0139] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0140] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A processing method for an average ribosome load prediction system, characterized in that: The method comprises: Pre-constructing an average ribosome load prediction system as a corresponding first prediction system; constructing a data set as a corresponding first data set; and training the first prediction system based on the first data set; After the system training is completed, a system application mode and corresponding system application data input by the user are received; the system application mode includes a single sequence prediction mode, a multiple sequence sorting mode and a single sequence optimization mode; when the system application mode is the single sequence prediction mode, the system application data is the corresponding first target RNA sequence; when the system application mode is the multiple sequence sorting mode, the system application data is the corresponding multiple second target RNA sequences; when the system application mode is the single sequence optimization mode, the system application data is the corresponding first target mRNA sequence; When the system application mode is a single sequence prediction mode, performing average ribosome load prediction processing on the first target RNA sequence based on the first prediction system; When the system application mode is a multiple sequence sorting mode, performing target RNA sequence sorting processing on the multiple second target RNA sequences based on the first prediction system; When the system application mode is a single sequence optimization mode, performing sequence optimization processing on the first target mRNA sequence based on the first prediction system; Wherein, the first prediction system includes an RNA sequence pre-training unit, an RNA sequence feature dimensionality reduction unit, a ribosome load prediction unit and an average ribosome load prediction unit; The first data set includes a first-stage data set, a second-stage data set, and a third-stage data set; the first-stage data set is an RNA sequence database that integrates the RNAcentral database, the MG-RAST database, the Genome Warehouse database, the MGnify database, the NCBI nucleic acid sequence database, and its subsets; the first-stage data set includes multiple first RNA sequences; the second-stage data set includes multiple first non-coding RNA sequences; the third-stage data set includes multiple first RNA modification sequences; The training of the first prediction system based on the first data set specifically includes: The first-stage data set is used as the corresponding first data set; each of the first RNA sequences in the first data set is used as the corresponding first sequence; the first data set is divided into a training data set and a test data set according to a preset training-test data ratio, where the training-test data ratio = the number of first sequences in the training data set: the number of first sequences in the test data set; and a round of system training is performed on the first prediction system based on the training data set and the test data set obtained this time; At the end of this round of system training, the second-stage data set is used as the new first data set; each of the first non-coding RNA sequences in the new first data set is used as the corresponding first sequence; the new first data set is divided into the new training data set and the test data set according to the training-test data ratio; and the first prediction system is trained for one round based on the training data set and the test data set obtained this time; At the end of this round of system training, the three-stage dataset is used as the new first dataset; each of the first RNA modification sequences in the new first dataset is used as the corresponding first sequence; the new first dataset is divided into the new training dataset and the test dataset according to the training-test data ratio; and the first prediction system is trained for one round based on the training dataset and the test dataset obtained this time; And at the end of this round of system training, it is confirmed that the system training of the first prediction system is completed.

2. The processing method of the average ribosome load prediction system according to claim 1, characterized in that: The input end of the RNA sequence pre-training unit is the input end of the first prediction system, and the output end is connected to the input end of the RNA sequence feature dimensionality reduction unit; the output end of the RNA sequence feature dimensionality reduction unit is connected to the input end of the ribosome load prediction unit; the output end of the ribosome load prediction unit is connected to the input end of the average ribosome load prediction unit; the output end of the average ribosome load prediction unit is the output end of the first prediction system; The RNA sequence pre-training unit is used to perform feature encoding on the input sequence of the first prediction system and output a corresponding feature tensor; the RNA sequence pre-training unit is implemented based on a preset pre-training model; the preset pre-training model is a pre-training model based on a Transformer structure, a pre-training model based on a CNN structure, a pre-training model based on an RNN structure, a pre-training model based on an autoencoder structure, a pre-training model based on a variational autoencoder structure, or a pre-training model based on an autoregressive model structure; The RNA sequence feature dimensionality reduction unit is used to perform feature dimension reduction processing on the feature tensor output by the RNA sequence pre-training unit and output a corresponding reduced dimension tensor; the RNA sequence feature dimensionality reduction unit is implemented based on a type of linear neural network; the type of linear neural network includes a fully connected neural network; The ribosome load prediction unit is used to perform ribosome load prediction processing based on the dimensionality reduction tensor output by the RNA sequence feature dimensionality reduction unit to generate a corresponding prediction tensor; the ribosome load prediction unit is implemented based on a preset prediction model; the preset prediction model is a prediction model based on a multi-layer ResNet structure, a prediction model based on a Transformer structure, a prediction model based on a CNN structure, or a prediction model based on a GNN structure; The average ribosome load prediction unit is used to perform average ribosome load estimation processing based on the prediction tensor output by the ribosome load prediction unit and output the obtained estimation data as the average ribosome load prediction result of the first prediction system; the average ribosome load prediction unit is implemented based on a second-class linear neural network; the second-class linear neural network includes a fully connected neural network.

3. The processing method of the average ribosome load prediction system according to claim 2, characterized in that: The performing a round of system training on the first prediction system based on the training data set and the test data set obtained this time specifically includes: Step 51: Use the training data set as the corresponding current data set; and set the first state to be the training state; the first state includes the training state and the test state; Step 52: Select the first of the first sequences from the current data set as the corresponding current sequence; Step 53: using the ribosome load of the current sequence as the corresponding current sequence tag; and performing standardization processing on the current sequence tag to generate corresponding first tag data; Step 54: input the current sequence into the first prediction system; the RNA sequence pre-training unit performs feature encoding on the current sequence and outputs a corresponding first feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the first feature tensor and outputs a corresponding first reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the first reduced dimensionality tensor to generate a corresponding first prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the first prediction tensor to generate corresponding first estimation data; and the first estimation data is output as the average ribosome load prediction result of the first prediction system; Step 55: Substitute the first estimated data and the first label data into a preset error function to calculate a corresponding first prediction error; Step 56, identifying the first state; when the first state is a training state, go to step 57; when the first state is a test state, go to step 58; Step 57: Identify whether the first prediction error satisfies a preset first reasonable error range; if so, proceed to step 59; if not, perform full model parameter modulation on the RNA sequence pre-training unit and the ribosome load prediction unit of the first prediction system based on a preset system optimization algorithm, and return to step 54 to continue training after parameter modulation is completed; the system optimization algorithm includes a gradient descent method, a genetic optimization algorithm, a simulated annealing algorithm, a particle swarm optimization algorithm, and a differential evolution algorithm; the full model parameter modulation is to modulate all model parameters of the RNA sequence pre-training unit and the ribosome load prediction unit; Step 58: Identify whether the first prediction error meets a preset second reasonable error range; if so, proceed to step 59; if not, modulate the specified model parameters of the ribosome load prediction unit of the first prediction system based on the system optimization algorithm and return to step 54 to continue training after the parameter modulation is completed; the specified model parameter modulation is to modulate a portion of the model parameters pre-specified in the ribosome load prediction unit; Step 59: Identify whether the current sequence is the last first sequence in the current data set; if so, proceed to step 60; if not, select the next first sequence from the current data set as the new current sequence and return to step 53 to continue training; Step 60, identifying the first state; when the first state is the training state, using the test data set as the new current data set, setting the first state to the test state, and returning to step 52 to continue training; when the first state is the test state, stopping training and confirming the end of this round of system training.

4. The processing method of the average ribosome load prediction system according to claim 2, characterized in that: The performing average ribosome load prediction processing on the first target RNA sequence based on the first prediction system specifically includes: The first target RNA sequence is input into the first prediction system; the RNA sequence pre-training unit performs feature encoding on the first target RNA sequence and outputs a corresponding second feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the second feature tensor and outputs a corresponding second reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the second reduced dimensionality tensor to generate a corresponding second prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the second prediction tensor to generate corresponding second estimation data; and the second estimation data is output as the processing result of the average ribosome load prediction processing.

5. The processing method of the average ribosome load prediction system according to claim 2, characterized in that: The performing target RNA sequence sorting processing on the plurality of second target RNA sequences based on the first prediction system specifically includes: Input each second target RNA sequence into the first prediction system respectively; the RNA sequence pre-training unit performs feature encoding on the currently input second target RNA sequence and outputs a corresponding third feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the third feature tensor and outputs a corresponding third reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the third reduced dimensionality tensor to generate a corresponding third prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the third prediction tensor to generate corresponding third estimation data; and the third estimation data is output as the average ribosome load prediction result of the first prediction system; According to the preset evaluation parameter rules, the evaluation parameters are calculated according to each of the third estimation data to obtain the corresponding first evaluation parameters; and the first evaluation parameters are sorted according to the sorting rules corresponding to the evaluation parameter rules; and the second target RNA sequences corresponding to each of the first evaluation parameters are sorted based on the sorting order of all the first evaluation parameters to obtain the corresponding second target RNA sequence sorting array; and the obtained second target RNA sequence sorting array is output as the processing result of the target RNA sequence sorting process; the evaluation parameter rules include RNA translation efficiency parameter rules, RNA translation rate parameter rules and RNA ribosome retention frequency parameter rules; the RNA translation efficiency parameter rules are corresponding rules for calculating translation efficiency parameters based on average ribosome load, and the RNA translation The sorting rule corresponding to the translation efficiency parameter rule is to sort all the second target RNA sequences in order of translation efficiency from high to low; the RNA translation rate parameter rule is the corresponding rule for calculating the translation rate parameter based on the average ribosome load, and the sorting rule corresponding to the RNA translation rate parameter rule is to first sort all the second target RNA sequences by level based on a pre-specified multi-level translation rate range, and then sort multiple second target RNA sequences of the same level in order of translation rate from high to low; the RNA ribosome retention frequency parameter rule is the corresponding rule for calculating the ribosome retention frequency parameter based on the average ribosome load, and the sorting rule corresponding to the RNA ribosome retention frequency parameter rule is to sort all the second target RNA sequences in order of ribosome retention frequency from low to high.

6. The processing method of the average ribosome load prediction system according to claim 2, characterized in that: The performing sequence optimization processing on the first target mRNA sequence based on the first prediction system specifically includes: Step 81, using the first target mRNA sequence as the corresponding current mRNA sequence; Step 82: Input the current mRNA sequence into the first prediction system; the RNA sequence pre-training unit performs feature encoding on the current mRNA sequence and outputs a corresponding fourth feature tensor; the RNA sequence feature dimensionality reduction unit performs feature dimension reduction processing on the fourth feature tensor and outputs a corresponding fourth reduced dimensionality tensor; the ribosome load prediction unit performs ribosome load prediction processing based on the fourth reduced dimensionality tensor to generate a corresponding fourth prediction tensor; the average ribosome load prediction unit performs average ribosome load estimation processing based on the fourth prediction tensor to generate corresponding fourth estimation data; and the fourth estimation data is output as the average ribosome load prediction result of the first prediction system; Step 83, identifying whether the fourth estimated data exceeds a preset optimal average ribosome load threshold; if not, proceeding to step 84; if so, proceeding to step 85; Step 84, performing sequence optimization processing on the current mRNA sequence based on a preset mRNA sequence optimization algorithm in a direction of maximizing the ribosome load prediction of the ribosome load prediction unit to obtain a corresponding mRNA optimized sequence; and using the mRNA optimized sequence as the new current mRNA sequence and returning to step 82; the mRNA sequence optimization algorithm includes a sequence optimization algorithm implemented based on a simulated annealing algorithm and a sequence optimization algorithm implemented based on a genetic optimization algorithm; Step 85: output the latest current mRNA sequence as the processing result of the sequence optimization process.

7. A device for executing the processing method of the average ribosome load prediction system according to any one of claims 1 to 6, characterized in that: The device includes: a prediction system construction and training module, an application information acquisition module, a first-class application processing module, a second-class application processing module, and a third-class application processing module; The prediction system construction and training module is used to construct an average ribosome load prediction system, which is recorded as a corresponding first prediction system; and to construct a data set, which is recorded as a corresponding first data set; and to train the first prediction system based on the first data set; The application information acquisition module is used to receive a system application mode and corresponding system application data input by a user after the system training is completed; the system application mode includes a single sequence prediction mode, a multiple sequence sorting mode, and a single sequence optimization mode; when the system application mode is the single sequence prediction mode, the system application data is the corresponding first target RNA sequence; when the system application mode is the multiple sequence sorting mode, the system application data is the corresponding multiple second target RNA sequences; when the system application mode is the single sequence optimization mode, the system application data is the corresponding first target mRNA sequence; The first type of application processing module is used for performing average ribosome load prediction processing on the first target RNA sequence based on the first prediction system when the system application mode is a single sequence prediction mode; The second type of application processing module is used for performing target RNA sequence sorting processing on the multiple second target RNA sequences based on the first prediction system when the system application mode is a multiple sequence sorting mode; The three types of application processing modules are used to perform sequence optimization processing on the first target mRNA sequence based on the first prediction system when the system application mode is a single sequence optimization mode.

8. An electronic device, characterized in that: include: memory, processors, and transceivers; The processor is configured to be coupled to the memory, read and execute instructions in the memory, so as to implement the method according to any one of claims 1 to 6; The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a computer, the computer is caused to execute the method according to any one of claims 1 to 6.

Images