Recommendation method and device for recommending mutable sites through small sample experiment data

Abstract

The invention discloses a recommendation method and device for recommending mutable sites from small sample experiment data, and the method comprises the following steps: obtaining the small sample experiment data, including sequences and data of wild enzymes and mutants, and obtaining an optimal mutant sequence according to the data type of the mutants; predicting according to the optimal mutant sequence to obtain a mutant structure; predicting according to the mutant structure to obtain a substrate binding pocket; predicting a single mutation site based on a protein structure model, and selecting a residue with a distance from the center of a substrate binding pocket smaller than a threshold value 1 as a recommended single mutation site; selecting sites in small sample experimental data, mutating each site in a site set into 19 other amino acids, combining every two other amino acids to construct a double mutation set, predicting a mutation result, and obtaining recommended double mutation sites according to a sorting result; selecting sites in small sample experiment data, obtaining coordinates of the sites in a variant structure for clustering analysis, and selecting one site from each cluster to be combined with other clusters to construct multiple mutations; predicting mutation results of multiple mutations, and obtaining recommended multiple mutation sites according to a sorting result; the method is effectively operated under the condition of small sample data; through the strong generalization ability of the large language model, the structure and function information in the biomolecules can be captured, so that the sequence of the enzyme is effectively coded for more accurate inference.

Description

Technical Field

[0001] The present invention relates to the field of bioinformatics, and in particular to a method and device for recommending mutable sites based on small sample experimental data. Background Art

[0002] Enzymes are biological catalysts composed of proteins or RNA that can accelerate chemical reactions in organisms. By reducing the activation energy of the reaction, enzymes achieve efficient catalysis under milder conditions and are highly specific to substrates. They are widely present in organisms and regulate almost all life processes.

[0003] The limitations of wild enzymes are mainly reflected in their catalytic efficiency, stability and substrate specificity. They usually can only work efficiently under the natural conditions of the organism and are sensitive to temperature, pH and other environmental factors. In addition, the substrate range of wild enzymes is limited and they are often unable to adapt to a wide range of reaction requirements under industrial or experimental conditions.

[0004] Enzyme modification optimizes enzyme performance through techniques such as directed evolution or rational design, but faces many challenges. The problem is particularly prominent when recommending mutation sites through small sample experimental data. The structure of enzymes is complex, and a small number of mutations may lead to unpredictable changes in global performance. Since the mutation effects of enzymes often have synergistic or antagonistic effects, mutation prediction based on small-scale data is difficult to accurately measure the global impact of mutations. In addition, when the data sample is insufficient, it may not be possible to capture the complex relationship between enzyme function and structure, thus limiting the effect of mutation site optimization. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for recommending mutable sites based on small sample experimental data, so as to overcome the deficiencies in the prior art.

[0006] To achieve the above object, the present invention provides the following technical solutions: The present application discloses a method for recommending mutable sites based on small sample experimental data, comprising the following steps: S1. Obtaining small sample experimental data, wherein the small sample experimental data includes sequences and data of wild enzymes and mutants, obtaining an optimal mutant sequence according to the data type of the mutant; predicting a mutant structure according to the optimal mutant sequence; and predicting a substrate binding pocket according to the mutant structure; S2. Predicting single mutation sites based on the protein structure model to obtain prediction results, and selecting residues whose distance from the center of the substrate binding pocket in step S1 is less than a threshold value one as recommended single mutation sites in the prediction results; S3, based on the small sample experimental data of step S1, training the neural network model through the protein large language model; S4. According to the data type of the mutant, the sites corresponding to the mutant sequences that are better than the wild enzyme data are selected in the small sample experimental data, as well as the sites whose distance from the center of the substrate binding pocket in step S1 is less than the threshold 2, to obtain a site set, and each site in the site set is mutated to 19 other amino acids, and two sites are combined to construct a double mutation set, and the neural network model of step S3 is used to predict the mutation results, and the predicted results are sorted, and the recommended double mutation sites are obtained according to the sorting results; S5. According to the data type of the mutant, the site corresponding to the mutant sequence that is better than the wild enzyme data is selected in the small sample experimental data, and the coordinates of the site are obtained in the mutant structure of step S1 for cluster analysis to obtain several clusters; S51, selecting one site from each cluster and combining it with other clusters to construct multiple mutations; and using the neural network model of step S3 to predict the mutation results of the multiple mutations, sorting them according to the predicted results, and obtaining recommended multiple mutation sites according to the sorting results; S52. From each cluster, mutations whose mutation results are at least 1.5 times better than those of the wild enzyme are selected according to the sorting of mutation results. The mutations selected by each cluster are combined with the mutations selected by other clusters as recommended multiple mutation sites.

[0007] Preferably, the small sample experimental data is provided by biological experimenters or obtained from a public mutation data set.

[0008] Preferably, the data types include enzyme activity data, ligand binding data or temperature tolerance data; the method for obtaining the optimal mutant sequence according to the data type of the mutant in step S1 is as follows: according to the data type of the mutant, the data are sorted from best to worst, and the sequence of the mutant with the best data is selected.

[0009] Preferably, in step S1, AlphaFold, chai_lab or helixfold3 is used to predict the protein structure corresponding to the optimal mutant sequence to obtain the mutant structure.

[0010] Preferably, in step S1, p2rank is used to predict the substrate binding pocket of the mutant structure.

[0011] Preferably, in step S2, the threshold value is 15 angstroms.

[0012] Preferably, step S3 specifically includes the following steps: S31, using ESM or ProtTrans protein language model; S32, using a large protein language model to convert sequences in small sample experimental data into numbers as training features X, using mutation results as Y, training a neural network model, and calculating the correlation between mutation results in the test set and predicted mutation results; S33, obtaining homologous sequences in small sample experimental data to fine-tune the protein large language model, training the neural network model and calculating the correlation between the mutation results in the test set and the predicted mutation results; S34. Compare the correlation results of S32 and S33, and select the neural network model with the largest correlation as the final neural network model.

[0013] Preferably, the threshold value in step S4 is 15 angstroms, the mutation results are sorted from large to small, and topN is selected as the recommended double mutation site according to the sorting results.

[0014] Preferably, in step S51, the mutation results are sorted from large to small, and topN is selected as the recommended multiple mutation sites according to the sorting results.

[0015] The present invention also discloses a device for recommending mutable sites based on small sample experimental data, comprising a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, it is used for the above-mentioned method for recommending mutable sites based on small sample experimental data.

[0016] Beneficial effects of the present invention: Small sample adaptability: Traditional methods often require a large amount of experimental data to capture the complex relationship between enzyme function and structure, while the method of the present invention can also operate effectively under small sample data conditions; through the powerful generalization ability of the large language model, the structural and functional information in biological molecules can be captured, thereby effectively encoding the enzyme sequence for more accurate inference.

[0017] Efficient mutation screening: The method of the present invention can quantify the effect of mutations on enzyme performance through neural networks, significantly improving the efficiency of mutation site screening; compared with traditional experimental exploration, the method of the present invention can propose more promising mutation combinations in a short period of time, thereby reducing the blindness of the experiment.

[0018] Rapid experimental verification: Through the predictive ability of the model, the activity of the target enzyme was increased by more than 6 times in 1 week of calculation time and 2 weeks of experimental time. This rapid feedback capability not only has theoretical advantages, but also shows excellent results in actual experimental verification; The features and advantages of the present invention will be described in detail through embodiments in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic flow chart of a method for recommending mutable sites based on small sample experimental data of the present invention; Figure 2 is the pocket information of the substrate binding pocket of the embodiment of the present invention; Figure 3 is the result of predicting the mutation point by the protein structure model of the embodiment of the present invention; Figure 4 This is the result of selecting residues within 15 angstroms from the center of the pocket in an embodiment of the present invention; Figure 5 It is the result of the double mutation site of the embodiment of the present invention; Figure 6 is the clustering result of the embodiment of the present invention; Figure 7 is the prediction result of step S51 of the embodiment of the present invention; Figure 8 are all mutation recommendation results of the embodiment of the present invention; Fig. 9 It is a wet experiment to verify the activity of the mutant in the embodiment of the present invention; Fig.10 It is a schematic diagram of the structure of a device for recommending mutable sites based on small sample experimental data of the present invention; DETAILED DESCRIPTION In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is further described in detail below through the accompanying drawings and embodiments. However, it should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the scope of the present invention. In addition, in the following description, the description of known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0020] like Figure 1 As shown, the present invention provides a method for recommending mutable sites based on small sample experimental data, comprising the following steps: S1. Obtaining small sample experimental data, wherein the small sample experimental data includes sequences and data of wild enzymes and mutants, obtaining an optimal mutant sequence according to the data type of the mutant; predicting a mutant structure according to the optimal mutant sequence; and predicting a substrate binding pocket according to the mutant structure; S2. Predicting single mutation sites based on the protein structure model to obtain prediction results, and selecting residues whose distance from the center of the substrate binding pocket in step S1 is less than a threshold value one as recommended single mutation sites in the prediction results; S3, based on the small sample experimental data of step S1, training the neural network model through the protein large language model; S4. According to the data type of the mutant, the sites corresponding to the mutant sequences that are better than the wild enzyme data are selected in the small sample experimental data, as well as the sites whose distance from the center of the substrate binding pocket in step S1 is less than the threshold 2, to obtain a site set, and each site in the site set is mutated to 19 other amino acids, and two sites are combined to construct a double mutation set, and the neural network model of step S3 is used to predict the mutation results, and the predicted results are sorted, and the recommended double mutation sites are obtained according to the sorting results; S5. According to the data type of the mutant, the site corresponding to the mutant sequence that is better than the wild enzyme data is selected in the small sample experimental data, and the coordinates of the site are obtained in the mutant structure of step S1 for cluster analysis to obtain several clusters; S51, selecting one site from each cluster and combining it with other clusters to construct multiple mutations; and using the neural network model of step S3 to predict the mutation results of the multiple mutations, sorting them according to the predicted results, and obtaining recommended multiple mutation sites according to the sorting results; S52. From each cluster, mutations whose mutation results are at least 1.5 times better than those of the wild enzyme are selected according to the sorting of mutation results. The mutations selected by each cluster are combined with the mutations selected by other clusters as recommended multiple mutation sites.

[0021] In a feasible embodiment, the small sample experimental data is provided by biological experimenters or obtained from a public mutation data set.

[0022] In a feasible embodiment, the data type includes enzyme activity data, ligand binding force data or temperature tolerance data; the method for obtaining the optimal mutant sequence according to the data type of the mutant in step S1 is as follows: according to the data type of the mutant, the data is sorted from best to worst, and the sequence of the mutant with the best data is selected.

[0023] In a feasible embodiment, in step S1, AlphaFold, chai_lab or helixfold3 is used to predict the protein structure corresponding to the optimal mutant sequence to obtain the mutant structure.

[0024] In a feasible embodiment, in step S1, p2rank is used to predict the substrate binding pocket of the mutant structure.

[0025] In a feasible embodiment, in step S2, the threshold value is 15 angstroms.

[0026] In a feasible embodiment, step S3 specifically includes the following steps: S31, using ESM or ProtTrans protein language model; S32, using a large protein language model to convert sequences in small sample experimental data into numbers as training features X, using mutation results as Y, training a neural network model, and calculating the correlation between mutation results in the test set and predicted mutation results; S33, obtaining homologous sequences in small sample experimental data to fine-tune the protein large language model, training the neural network model and calculating the correlation between the mutation results in the test set and the predicted mutation results; S34. Compare the correlation results of S32 and S33, and select the neural network model with the largest correlation as the final neural network model.

[0027] In a feasible embodiment, the threshold value in step S4 is 15 angstroms, the mutation results are sorted from large to small, and topN is selected as the recommended double mutation site according to the sorting results.

[0028] In a feasible embodiment, in step S51, the mutation results are sorted from large to small, and topN is selected as the recommended multiple mutation sites according to the sorting results.

[0029] Example: Step 1: The biological experimenter provides the mutation data of an enzyme or obtains the mutation data of an enzyme from a public mutation dataset (proteingym: https: / / proteingym.org / ); Obtain the optimal mutant sequence from the data type of a certain enzyme mutant (if it is enzyme activity, select the sequence with the largest enzyme activity value; if it is binding ability, select the sequence with the largest binding ability; if it is temperature tolerance, select the sequence with the highest temperature); Predict mutant structures from optimal mutant sequences; predict substrate binding pockets from mutant structures; The tools that can predict protein structure include AlphaFold (https: / / golgi.sandbox.google.com / ), chai_lab (https: / / lab.chaidiscovery.com / dashboard); helixfold3 (https: / / paddlehelix.baidu.com / app / all / helixfold3 / forecast). The present invention selects AlphaFold to predict the structure; the present invention uses p2rank (https: / / github.com / rdk / p2rank) to predict pockets. The pocket information is as follows: Figure 2 Shown Since the results of this experiment are enzyme activity data, the selected sites below refer to sites with good enzyme activity; Step 2: Predict single mutation sites based on the protein structure model, and select residues that are less than 15 angstroms away from the center of the substrate binding pocket predicted in S1 as recommended single mutation sites The present invention uses a structural model (https: / / github.com / drorlab / gvp-pytorch) to predict single mutation sites. The prediction results are as follows: like Figure 3 As shown, the structural model predicts mutation sites (367 in total); like Figure 4 As shown, the results for residues within 15 angstroms from the center of the pocket were selected; Step 3: Use the protein language model to train the neural network based on small sample experimental data 1: Protein large language model The present invention uses ESM, specifically esm2_t33_650M_UR50D 2: Fine-tune the homologous sequences of the large language model. Use Jackhmmer to obtain 11,297 homologous sequences from the UniRef90 database. The execution command is: jackhmmer -o / dev / null -A nit.sto \ --noali --F1 0.0005 --F2 5e-05 --F3 5e-07 --incE 0.0001 -E0.0001 --cpu 25 \ -N 1 uniref90.fasta 3: The neural network uses gaussian_process of the sklearn package. The Gaussian kernel is a composite kernel, specifically WhiteKernel and RationalQuadratic 4: Correlation calculation uses the pearsonr method of the scipy package No fine-tuning of the protein language model Model score 0.37; correlation between test set and experimental results 0.52 Fine-tuning the protein language model Model score 0.48; correlation between test set and experimental results 0.69 Therefore, we chose to use the fine-tuned protein large language model to train the Gaussian model; Step 4: Select sites with improved performance compared to the wild enzyme and sites within 15 angstroms from the center of the substrate binding pocket from the mutation data and mutate each site to 19 different amino acids. Combine them in pairs and use the neural network model in step 3 to predict the mutation results. Sort the predicted mutation results from large to small and select the top 5 as the recommended double mutation sites.

[0030] like Figure 5 As shown, there are 267,501 double mutation combinations in total, and the top 5 were selected by sorting and removing duplicate sites; Step 5: Select sites with improved performance compared to the wild-type enzyme from the mutation data, and obtain the coordinates of these sites from the predicted structure in S1 for cluster analysis, e.g. Figure 6 As shown, the structural clustering results; Step 5 (1) Select one site from each cluster and combine it with other clusters to construct multiple mutations. Use the neural network model in step 3 to predict the mutation results of the multiple mutants. Select the top 5 as the recommended multiple mutation sites according to the predicted mutation results from large to small. Model prediction, sorting and selection of multiple mutation combinations such as Figure 7 As shown; Step 5 (2): Based on the clustering results, select mutations from each cluster whose mutation results are at least 1.5 times better than those of the wild type according to the mutation results sorting. The mutations selected by each cluster are combined with the mutations selected by other clusters as recommended multi-mutation sites.

[0031] Clustering results and experimental data clust1 = ['K320S-1.10'] clust2=['V199H-1.20','L194M-1.08','A201Y-2.60','Q207N-1.02'] clust3=['K97N-1.18','D40G-1.05','I117L-1.12'] clust4=['M287L-1.79','A286T-1.23','A288H-2'] clust5=['A179C-2.09','A176C-2.13','M180Q-2.51'] The mutations in each cluster that are at least 1.5 times more likely to be effective than those in the wild type are selected and combined as the recommended multiple mutation sites: K320S_A201Y_K97N_A288H_M180Q A201Y_K97N_A288H_M180Q A201Q_K97N_A288H_M180Q A201Y_K97N_A288H_A179C A201Y_K97N_M287L_M180Q All mutations recommended in the end are as follows Figure 8 As shown; Next, wet experiments were performed to verify the activity of the recommended mutable sites.

[0032] Wet experiment to verify the activity of mutants Fig. 9 As shown; WT represents wild-type protease; at is the abbreviation of enzyme, and L60I represents the mutation of the 60th position of the protease from amino acid L to amino acid I [numbering starts from 1]; In the field of enzyme modification, although traditional methods such as directed evolution and rational design have made significant progress, they still face many challenges in optimizing enzyme performance. In particular, when relying solely on small sample experimental data to recommend mutation sites, the relationship between the complex structure and function of the enzyme often makes the mutation effect difficult to predict. Some mutations may lead to significant changes in global performance, and these changes are difficult to capture in small-scale data sets. In addition, the mutation effects of enzymes often have synergistic or antagonistic effects, and traditional methods have difficulty evaluating the impact of these interactions on overall activity or stability. This makes it further difficult to accurately optimize mutation sites when data is limited; in order to overcome these limitations, the present invention proposes an innovative method for recommending mutable sites based on small sample experimental data. Compared with traditional methods, this method exhibits several advantages: Small sample adaptability: Traditional methods often require a large amount of experimental data to capture the complex relationship between enzyme function and structure, while the method of the present invention can also operate effectively under small sample data conditions. Through the powerful generalization ability of the large language model, the structural and functional information in the biological molecules can be captured, thereby effectively encoding the enzyme sequence for more accurate inference.

[0033] Efficient mutation screening: The method of the present invention can quantify the effect of mutations on enzyme performance through neural networks, significantly improving the efficiency of mutation site screening. Compared with traditional experimental exploration, the method of the present invention can propose more promising mutation combinations in a short period of time, thereby reducing the blindness of the experiment.

[0034] Rapid experimental verification: Through the predictive ability of the model, the present invention increased the activity of the target enzyme by more than 6 times within 1 week of calculation time and 2 weeks of experimental time. This rapid feedback capability makes the method of the present invention not only have theoretical advantages, but also show excellent results in actual experimental verification.

[0035] An embodiment of the device for recommending mutable sites based on small sample experimental data of the present invention can be applied to any device with data processing capabilities, and the device with data processing capabilities can be a device or apparatus such as a computer. The device embodiment can be implemented through software, or through hardware or a combination of software and hardware. Taking software implementation as an example, as a device in a logical sense, it is formed by the processor of any device with data processing capabilities in which it is located reading the corresponding computer program instructions in the non-volatile memory into the internal memory for execution. From a hardware perspective, if Fig.10As shown, it is a hardware structure diagram of any device with data processing capability where a device for recommending mutable sites based on small sample experimental data of the present invention is located, except Fig.10 In addition to the processor, memory, network interface, and non-volatile memory shown, any device with data processing capabilities in the embodiment of the device may also include other hardware according to the actual function of the device with data processing capabilities, which will not be described in detail. The implementation process of the functions and effects of each unit in the above device is specifically detailed in the implementation process of the corresponding steps in the above method, which will not be described in detail here.

[0036] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of the present invention. Ordinary technicians in this field can understand and implement it without paying creative work.

[0037] An embodiment of the present invention further provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, a recommendation device for recommending mutable sites based on small sample experimental data in the above embodiment is implemented.

[0038] The computer-readable storage medium may be an internal storage unit of any device with data processing capability described in any of the aforementioned embodiments, such as a hard disk or a memory. The computer-readable storage medium may also be an external storage device of any device with data processing capability, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, a flash card (Flash Card), etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit and an external storage device of any device with data processing capability. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capability, and may also be used to temporarily store data that has been output or is to be output.

[0039] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modification, equivalent substitution or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for recommending mutable sites based on small sample experimental data, characterized in that: The steps include: S1. Obtaining small sample experimental data, wherein the small sample experimental data includes sequences and data of wild enzymes and mutants, obtaining an optimal mutant sequence according to the data type of the mutant; predicting a mutant structure according to the optimal mutant sequence; and predicting a substrate binding pocket according to the mutant structure; S2. Predicting single mutation sites based on the protein structure model to obtain prediction results, and selecting residues whose distance from the center of the substrate binding pocket in step S1 is less than a threshold value one as recommended single mutation sites in the prediction results; S3, based on the small sample experimental data of step S1, training the neural network model through the protein large language model; S4. According to the data type of the mutant, the sites corresponding to the mutant sequences that are better than the wild enzyme data are selected in the small sample experimental data, as well as the sites whose distance from the center of the substrate binding pocket in step S1 is less than the threshold 2, to obtain a site set, and each site in the site set is mutated to 19 other amino acids, and two sites are combined to construct a double mutation set, and the neural network model of step S3 is used to predict the mutation results, and the predicted results are sorted, and the recommended double mutation sites are obtained according to the sorting results; S5. According to the data type of the mutant, the site corresponding to the mutant sequence that is better than the wild enzyme data is selected in the small sample experimental data, and the coordinates of the site are obtained in the mutant structure of step S1 for cluster analysis to obtain several clusters; S51, selecting one site from each cluster and combining it with other clusters to construct multiple mutations; and using the neural network model of step S3 to predict the mutation results of the multiple mutations, sorting them according to the predicted results, and obtaining recommended multiple mutation sites according to the sorting results; S52. From each cluster, mutations whose mutation results are at least 1.5 times better than those of the wild enzyme are selected according to the sorting of mutation results. The mutations selected by each cluster are combined with the mutations selected by other clusters as recommended multiple mutation sites.

2. A method for recommending mutable sites based on small sample experimental data as claimed in claim 1, characterized in that: The small sample experimental data is provided by biological experimenters or obtained from a public mutation data set.

3. The method for recommending mutable sites based on small sample experimental data according to claim 1, characterized in that: The data types include enzyme activity data, ligand binding data or temperature tolerance data; the method for obtaining the optimal mutant sequence according to the data type of the mutant in step S1 is as follows: according to the data type of the mutant, the data is sorted from best to worst, and the sequence of the mutant with the best data is selected.

4. The method for recommending mutable sites based on small sample experimental data according to claim 1, characterized in that: In step S1, AlphaFold, chai_lab or helixfold3 is used to predict the protein structure corresponding to the optimal mutant sequence to obtain the mutant structure.

5. The method for recommending mutable sites based on small sample experimental data as claimed in claim 1, characterized in that: In step S1, p2rank was used to predict the substrate binding pocket of the mutant structure.

6. The method for recommending mutable sites based on small sample experimental data as claimed in claim 1, characterized in that: In step S2, the threshold value is set to 15 angstroms.

7. The method for recommending mutable sites based on small sample experimental data as claimed in claim 1, characterized in that: Step S3 specifically includes the following steps: S31, using ESM or ProtTrans protein language model; S32, using a large protein language model to convert sequences in small sample experimental data into numbers as training features X, using mutation results as Y, training a neural network model, and calculating the correlation between mutation results in the test set and predicted mutation results; S33, obtaining homologous sequences in small sample experimental data to fine-tune the protein large language model, training the neural network model and calculating the correlation between the mutation results in the test set and the predicted mutation results; S34. Compare the correlation results of S32 and S33, and select the neural network model with the largest correlation as the final neural network model.

8. The method for recommending mutable sites based on small sample experimental data as claimed in claim 1, characterized in that: In step S4, the threshold value is 15 angstroms, the mutation results are sorted from large to small, and topN is selected as the recommended double mutation site according to the sorting results.

9. The method for recommending mutable sites based on small sample experimental data as claimed in claim 1, characterized in that: In step S51, the mutation results are sorted from large to small, and topN is selected as the recommended multiple mutation sites according to the sorting results.

10. A device for recommending mutable sites based on small sample experimental data, characterized in that: It comprises a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, it is used to implement a method for recommending mutable sites based on small sample experimental data as described in any one of claims 1-9.

Images