Method, device, medium and equipment for generating protein sequence of highly thermostable enzyme

Abstract

The present invention discloses a method, device, medium and equipment for generating protein sequences of highly thermostable enzymes. The protein sequence generation method comprises: obtaining training samples, the training samples comprising protein sequence data of a specific type of enzymes with a tolerance temperature greater than a predetermined value; using the training samples to train a pre-constructed generative adversarial network model to obtain a protein sequence generation model; and using the protein sequence generation model to generate batch protein sequence data. This method uses the existing protein sequence data of a specific type of enzymes with a tolerance temperature greater than a predetermined value to train a generative adversarial network model to obtain a protein sequence generation model, which can generate sequences of a specific type of enzymes with high thermostable stability in batches. The generation method is simple, starting from the protein sequence only and completing the design on a computer, and can be further experimentally verified according to the similarity distribution, and the results are more reliable and easy to analyze.

Description

Technical Field

[0001] The present invention belongs to the field of biomedicine technology, and specifically relates to a method for generating a protein sequence of a highly thermostable enzyme, a protein sequence generating device, a computer-readable storage medium, and a computer device. Background Art

[0002] Highly thermostable enzymes play an extremely important role in the fields of biofuels and biochemical engineering. The traditional method of obtaining them is mainly to separate them from thermophilic bacteria, and then put them into industrial use after experimental verification, improvement and optimization. However, the number of enzymes separated by traditional separation methods is limited, which cannot meet the increasingly abundant needs in specific industrial scenarios. It is particularly important to design some new highly thermostable enzymes from scratch. At present, there are two mainstream methods, one is based on rational design, which is mainly transformed through structural modification, and the other is based on directed evolution. However, the ability of these transformations has certain limitations, and the number of highly thermostable enzymes obtained is limited. Among them, the method based on rational design requires a more detailed understanding of information such as enzyme structure, and is very familiar with existing transformation methods. The transformation process is relatively complicated and it is difficult to generate in batches; the method based on directed evolution selects the corresponding enzymes by establishing a random mutation library, but the success rate of this method is low, and the workload is huge, so it is difficult to generate in batches. At the same time, neither method conducts a systematic comparative analysis of sequences from a theoretical perspective.

[0003] Therefore, it is urgent to develop a method for mass-designing new enzymes with high thermal stability. Summary of the invention

[0004] (I) Technical Problems to be Solved by the Present Invention

[0005] The technical problem solved by the present invention is: how to quickly and in batches generate the protein sequence of a high-thermostability enzyme.

[0006] (II) Technical solution adopted by the present invention

[0007] A method for generating a protein sequence of a highly thermostable enzyme, characterized in that the method comprises:

[0008] Acquire a training sample, wherein the training sample includes protein sequence data of a specific type of enzyme that tolerates a temperature greater than a predetermined value;

[0009] The pre-built generative adversarial network model is trained using training samples to obtain a protein sequence generation model;

[0010] The protein sequence generation model is used to generate batch protein sequence data.

[0011] The generative adversarial network model includes a generator and a discriminator. The specific method of using training samples to train the pre-built generative adversarial network model to obtain a protein sequence generation model includes:

[0012] Inputting random noise into a generator, the generator outputs generated data, and selecting part of the data from the training samples as real data;

[0013] The generated data and the real data are inputted into the discriminator, and the discriminator outputs a discrimination result;

[0014] Adjusting the network parameters of the generator and the discriminator according to the discrimination result to complete a round of training;

[0015] The above training steps are repeated until the predetermined training conditions are met to obtain the protein sequence generation model.

[0016] Preferably, the method for obtaining training samples includes:

[0017] Obtaining protein group sequences of various microorganisms with tolerance temperatures greater than a predetermined value;

[0018] Determine a target enzyme in the same category as the specific enzyme from the protein group sequence, and extract a target protein sequence of the target enzyme;

[0019] The target protein sequence is clustered using a sequence clustering algorithm to obtain protein sequences of multiple clusters, and protein sequences with clusters greater than a threshold are selected as training samples.

[0020] Preferably, the training sample also includes protein sequence data of an initial sample enzyme in the same category as the specific enzyme. Before using the training sample to train the pre-constructed generative adversarial network model, the protein sequence generation method also includes:

[0021] The generative adversarial network model is pre-trained using the protein sequence data of the initial sample enzyme to obtain the pre-constructed generative adversarial network model.

[0022] Preferably, the specific type of enzyme and the initial sample enzyme both belong to any enzyme under the fourth level enzymatic number in the Enzyme Commission Nomenclature.

[0023] Preferably, the protein sequence generation method further comprises:

[0024] Comparing each protein sequence in the batch protein sequence data with the partial protein sequence data to determine the similarity of each protein sequence;

[0025] Sort each protein sequence in descending order according to similarity and divide it into multiple intervals;

[0026] Selecting a number of protein sequences from each interval, and synthesizing and expressing each selected protein sequence to generate a corresponding enzyme;

[0027] The dissolution temperature of each enzyme is determined, and the protein sequence corresponding to the enzyme with a dissolution temperature greater than or equal to a predetermined temperature is screened out as a new protein sequence with high thermal stability.

[0028] Preferably, the method of comparing each protein sequence in the batch protein sequence data with the partial protein sequence data to determine the similarity of each protein sequence comprises:

[0029] Calculating the similarity between each protein sequence and each sequence in the partial protein sequence data respectively to obtain a set of similarity data;

[0030] The maximum similarity in the set of similarity data is taken as the similarity of each protein sequence.

[0031] The present application also discloses a protein sequence generation device for a high thermostable enzyme, the protein sequence generation device comprising:

[0032] A sample acquisition unit, which acquires a training sample, wherein the training sample includes protein sequence data of a specific type of enzyme that tolerates a temperature greater than a predetermined value;

[0033] A training unit, used to train a pre-built generative adversarial network model using training samples to obtain a protein sequence generation model;

[0034] The batch generation unit is used to generate batch protein sequence data using the protein sequence generation model.

[0035] The present application also discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a protein sequence generation program for a highly thermostable enzyme. When the protein sequence generation program for a highly thermostable enzyme is executed by a processor, the above-mentioned protein sequence generation method for a highly thermostable enzyme is implemented.

[0036] The present application also discloses a computer device, which includes a computer-readable storage medium, a processor, and a protein sequence generation program for a highly thermostable enzyme stored in the computer-readable storage medium. When the protein sequence generation program for a highly thermostable enzyme is executed by the processor, the above-mentioned protein sequence generation method for a highly thermostable enzyme is implemented.

[0037] (III) Beneficial effects

[0038] The present invention discloses a method for generating a protein sequence of a highly thermostable enzyme, which has the following technical effects compared with the traditional generation method:

[0039] By using the existing protein sequence data of a specific type of enzyme with a tolerance temperature greater than a predetermined value to train a generative counterpart model, a protein sequence generation model can be obtained, which can batch generate sequences of a specific type of enzyme with high thermal stability. The generation method is simple, starting from the protein sequence only and completing the design on a computer, and can be further experimentally verified according to the similarity distribution. The results are more reliable and easier to analyze. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a flow chart of the method for generating the protein sequence of a highly thermostable enzyme according to Example 1 of the present invention;

[0041] Figure 2 Another flow chart of another protein sequence generation method according to embodiment 1 of the present invention;

[0042] Figure 3 This is a flow chart of the method for generating the protein sequence of a highly thermostable enzyme according to Example 2 of the present invention;

[0043] Figure 4 This is a schematic diagram of a device for generating a protein sequence of a highly thermostable enzyme according to Example 3 of the present invention;

[0044] Figure 5 Schematic diagram of a computer device according to Embodiment 5 of the present invention. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0046] Before describing the various embodiments of the present application in detail, the inventive concept of the present application is first briefly described: the existing design method of highly thermostable enzymes requires a relatively good understanding of information such as enzyme structure, and the modification process is relatively complicated and difficult to generate in batches. The protein sequence generation method of highly thermostable enzymes provided in the present application first screens out protein sequence data of a specific type of enzyme whose tolerance temperature is greater than a predetermined value as a training sample, and then uses the training sample to train a pre-built generative adversarial network model to obtain a protein sequence generation model, so that the model learns the basic characteristics of highly thermostable enzymes, and finally uses the trained protein sequence generation model to generate protein sequence data in batches. The generation method is simple and can be completed only on a computer, which improves the efficiency of automated design.

[0047] Specifically, if Figure 1As shown, this embodiment 1 provides a method for generating a protein sequence of a highly thermostable enzyme, comprising the following steps:

[0048] Step S10, obtaining a training sample, wherein the training sample includes protein sequence data of a specific type of enzyme having a tolerance temperature greater than a predetermined value;

[0049] Step S20: training the pre-built generative adversarial network model using the training samples to obtain a protein sequence generation model;

[0050] Step S30: Generate batch protein sequence data using the protein sequence generation model.

[0051] Specifically, the main purpose of step S10 is to construct a protein sequence database of highly thermostable enzymes for providing training samples. Step S10 includes: obtaining a protein group sequence of a microorganism with a tolerance temperature greater than a predetermined value; determining a target enzyme in the same category as the specific enzyme from the protein group sequence, and extracting a target protein sequence of the target enzyme; clustering the target protein sequence using a sequence clustering algorithm to obtain protein sequences of multiple clusters, and selecting the protein sequence of the largest cluster as a training sample.

[0052] Exemplarily, the growth temperatures of a large number of microorganisms can be obtained from existing databases, such as the Martin KM Engqvist database and other databases containing microbial growth temperatures; then, a suitable predetermined value is set, and microorganisms with a tolerance temperature above the predetermined value, such as above 40 degrees Celsius, are selected; next, the proteome sequences of the above-mentioned microorganisms are extracted from the proteome database, such as the Uniprot / Proteomes database and other databases containing proteomes; next, the target protein sequences corresponding to the target enzymes in the same category as the specific enzyme in all the above-mentioned proteomes are extracted; finally, a sequence clustering algorithm, such as mmseq2 clustering and other clustering methods suitable for sequences, is used to set a reasonable similarity threshold ranging from 0.0 to 1.0, such as 0.5, and protein sequences with clusters greater than the threshold are selected from the target protein sequences as training samples.

[0053] Furthermore, in order to improve the accuracy of the protein sequence generation model, the protein sequences used as training samples should have a high degree of similarity. The specific enzyme of Example 1 and the target enzyme both belong to any enzyme under the fourth-level enzymatic number in the Enzyme Commission nomenclature. For example, both belong to 3.5.4.5-cytidine deaminase under the fourth-level enzymatic number. This can avoid large differences in training samples and affect the training effect. There are many types of enzymes under the fourth-level enzymatic number in the Enzyme Commission nomenclature. The protein sequences of various enzymes are different, but they can all catalyze the same type of reaction and can be considered to have a high degree of similarity. This can ensure that the protein sequence generation model obtained through training has a high degree of accuracy.

[0054] Furthermore, after step S10, the training samples need to be preprocessed, the protein sequence data of the specific enzyme class is aligned, all protein sequences are padded with non-amino acid characters, and converted into one-hot encoding form as input data suitable for model training.

[0055] In step S20, the generative adversarial network model includes a generator and a discriminator, and the method of using training samples to train the pre-constructed generative adversarial network model includes: inputting random noise into the generator, the generator outputs generated data, and selecting part of the data from the training samples as real data; inputting the generated data and the real data into the discriminator together, and the discriminator outputs a discrimination result; adjusting the network parameters of the generator and the discriminator according to the discrimination result to complete a round of training; repeating the above training steps until the predetermined training conditions are met to obtain the generative model. Exemplarily, the generative adversarial network model adopts a WGAN-GP network.

[0056] In each round of training, the input format of the real data is [Batch_size, Seq_Len, charmap], where Batch_size is the number of sequences input to the model each time, Seq_Len is the unified sequence length, and charmap is the dimension of the dictionary, including 20 types of amino acids and padding characters. The input of the generator is random noise that satisfies the standard normal distribution. After multiple rounds of iterative learning, the generator of the model will generate data similar to the real protein sequence of a specific enzyme, thus completing the extraction of the high thermal stability characteristics of a specific enzyme, thereby completing the training of the protein sequence generation model.

[0057] As a preferred embodiment, Figure 2 As shown, in order to further determine the reliability of the protein sequences generated in batches, the protein sequence generation method further includes the following steps:

[0058] Step S40: Compare each protein sequence in the batch protein sequence data with the partial protein sequence data to determine the similarity of each protein sequence. Exemplarily, the method for determining the similarity of each protein sequence includes: calculating the similarity between each protein sequence and each sequence in the partial protein sequence data to obtain a set of similarity data; and taking the maximum similarity in the set of similarity data as the similarity of each protein sequence.

[0059] Step S50: sort the protein sequences in descending order of similarity and divide them into multiple intervals. For example, a similarity of 90% or more is regarded as an interval, a similarity of 80% to 90% is regarded as an interval, a similarity of 70% to 80% is regarded as an interval, and so on, to form multiple intervals.

[0060] Step S60: selecting a number of protein sequences from each interval, and performing synthetic expression according to each selected protein sequence to generate a corresponding enzyme;

[0061] Step S70, determining the dissolution temperature of each enzyme, and screening out protein sequences corresponding to enzymes having a dissolution temperature greater than or equal to a predetermined temperature as new protein sequences with high thermal stability.

[0062] The method for generating the protein sequence of a highly thermostable enzyme disclosed in Example 1 can generate sequences of a specific class of highly thermostable enzymes in batches. The generation method is simple, starting from the protein sequence only and completing the design on a computer. It can be further experimentally verified according to the similarity distribution, and the results are more reliable and easier to analyze.

[0063] Furthermore, the main difference between the method for generating protein sequences of highly thermostable enzymes in Example 2 and Example 1 is that the model is initially trained, which can better initialize the model parameters and make up for the impact caused by the small amount of protein sequence data of a specific type of enzyme with a tolerance temperature greater than a predetermined value. Specifically, Figure 3 As shown, the method for generating the protein sequence of the highly thermostable enzyme in the second embodiment includes the following steps:

[0064] Step S10', obtaining a training sample, wherein the training sample includes protein sequence data of a specific type of enzyme whose tolerance temperature is greater than a predetermined value and protein sequence data of an initial sample enzyme in the same category as the specific type of enzyme;

[0065] Step S20′, pre-training the generative adversarial network model using the protein sequence data of the initial sample enzyme to obtain the pre-constructed generative adversarial network model, and then training the pre-constructed generative adversarial network model using the protein sequence data of a specific type of enzyme to obtain a protein sequence generation model;

[0066] Step S30': Generate batch protein sequence data using the protein sequence generation model.

[0067] Specifically, when there is less protein sequence data of a specific type of enzyme whose tolerance temperature is greater than a predetermined value, other enzymes in the same category as the specific type of enzyme can be selected as initial sample enzymes, and the generative adversarial network model can be pre-trained using the protein sequence data of the initial sample enzyme, so that the model learns the basic characteristics of the specific type of enzyme. As a preferred embodiment, in order to avoid a large difference between the initial sample enzyme and the specific type of enzyme, the initial sample enzyme and the specific type of enzyme are limited to any enzyme under the fourth-level enzymatic number in the Enzyme Commission nomenclature, for example, both belong to 3.5.4.5-cytidine deaminase under the fourth-level enzymatic number. At this time, the tolerance temperature of the initial sample enzyme does not need to be considered, as long as the protein sequences of other enzymes under the same fourth-level enzymatic number as the specific type of enzyme can be used for pre-training.

[0068] Exemplarily, the method for obtaining the protein sequence of the initial sample enzyme is: first download all naturally occurring sequences of the initial sample enzyme from a protein database, such as the Uniprot database and other databases where protein sequences can be obtained; then extract the target sequence belonging to the initial sample enzyme based on the classic five-kingdom division criteria, i.e., one of the prokaryotes, protists, fungi, plants, and animals. In the second embodiment, the protein sequence of the initial sample enzyme in the prokaryotes is preferably used for pre-training.

[0069] In step S20', the process of pre-training the generative adversarial network model using the protein sequence data of the initial sample enzyme is as follows: the generator of the generative adversarial network model obtains generated data according to random noise, and selects part of the data from the protein sequence data of the initial sample enzyme as real data; the generated data and the real data are input into the discriminator of the generative adversarial network model together to obtain a discrimination result; the network parameters of the discriminator and the generator of the generative adversarial network model are updated according to the discrimination result to complete a round of training; the above training steps are repeated until the initial training conditions are met to obtain a pre-constructed generative adversarial network model. Then, the protein sequence data of a specific class of enzymes with a tolerance temperature greater than a predetermined value are used to continue training the pre-constructed generative adversarial network model to obtain a final protein sequence generation model. Among them, the details of data preprocessing, data input format, etc. in the training process of step S20' can refer to the training process of step S20 of embodiment 1, and will not be repeated here.

[0070] Furthermore, step S30' in the second embodiment is the same as step S30 in the first embodiment, and after step S30' in the second embodiment, the protein sequence generation method further includes the contents of steps S40 to S70 in the first embodiment, which will not be described in detail here.

[0071] The method for generating the protein sequence of a highly thermostable enzyme disclosed in the second embodiment has the following advantages over the method in the first embodiment: it makes up for the impact caused by the small amount of protein sequence data of a specific type of enzyme that tolerates a temperature greater than a predetermined value, thereby improving the accuracy of the protein sequence generation model to generate protein sequences with higher similarity.

[0072] Furthermore, if Figure 4 As shown, this embodiment three also discloses a protein sequence generation device for a thermostable enzyme, and the protein sequence generation device includes a sample acquisition unit 100, a training unit 200, and a batch generation unit 300. The sample acquisition unit 100 is used to obtain training samples, and the training samples include protein sequence data of a specific type of enzyme that tolerates a temperature greater than a predetermined value; the training unit 200 uses the training samples to train a pre-constructed generative adversarial network model to obtain a protein sequence generation model; the batch generation unit 300 is used to generate batch protein sequence data using the protein sequence generation model. Among them, the specific working process of the sample acquisition unit 100, the training unit 200, and the batch generation unit 300 can refer to the relevant description of the embodiment one, and will not be repeated here.

[0073] This fourth embodiment also discloses a computer-readable storage medium, which stores a program for generating a protein sequence of a highly thermostable enzyme. When the program for generating a protein sequence of a highly thermostable enzyme is executed by a processor, the above-mentioned method for generating a protein sequence of a highly thermostable enzyme is implemented.

[0074] This fifth embodiment also discloses a computer device, at the hardware level, such as Figure 5 As shown, the terminal includes a processor 12, an internal bus 13, a network interface 14, and a computer-readable storage medium 11. The processor 12 reads the corresponding computer program from the computer-readable storage medium and then runs it, forming a request processing device at the logical level. Of course, in addition to software implementations, one or more embodiments of this specification do not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is to say, the execution subject of the following processing flow is not limited to each logical unit, but can also be hardware or logic devices. The computer-readable storage medium 11 stores a protein sequence generation program for a highly thermostable enzyme, and when the protein sequence generation program for a highly thermostable enzyme is executed by the processor, the above-mentioned protein sequence generation method for a highly thermostable enzyme is implemented.

[0075] Computer-readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, modules of programs or other data. Examples of computer-readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage, quantum memory, graphene-based storage media or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0076] The specific implementation methods of the present invention are described in detail above. Although some embodiments have been shown and described, those skilled in the art should understand that these embodiments can be modified and improved without departing from the principles and spirit of the present invention, the scope of which is defined by the claims and their equivalents. These modifications and improvements should also be within the scope of protection of the present invention.

Claims

1. A method for generating a protein sequence of a highly thermostable enzyme, characterized in that: The protein sequence generation method comprises: Acquire a training sample, wherein the training sample includes protein sequence data of a specific type of enzyme that tolerates a temperature greater than a predetermined value; The pre-built generative adversarial network model is trained using training samples to obtain a protein sequence generation model; Generate batch protein sequence data using the protein sequence generation model; The protein sequence generation method further comprises: Comparing each protein sequence in the batch protein sequence data with the partial protein sequence data to determine the similarity of each protein sequence, including: calculating the similarity between each protein sequence and each sequence in the partial protein sequence data to obtain a set of similarity data; taking the maximum similarity in the set of similarity data as the similarity of each protein sequence; Sort each protein sequence in descending order according to similarity and divide it into multiple intervals; Selecting a number of protein sequences from each interval, and synthesizing and expressing each selected protein sequence to generate a corresponding enzyme; Determine the dissolution temperature of each enzyme, and screen out the protein sequence corresponding to the enzyme whose dissolution temperature is greater than or equal to the predetermined temperature as a new protein sequence with high thermal stability; The training sample also includes protein sequence data of an initial sample enzyme in the same category as the specific enzyme. Before using the training sample to train the pre-built generative adversarial network model, the protein sequence generation method also includes: The generative adversarial network model is pre-trained using the protein sequence data of the initial sample enzyme to obtain the pre-constructed generative adversarial network model.

2. The method for generating a protein sequence of a highly thermostable enzyme according to claim 1, characterized in that: The generative adversarial network model includes a generator and a discriminator. The specific method of using training samples to train the pre-built generative adversarial network model to obtain a protein sequence generation model includes: Inputting random noise into a generator, the generator outputs generated data, and selecting part of the data from the training sample as real data; The generated data and the real data are inputted into the discriminator, and the discriminator outputs a discriminant result; Adjust the network parameters of the generator and the discriminator according to the discrimination result to complete a round of training; The above training steps are repeated until the predetermined training conditions are met to obtain the protein sequence generation model.

3. The method for generating a protein sequence of a highly thermostable enzyme according to claim 2, characterized in that: Methods for obtaining training samples include: Obtaining protein group sequences of various microorganisms with tolerance temperatures greater than a predetermined value; Determine a target enzyme in the same category as the specific enzyme from the protein group sequence, and extract a target protein sequence of the target enzyme; The target protein sequence is clustered using a sequence clustering algorithm to obtain protein sequences of multiple clusters, and protein sequences with clusters greater than a threshold are selected as training samples.

4. The method for generating a protein sequence of a highly thermostable enzyme according to claim 1, characterized in that: The specific enzyme and the initial sample enzyme both belong to any enzyme under the fourth-level enzymatic number in the Enzyme Commission Nomenclature.

5. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program for generating a protein sequence of a thermostable enzyme, and when the program for generating a protein sequence of a thermostable enzyme is executed by a processor, the method for generating a protein sequence of a highly thermostable enzyme according to any one of claims 1 to 4 is implemented.

6. A computer device, characterized in that: The computer device includes a computer-readable storage medium, a processor, and a program for generating a protein sequence of a thermostable enzyme stored in the computer-readable storage medium. When the program for generating a protein sequence of a thermostable enzyme is executed by the processor, the method for generating a protein sequence of a highly thermostable enzyme according to any one of claims 1 to 4 is implemented.

Images