Method for estimating mutation introduction efficiency by genome edit tool

Abstract

To estimate mutation introduction efficiency with higher accuracy than before.SOLUTION: This method for estimating mutation introduction efficiency by the genome edit tool comprises: a virtual sequence generation step of generating a virtual sequence obtainable through mutation introduction from comparison sequence information in a target genome region of a cell to receive mutation and recognition sequence information recognized by the genome edit tool; a model generation step of repeating, by using a combination of virtual sequences varied from one operation to another, a step of generating, by using parts of the virtual sequences, regression models capable of estimating types and frequencies of mutations included in sequence analysis information in a target genome region of a group of cells into which the mutations have been introduced, and calculating accuracy evaluation indexes and completion degree evaluation indexes of the regression models; and an estimation step of estimating mutation introduction efficiency by using a regression model higher in evaluation of accuracy evaluation index and completion degree evaluation index than prescribed values.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a method for estimating the efficiency of mutation introduction using a genome editing tool. [Background technology]

[0002] Genome editing is a technique for introducing mutations such as deletions, substitutions, or insertions into the DNA sequence of any target gene using a genome editing tool capable of recognizing the DNA sequence of a target genome region of a cell and cleaving the DNA sequence. Known genome editing tools include, for example, zinc finger nuclease (ZFN), transcription activator-like effector nuclease (TALE nuclease), clustered regularly interspaced short palindromic repeats (CRISPR) / Cas protein (CRISPR-associated protein), etc.

[0003] Introduction of mutations using genome editing tools is generally performed on a cell population. In this case, the same mutation is not usually introduced uniformly into the genomic DNA of all the target cells. Since the type of mutation introduced varies from cell to cell, the ratio of cells into which the desired type of mutation has been introduced, in other words, the efficiency of introduction of mutations into the genomic DNA of a cell population, is an important indicator for considering the optimal conditions for genome editing.

[0004] For example, Non-Patent Document 1 proposes a method for estimating the efficiency of mutation introduction using a genome editing tool. [Prior art documents] [Non-patent literature]

[0005] [Non-Patent Document 1] David Conant et al., Inference of CRISPR Edits from Sanger Trace Data, The CRISPR Journal, 5(1) 123-130, 2022 Summary of the Invention [Problem to be solved by the invention]

[0006] Here, the method described in Non-Patent Document 1 does not take into account the possibility of deletions of more than 40 base pairs, and it is possible that the expected accuracy may not be achieved depending on the genome editing conditions, leaving room for improvement.

[0007] An object of one embodiment of the present invention is to provide a method for estimation that can estimate the efficiency of mutation introduction with higher accuracy than conventional methods. [Means for solving the problem]

[0008] In order to solve the above-mentioned problems, an estimation method according to one embodiment of the present invention is a method for estimating the efficiency of mutation introduction by a genome editing tool, and includes: a virtual sequence generation step of generating, for each mutation, a virtual sequence including the mutation that may occur in the target genome region by the introduction of the mutation, from control sequence information in a target genome region where the mutation is to be introduced of a cell to be subjected to the mutation and recognition sequence information of at least one site in the target genome region recognized by the genome editing tool; a model generation step of generating a regression model capable of estimating the type and frequency of the mutation included in sequence analysis information in the target genome region of nucleic acid obtained from the population of cells in which the mutation was introduced, using a combination of the virtual sequences that is a part of the generated virtual sequences, and calculating an accuracy evaluation index and a completeness evaluation index of the regression model, the steps of repeating this step a predetermined number of times using a different combination each time; and an estimation step of estimating the efficiency of mutation introduction using the regression model whose evaluation indicated by the completeness evaluation index was higher than a predetermined threshold among the regression models whose evaluation indicated by the accuracy evaluation index was higher than a predetermined threshold among the multiple regression models generated.

[0009] In order to solve the above-mentioned problems, an estimation system according to one embodiment of the present invention is an estimation system for estimating the efficiency of mutation introduction by a genome editing tool, and includes: a virtual sequence generation unit that generates, for each mutation, a virtual sequence including the mutation that may occur in the target genome region by the introduction of the mutation, from control sequence information in a target genome region where the mutation is to be introduced of a cell to be subjected to the mutation and recognition sequence information of at least one location in the target genome region recognized by the genome editing tool; a model generation unit that generates a regression model capable of estimating the type and frequency of the mutation included in sequence analysis information in the target genome region of nucleic acid obtained from the population of cells in which the mutation was introduced, using a combination of the virtual sequences that is a part of the generated virtual sequences, and repeats a process of calculating an accuracy evaluation index and a completeness evaluation index of the regression model a predetermined number of times using a different combination each time; and an estimation unit that estimates the efficiency of mutation introduction using a regression model whose evaluation indicated by the completeness evaluation index was higher than a predetermined threshold among the multiple regression models generated, among the regression models whose evaluation indicated by the accuracy evaluation index was higher than a predetermined threshold.

[0010] The estimation system according to each aspect of the present invention may be realized by a computer. In this case, the control program for the estimation system, which causes a computer to operate as each part (software element) of the estimation system, and the computer-readable recording medium on which the control program is recorded, also fall within the scope of the present invention. Effect of the Invention

[0011] According to one aspect of the present invention, it is possible to provide a method for estimating the efficiency of mutation introduction with higher accuracy than conventional methods. [Brief description of the drawings]

[0012] [Figure 1] FIG. 1 is a block diagram showing an example of a schematic configuration of an estimation system according to a first embodiment. [Diagram 2] 4 is a flowchart showing an example of a flow of processing executed by the estimation device according to the first embodiment. [Diagram 3] FIG. 11 is a diagram illustrating an example of a schematic configuration of an estimation system according to a second embodiment. [Figure 4] FIG. 11 is a block diagram showing an example of a schematic configuration of an estimation device according to a second embodiment. [Diagram 5] FIG. 1 is a box plot showing the coefficient of determination and the BIC value of the regression models obtained in the conventional example and the reference example. [Figure 6] FIG. 2 is a box plot showing the coefficient of determination and the BIC value of the regression models obtained in the conventional example, the reference example, and the example of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] [Embodiment 1] Hereinafter, one embodiment of the present invention will be described in detail.

[0014] (Features of Estimation System 100) The estimation system 100 according to the first embodiment of the present invention is a system that estimates the efficiency of introducing a mutation using a genome editing tool.

[0015] A genome editing tool is a tool that can specifically bind to a specific genome region and introduce a mutation such as a deletion, substitution, or insertion by cutting the DNA sequence of the genome region. When a mutation is introduced into a genome region using a genome editing tool, it is not always possible to reliably introduce the desired type of mutation. When a mutation is introduced into a cell population containing multiple cells using a genome editing tool, the type of mutation introduced into each cell is usually different.

[0016] The ratio of cells into which a desired type of mutation has been introduced, in other words, the efficiency of introducing a mutation into the genomic DNA of a cell population, can be used as an index showing the suitability of the conditions for introducing a mutation using a genome editing tool, for example. Therefore, if the accuracy of estimating the efficiency of introducing such a mutation can be improved, it becomes possible to efficiently find the optimal conditions for genome editing, and it becomes easier to obtain cells into which a desired mutation has been introduced.

[0017] The genome editing tool may be a tool composed of a combination of a DNA binding tool and a DNA cleavage tool such as a nuclease. Examples of the DNA binding tool include a zinc finger, a polypeptide including a TALE (Transcription Activator-Like Effector) or a pentatricopeptide repeat (PPR), and a guide nucleic acid that forms a complex with a wild-type or modified Cas protein (CRISPR-associated protein).

[0018] Genome editing tools can be fusion polypeptides of these DNA binding tools and nucleases such as Cas proteins, or polypeptide-nucleic acid complexes such as RNPs (ribonucleoproteins). Such genome editing tools introduce mutations by utilizing repair mechanisms such as non-homologous end joining or homologous recombination repair on DNA sequences cut by DNA cutting tools. In such repair mechanisms, the DNA sequences after repair are not necessarily the same, so there is variation in the types of mutations introduced in each cell.

[0019] On the other hand, it is known that the probability that the repaired DNA sequence will have the desired mutation can be increased by setting conditions such as the design of the DNA binding tool, i.e., the efficiency of introducing mutations can be increased by setting the conditions. This is because the type and frequency of mutations that occur during the repair process differ depending on the binding specificity of the DNA binding tool to the recognition sequence, the cleavage site of the DNA sequence, and the sequence pattern within the recognition sequence.

[0020] Therefore, it is important to consider conditions when using a genome editing tool, and if the accuracy of estimating the efficiency of mutation introduction can be improved, the condition consideration can be performed efficiently. The estimation system 100 according to this embodiment can estimate the efficiency of mutation introduction with higher accuracy than the conventional method described in Non-Patent Document 1 by executing the following processes (a) to (c).

[0021] (a) A virtual sequence including a mutation that may occur in the target genome region by introducing a mutation is generated for each mutation from control sequence information in the target genome region where a mutation is to be introduced of a cell to be subjected to the mutation and information of at least one recognition sequence in the target genome region recognized by the genome editing tool.

[0022] (b) Using a combination of virtual sequences that are part of the generated virtual sequences, a regression model capable of estimating the type and frequency of mutations contained in sequence analysis information in the target genomic region of nucleic acid obtained from a population of cells into which mutations have been introduced is generated, and the process of calculating an accuracy evaluation index and a completeness evaluation index of the regression model is repeated a predetermined number of times using a different combination each time.

[0023] (c) Among the multiple regression models generated, the efficiency of mutation introduction is estimated using a regression model whose accuracy evaluation index is higher than a predetermined threshold and whose completeness evaluation index is higher than a predetermined threshold.

[0024] (Schematic configuration of estimation system 100) Hereinafter, a schematic configuration of the estimation system 100 will be described with reference to Fig. 1. Fig. 1 is a block diagram showing an example of a schematic configuration of the estimation system 100.

[0025] As shown in Fig. 1, the estimation system 100 may include an estimation device 1 and a display device 4. Fig. 1 shows the estimation system 100 including one each of the estimation device 1 and the display device 4. However, the configuration of the estimation system 100 is not limited thereto. For example, the number of display devices 4 included in the estimation system 100 may be zero or more.

[0026] In the estimation system 100, the estimation device 1 and the display device 4 are connected to each other so that they can communicate with each other. The estimation device 1 and the display device 4 may be directly connected to each other by wire or wirelessly, or may be connected to each other via a communication network. The type of the communication network is not limited, and may be a local area network (LAN) or the Internet.

[0027] The estimation device 1 is a device that estimates the efficiency of mutation introduction by a genome editing tool using sequence analysis information of a cell population that has undergone genome editing. The generated estimation result may be transmitted from the estimation device 1 to the display device 4.

[0028] The display device 4 may typically be a computer, a smartphone, a tablet terminal, or the like used by a user who uses the estimation system 100. Note that FIG. 1 illustrates the estimation system 100 in which the display device 4 is separate from the estimation device 1. However, the configuration of the estimation system 100 is not limited to this. For example, the display device 4 may be a device integrated with the estimation device 1, and in this case, the display device 4 may be a display unit (such as a display) included in the estimation device 1.

[0029] (Configuration of Estimation Device 1) Next, a description will be given of the configuration of the estimation device 1. The estimation device 1 includes a control unit 10, a storage unit 20, and an input unit 30.

[0030] In one example, the control unit 10 may be a CPU (Central Processing Unit). The control unit 10 reads a control program, which is software stored in the storage unit 20, and expands it in a memory such as a RAM (Random Access Memory) to execute various functions. For the sake of simplicity, the storage unit 20 shown in FIG. 1 does not show stored data such as the control program.

[0031] The storage unit 20 may be, for example, a hard disk drive (HDD) or a solid state drive (SSD). The storage unit 20 may store information transmitted from the control unit 10, and the stored information may be read by the control unit 10.

[0032] The input unit 30 is configured to input information to the estimation device 1 by a user. The input unit 30 may be, for example, at least one of a keyboard, a mouse, or a touchpad. The input unit 30 may also be an interface that accepts input of information from an external device such as a sequencer or an external storage device such as a USB memory.

[0033] The control unit 10 includes an array information receiving unit 11, a virtual array generating unit 12, a model generating unit 13, and an estimating unit .

[0034] The sequence information receiving unit 11 receives input sequence information input using the input unit 30. The sequence information receiving unit 11 may store the received input sequence information in the storage unit 20.

[0035] The input sequence information received by the sequence information receiving unit 11 includes reference sequence information, recognition sequence information, template sequence information, and sequence analysis information, and may be at least one of these. Details of the reference sequence information, recognition sequence information, and template sequence information will be described later in the description of the virtual sequence generating unit 12. Details of the sequence analysis information will be described later in the description of the model generating unit 13.

[0036] The virtual sequence generating unit 12 is a process for generating a plurality of virtual sequences. The virtual sequence is information on a virtual sequence including possible mutations in a target genome region of a cell to be subjected to mutation introduction, where a mutation is introduced by a genome editing tool. Usually, various types of mutations can occur in a target genome region, so that there are multiple types of mutations that can occur. The estimation device 1 according to this embodiment can estimate the type and frequency of a mutation that matches a purpose among various mutations that have occurred in a target genome region in a population of cells as the efficiency of mutation introduction.

[0037] The types of mutations that can occur and are used to generate a virtual sequence may include at least any of the following (1) to (3).

[0038] (1) When the mutation is a deletion: The position of the deletion in the target genome region and the number of deleted bases each become an individual type of mutation. The number of deleted bases is not limited, but may be, for example, 1 base pair or more and 100 base pairs or less. In addition, the virtual sequence generation unit 12 may include a deletion of a length exceeding 40 base pairs, for example, a deletion of more than 40 base pairs and 100 base pairs or less, as at least one of the possible mutations. The virtual sequence generation unit 12 may include a deletion of a number of bases exceeding 100 base pairs in the type of mutation.

[0039] (2) When the mutation is a substitution: In the target genome region, the position where the substitution occurs, the type of the substituted base, and the number of substitutions each result in an individual type of mutation. The type of the substituted base may be, for example, any of A (adenine), T (thymine), G (guanine), or C (cytosine) other than the base before substitution, or may be a modified base of these bases, or may be other so-called special bases.

[0040] (3) When the mutation is an insertion: the position where the base is inserted in the target genome region, the number of inserted bases, and the base sequence each result in a different type of mutation.

[0041] The virtual sequence generation unit 12 uses the control sequence information and the recognition sequence information to estimate which type of mutation may occur in the target genome region, and generates a virtual sequence. The control sequence information is sequence information of the target genome region into which a mutation is to be introduced, of a cell into which a mutation is to be introduced. The recognition sequence information is sequence information of at least one site in the target genome region that is recognized by the genome editing tool.

[0042] The control sequence information is information showing the base sequence of a genomic region including a site recognized by a genome editing tool in the genomic DNA of a cell to which a mutation is to be introduced. The control sequence information may be text-format information showing the base sequence, or may be sequence analysis information including a waveform obtained by analyzing the base sequence of the target genomic region by Sanger sequencing.

[0043] The control sequence information preferably includes analysis information obtained by the same analysis method (sequencing method) as the sequence analysis information used by the model generation unit 13 described below. With such a configuration, the regression model generated by the model generation unit 13, including information such as waveform, can easily estimate with high accuracy the type and frequency of mutations actually introduced into the target genome region, and therefore the estimation accuracy by the estimation unit 14 can be easily improved.

[0044] The recognition sequence information is information on a sequence in a target genome region recognized by a DNA binding tool in a genome editing tool. The recognition sequence information may be, for example, text information showing any one of the following base sequences (i) to (iv).

[0045] (i) When the DNA-binding tool is a zinc finger, the base sequence recognized by a domain having a zinc finger protein motif.

[0046] (ii) When the DNA binding tool is a TALE, the base sequence recognized by the region to which the TAL module (Transcription Activator-Like Module) is bound.

[0047] (iii) When the DNA binding tool is a PPR, the base sequence recognized by a region of consecutive PPR motifs.

[0048] (iv) When the DNA-binding tool is a wild-type or mutant CRISPR / Cas protein, a nucleotide sequence complementary to a guide RNA (gRNA) complexed with the Cas protein and a protospacer adjacent motif (PAM) sequence recognized by the Cas protein.

[0049] In addition, information showing a sequence complementary to the base sequences shown in (i) to (iv) above may be regarded as information showing a sequence recognized by a DNA binding tool, and may be regarded as recognition sequence information. For example, in (iv) above, information on the base sequence of the guide RNA itself may be regarded as recognition sequence information.

[0050] The virtual sequence generating unit 12 estimates the position in the reference sequence information where the DNA cleavage tool will cleave, for example, based on the reference sequence information and the recognition sequence information. The virtual sequence generating unit 12 may then generate a virtual sequence for each type of deletion, substitution, or insertion that may occur at or around the cleavage position.

[0051] In addition, the virtual sequence generating unit 12 may generate a virtual sequence by using template sequence information for introducing a mutation in addition to the control sequence information and the recognition sequence information. The template sequence information is information including the base sequence of an insert to be incorporated into genomic DNA by homologous recombination, for example, when a design is made to use a homologous recombination repair mechanism for genome editing. The insert may be designed for the purpose of inserting, deleting, or substituting a base sequence using a homologous recombination repair mechanism or the like.

[0052] By using such template sequence information, the virtual sequence generation unit 12 can generate a virtual sequence by limiting the number of types of mutations that can occur, which enables the estimation device 1 to estimate the mutation efficiency with high accuracy while reducing the processing load.

[0053] The virtual sequence generation unit 12 may generate virtual sequences for each of all types of mutations that are theoretically possible to occur in the target genome region, assuming that each type of mutation may occur. Alternatively, the virtual sequence generation unit 12 may exclude types of mutations that are determined to have a low probability of occurring from among the types of mutations that are theoretically possible to occur in the target genome region, and generate virtual sequences for each of the remaining types of mutations.

[0054] For example, when the virtual sequence generating unit 12 uses template sequence information, the virtual sequence generating unit 12 may exclude insertions of base sequences not included in the template sequence information from the types of possible mutations, since such insertions are unlikely to occur. In addition, the virtual sequence generating unit 12 may exclude from the types of possible mutations types that are empirically considered to be unlikely to occur based on the type of genome editing tool and the DNA sequence of the target genome region. In this way, rules for excluding some types of mutations from the types of possible mutations may be stored, for example, in the storage unit 20.

[0055] The virtual sequence generating unit 12 may receive the control sequence information, the recognition sequence information, and the template sequence information through the sequence information receiving unit 11, or may read at least any of them from the storage unit 20. For example, in a condition study using a genome editing tool, various conditions are tried for the same target genome region. At this time, the recognition sequence information and the template sequence information are input from the input unit 30 each time a condition is studied, while the control sequence information may be repeatedly read from the information previously stored in the storage unit 20.

[0056] The length of the virtual sequence generated by the virtual sequence generation unit 12 is not particularly limited, but may be, for example, 1000 base pairs or less, preferably 500 base pairs or less, and more preferably 250 base pairs or less so as not to cause an excessive processing load. Also, from the viewpoint of estimating the mutation efficiency with high accuracy, the length may be 10 base pairs or more, 50 base pairs or more, or 100 base pairs or more. The virtual sequence generation unit 12 may generate virtual sequences with different lengths depending on the type of mutation that may occur, or may make all virtual sequences the same length.

[0057] The model generation unit 13 repeatedly generates regression models for estimating the efficiency of mutation introduction by the genome editing tool. Each regression model generated by the model generation unit 13 can estimate the type and frequency of mutations contained in sequence analysis information in a target genome region of nucleic acids obtained from a population of cells into which mutations have been introduced.

[0058] The sequence analysis information is information included in the input sequence information received by the sequence information receiving unit 11. The sequence analysis information includes, for example, information such as waveforms obtained by analyzing a target genome region of genomic DNA, which is a nucleic acid obtained from a population of cells into which a mutation has been introduced, by a method such as Sanger sequencing. Here, a population of cells into which a mutation has been introduced may have different types of mutations introduced into each cell. Therefore, the sequence analysis information is usually information obtained by analyzing multiple base sequences into which different mutations have been introduced in a redundant manner. For example, in the case of Sanger sequencing, the sequence analysis information includes multiple waveform information in a redundant state.

[0059] The model generating unit 13 generates a regression model that estimates the type and frequency of mutations included in the sequence analysis information, using a combination of virtual sequences that are part of the multiple virtual sequences generated by the virtual sequence generating unit 12. The generated regression model may be, for example, a regression model that is capable of separating and analyzing multiple waveforms included in the sequence analysis information and estimating the type and frequency of mutations that have actually occurred in the genomic DNA of a population of cells into which a mutation has been introduced.

[0060] The regression model generated by the model generation unit 13 may be a linear regression model or a nonlinear regression model. Examples of the linear regression model include a nonnegative least squares regression model, a least squares regression model, a lasso regression model, a ridge regression model, a principal component regression model, and a Bayesian linear regression model. Examples of the nonlinear regression model include a neural network regression model, a support vector regression model, and a random forest regression model. The regression model generated by the model generation unit 13 may be a regression model that combines these regression models. If the regression model is a linear regression model, a regression coefficient may be used in estimating the frequency of mutation. Also, even if the regression model is a nonlinear regression model, an alternative index of the regression coefficient may be used in estimating the frequency of mutation. For example, if the regression model is a random forest regression model, an index representing impurity such as Gini impurity may be used as the alternative index.

[0061] The model generation unit 13 calculates an accuracy evaluation index that indicates the accuracy of the generated regression model. The accuracy evaluation index may be anything that indicates the accuracy of the regression model, such as the coefficient of determination (R 2 ), least square error, absolute mean error, cross entropy error, squared logarithmic error, and AUC (Area Under Curve). For example, when the accuracy evaluation index is the coefficient of determination, the larger the calculated value, the higher the evaluation of the accuracy of the regression model. When the accuracy evaluation index is the least square error, the level of evaluation indicated by the accuracy evaluation index is represented by the small value of the least square error. In other words, when the accuracy evaluation index is the least square error, the smaller the calculated value, the higher the evaluation of the accuracy of the regression model.

[0062] Moreover, the model generating unit 13 calculates a completeness evaluation index indicating the completeness of the generated regression model. The completeness evaluation index may be any index indicating the completeness of the regression model, and may be, for example, BIC (Bayesian Information Criterion), AIC (Akaike Information Criterion), WBIC (Widely Applicable Bayesian Information Criterion), MDL (Minimum Description Length), and Cp criterion. For example, when the completeness evaluation index is BIC, the degree of evaluation indicated by the completeness evaluation index is expressed by the smaller the BIC value. That is, when the completeness evaluation index is BIC, the smaller the calculated value (BIC value), the higher the completeness of the regression model.

[0063] The model generation unit 13 repeats the process of generating such a regression model a predetermined number of times. At this time, the model generation unit 13 uses a different combination of virtual sequences for each repetition. With this configuration, the model generation unit 13 can generate multiple regression models with different accuracy and completeness.

[0064] The model generation unit 13 may generate combinations of virtual arrays by randomly extracting them from the multiple virtual arrays generated by the virtual array generation unit 12, or may generate combinations by extracting them according to a predetermined rule. The predetermined rule is not particularly limited, and may be, for example, a rule that preferentially selects a virtual array that has not been used in the generation of a previous regression model. Such a predetermined rule for generating combinations of virtual arrays may be stored in the storage unit 20, for example.

[0065] The predetermined number of times that the model generating unit 13 repeats the generation of the regression model is not particularly limited. From the viewpoint of obtaining a regression model with excellent accuracy and completeness, the predetermined number of times may be 2 times or more, 5 times or more, 10 times or more, or 100 times or more. On the other hand, from the viewpoint of reducing the processing load of the model generating unit 13, the predetermined number of times may be 10,000 times or less, or 1,000 times or less. The predetermined number of times may be stored in the storage unit 20 as setting information, or may be input by the user from the input unit 30.

[0066] Furthermore, the model generation unit 13 may, in a predetermined case, treat a combination of virtual arrays to be used in the next or subsequent iteration as a part of the virtual array used this time. The predetermined case may be, for example, a case where the evaluation indicated by the completeness evaluation index of a regression model generated in the second or subsequent iteration of regression model generation is lower than the evaluation indicated by the completeness evaluation index of any of the regression models generated before that.

[0067] Specifically, the model generation unit 13 may set the combination of virtual arrays used this time as a population for generating combinations of virtual arrays from the next time onward. In this case, the model generation unit 13 does not use the virtual arrays not used this time among the virtual arrays generated by the virtual array generation unit 12 in generating the subsequent regression models.

[0068] According to such a configuration, when the evaluation of a regression model generated in a certain iteration is lower than any of the regression models generated previously, the model generation unit 13 limits the virtual array to be used in the next iteration to a part of the virtual array used this time. This reduces the possibility of reusing a combination similar to the combination of virtual arrays used to generate a regression model with a low evaluation. Therefore, the model generation unit 13 can increase the possibility of generating a regression model with a higher evaluation in the subsequent iterations. In addition, the more the model generation unit 13 repeats the generation of the regression model, the less the amount of virtual arrays used can be, and therefore the processing load of the model generation unit 13 can be reduced.

[0069] The estimation unit 14 estimates the efficiency of mutation introduction by selecting a single appropriate regression model from the multiple regression models generated by the model generation unit 13. Specifically, the estimation unit 14 estimates the efficiency of mutation introduction by using a regression model whose evaluation indicated by a completeness evaluation index was higher than a predetermined threshold among the multiple regression models generated and whose evaluation indicated by an accuracy evaluation index was higher than a predetermined threshold.

[0070] The process of determining that the evaluation indicated by the accuracy evaluation index is "higher than a predetermined threshold" may differ depending on the type of accuracy evaluation index calculated by the model generation unit 13. For example, if the accuracy evaluation index is an index that becomes a larger numerical value as the evaluation of the regression model becomes higher, the estimation unit 14 may determine that the accuracy of the regression model is "higher than a predetermined threshold" when the value of the accuracy evaluation index is greater than a predetermined threshold. Also, if the accuracy evaluation index is an index that becomes a smaller numerical value as the evaluation of the regression model becomes higher, the estimation unit 14 may determine that the accuracy of the regression model is "higher than a predetermined threshold" when the value of the accuracy evaluation index is smaller than a predetermined threshold.

[0071] The same applies to the evaluation indicated by the completeness evaluation index being "higher than a predetermined threshold value." For example, if the completeness evaluation index is an index that becomes a larger numerical value as the evaluation of the regression model becomes higher, the estimation unit 14 may determine that the completeness of the regression model is "higher than a predetermined threshold value" when the value of the completeness evaluation index is greater than a predetermined threshold value. Also, if the completeness evaluation index is an index that becomes a smaller numerical value as the evaluation of the regression model becomes higher, the estimation unit 14 may determine that the completeness of the regression model is "higher than a predetermined threshold value" when the value of the completeness evaluation index is less than a predetermined threshold value.

[0072] The predetermined threshold value of the evaluation indicated by the accuracy evaluation index may be set appropriately. For example, when the accuracy evaluation index is a coefficient of determination, the predetermined threshold value may be set to a value relatively close to 1, such as 0.8, and the estimation unit 14 may determine that a regression model having a coefficient of determination larger than the threshold value has a high evaluation of accuracy.

[0073] Similarly, a predetermined threshold value of the evaluation indicated by the completeness evaluation index may be appropriately set. The predetermined threshold value of the completeness evaluation index may be set to a numerical value similar to that of the accuracy evaluation index. Furthermore, a relative condition such as "the one having the largest completeness evaluation index" among the generated regression models may be set as the predetermined threshold value.

[0074] The predetermined thresholds for each of the accuracy evaluation index and the completeness evaluation index may be stored in the storage unit 20, for example, or values ​​input by the user through the input unit 30 may be used.

[0075] According to such a configuration of the estimation device 1, the generation of regression models is repeated a predetermined number of times, and the efficiency of mutation introduction can be estimated using a regression model that is evaluated excellently in terms of the accuracy evaluation index and the completeness evaluation index among the generated regression models.

[0076] Conventionally, sequence analysis information including information of overlapping waveforms obtained by Sanger sequencing could be separated into individual base sequences by a technique such as that shown in Non-Patent Document 1, and the efficiency of mutation introduction could be estimated. However, in the technique described in Non-Patent Document 1, the number of deleted bases when the type of mutation is estimated to be a deletion is set to 40 base pairs. Therefore, it was unclear whether the efficiency of mutation introduction could be accurately estimated when a deletion of more than 40 base pairs has occurred.

[0077] As shown in the following examples as a conventional example, the present inventors conducted an investigation using samples in which deletions of more than 40 base pairs were introduced by a genome editing tool, and found that the accuracy of the regression model was reduced in the technology described in Non-Patent Document 1. In addition, when the condition settings of the technology described in Non-Patent Document 1 were changed so that deletions of more than 40 base pairs were also taken into consideration, the accuracy of the regression model was improved, but the index of the completeness of the regression model was found to deteriorate.

[0078] As described above, the estimation device 1 according to this embodiment can generate a regression model with excellent accuracy and completeness in order to estimate the efficiency of mutation introduction, and can estimate the efficiency of mutation introduction using such a regression model. Furthermore, the present inventors have found that such an estimation device 1 can generate a regression model that is both highly accurate and excellent in completeness even when a virtual sequence is generated including, for example, a deletion of more than 40 base pairs, and have completed the present invention.

[0079] (Flow of processing executed by the estimation device 1) A method for estimating mutation introduction efficiency according to one embodiment of the present invention will be described with reference to Fig. 2, taking as an example the flow of processing executed by the estimation device 1. Fig. 2 is a flowchart showing an example of the flow of processing executed by the estimation device 1. Fig. 2 also shows the flow of processing executed by an estimation system 100 including the estimation device 1.

[0080] First, the sequence information receiving unit 11 receives input of input sequence information by a user (step S1). The input sequence information includes control sequence information, recognition sequence information, and sequence analysis information, and may further include template sequence information.

[0081] Next, the virtual sequence generation unit 12 uses the control sequence information and recognition sequence information, and optionally template sequence information, to generate virtual sequences including mutations that may occur in the target genome region, for each type of mutation (step S2: virtual sequence generation process).

[0082] Subsequently, the model generation unit 13 executes a model generation process from step S3 to step S8. First, the model generation unit 13 generates a combination of virtual arrays from a part of the virtual arrays generated in the virtual array generation process (step S3).

[0083] Here, the population of virtual arrays used to generate a combination of virtual arrays is referred to as a “target virtual array.” The model generation unit 13 generates a regression model multiple times, and in at least the first generation of the regression model, the target virtual array may include all of the virtual arrays generated by the virtual array generation step.

[0084] Next, the model generating unit 13 generates a regression model capable of estimating the type and frequency of mutation contained in the sequence analysis information, using a combination of the generated virtual sequences (step S4).

[0085] Then, the model generation unit 13 calculates the coefficient of determination and the BIC value of the generated regression model (step S5). Hereinafter, the accuracy evaluation index calculated by the model generation unit 13 will be described as the coefficient of determination, and the completeness evaluation index will be described as the BIC value.

[0086] Next, the model generation unit 13 judges whether the BIC value of the regression model generated this time is larger (whether the evaluation is lower) than the BIC value of any of the regression models generated previously (step S6). If the answer is Yes in step S6, the model generation unit 13 limits the target virtual arrays to be used next time to the combination of virtual arrays used this time (step S7). In other words, the virtual arrays not used this time are excluded from the generation of the regression model from the next time onwards.

[0087] On the other hand, if the answer is No in step S6, the model generation unit 13 does not make any particular changes to the target virtual sequence. For example, if the model generation unit 13 generates a regression model for the first time, there is no "regression model generated before the last time." In this case, the model generation unit 13 may determine No in step S6.

[0088] Next, the model generating unit 13 determines whether the generation of the regression model has been repeated a predetermined number of times (step S8). The predetermined number of times may be, for example, 10 times, and in this case, the number of generated regression models is 10.

[0089] Next, the estimation unit 14 executes the estimation process of steps S9 and S10. First, the estimation unit 14 selects a regression model having a coefficient of determination greater than 0.8 from among the multiple regression models generated by the model generation process. At this time, if a regression model having a coefficient of determination greater than 0.8 is not found, the estimation device 1 may return the process to step S3 and further repeat the generation of the regression model by the model generation unit 13.

[0090] Next, the estimation unit 14 selects the regression model with the smallest BIC value (the regression model with the highest degree of completion) from among the selected regression models (step S9).

[0091] Then, the estimation unit 14 estimates the efficiency of mutation introduction in the population of cells into which the mutation has been introduced, using the regression model selected in step S9 (step S10).

[0092] The estimation system 100 executes the processes described above, for example, to generate a regression model that achieves both high accuracy and excellent completeness, and can estimate the mutation introduction efficiency with high accuracy using the regression model.

[0093] [Embodiment 2] Other embodiments of the present invention are described below.

[0094] 1, the estimation system 100 includes an estimation device 1 that includes an input unit 30 that accepts input of input sequence information by a user and outputs an estimation result to a display device 4, but is not limited to this. For example, as shown in FIG 3, the estimation system 100a may include an estimation device 1a that is communicatively connected to communication devices 5a and 5b used by each user via a communication network 9.

[0095] In the estimation system 100a shown in FIG. 3, the estimation device 1a receives input sequence information from each of the communication devices 5a and 5b. Then, the estimation device 1a transmits an estimation result corresponding to the input sequence information received from the communication device 5a to the communication device 5a, and transmits an estimation result corresponding to the input sequence information received from the communication device 5b to the communication device 5b. Note that FIG. 3 illustrates the estimation system 100a including the communication devices 5a and 5b and the estimation device 1a, but is not limited thereto. In the estimation system 100a, the estimation device 1a may be capable of communicating with two or more communication devices.

[0096] (Configuration of Estimation Device 1a) The configuration of the estimation device 1a will be described with reference to Fig. 4. Fig. 4 is a functional block diagram showing a configuration example of an estimation system 100a according to one embodiment of the present invention. For convenience of explanation, the same reference numerals are given to members having the same functions as those described in the above embodiment, and the explanations thereof will not be repeated.

[0097] As shown in Fig. 4, the estimation device 1a includes a communication unit 60 that functions as a communication interface with the communication devices 5a and 5b. The array information receiving unit 11 receives input array information via the communication unit 60. The estimation unit 14 transmits an estimation result to each of the communication devices 5a and 5b via the communication unit 60. The estimation device 1a may generate a web page showing an estimation result regarding the received input array information, and provide information for accessing the web page to a user who is a sender of the input array information.

[0098] Here, the communication device 5a and the communication device 5b may be communication devices used by users who are registered in advance as users of the estimation system 100a. In this case, the storage unit 20 may store user information including information about users who are registered in advance as users of the estimation system 100a. The user information may include contact information such as an email address for each user, and may further include information about a billing destination for the usage fee of the estimation system 100a for each user.

[0099] [Software implementation example] The functions of the estimation device 1 (hereinafter referred to as the "device") can be realized by a program for causing a computer to function as the device, and a program for causing a computer to function as each control block of the device (particularly each part included in the control unit 10).

[0100] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the program. The control device and storage device execute the program to realize each function described in each embodiment.

[0101] The program may be non-transitory and may be recorded in one or more computer-readable recording media. The recording media may or may not be included in the device. In the latter case, the program may be provided to the device via any wired or wireless transmission medium.

[0102] In addition, some or all of the functions of each of the control blocks can be realized by a logic circuit. For example, an integrated circuit in which a logic circuit that functions as each of the control blocks is formed is also included in the scope of the present invention. In addition, the functions of each of the control blocks can be realized by, for example, a quantum computer.

[0103] Furthermore, each process described in each embodiment may be executed by AI (Artificial Intelligence). In this case, the AI ​​may be operated by the control device or may be operated by another device (e.g., an edge computer or a cloud server).

[0104] 〔summary〕 The estimation method according to a first aspect of the present invention is a method for estimating efficiency of mutation introduction by a genome editing tool, and includes: a virtual sequence generation step of generating, for each mutation, a virtual sequence including the mutation that may occur in the target genome region by the introduction of the mutation, from control sequence information in a target genome region where the mutation is to be introduced of a cell to be subjected to the mutation and recognition sequence information of at least one site in the target genome region recognized by the genome editing tool; a model generation step of generating a regression model capable of estimating the type and frequency of the mutation included in sequence analysis information in the target genome region of nucleic acid obtained from the population of cells into which the mutation was introduced, using a combination of the virtual sequences that is a part of the generated virtual sequences, and calculating an accuracy evaluation index and a completeness evaluation index of the regression model, the steps of repeating this step a predetermined number of times using a different combination each time; and an estimation step of estimating efficiency of mutation introduction using a regression model whose evaluation indicated by the completeness evaluation index was higher than a predetermined threshold among the multiple regression models generated, the regression models whose evaluation indicated by the accuracy evaluation index was higher than a predetermined threshold.

[0105] In the estimation method according to aspect 2 of the present invention, in the virtual sequence generation step of aspect 1, the virtual sequence may be generated by further using template sequence information for introducing the mutation in addition to the control sequence information and the recognition sequence information.

[0106] In the estimation method according to aspect 3 of the present invention, in the model generation step of aspect 1 or 2, if the evaluation indicated by the completeness evaluation index of the regression model generated in the second or subsequent iteration is lower than the evaluation indicated by the completeness evaluation index of any of the regression models generated previously, the combination of the virtual arrays used in the next or subsequent iterations may be included as part of the virtual array used this time.

[0107] In the estimation method according to aspect 4 of the present invention, in any one of aspects 1 to 3, in the virtual sequence generation step, a plurality of virtual sequences may be generated, each of which includes at least one deletion of more than 40 base pairs and not more than 100 base pairs as a possible mutation.

[0108] A fifth aspect of the present invention relates to an estimation system for estimating the efficiency of mutation introduction by a genome editing tool, and includes: a virtual sequence generation unit that generates, for each mutation, a virtual sequence including the mutation that may occur in the target genome region by the introduction of the mutation, from control sequence information in a target genome region where the mutation is to be introduced of a cell to be subjected to the mutation and recognition sequence information at at least one location in the target genome region recognized by the genome editing tool; a model generation unit that generates a regression model capable of estimating the type and frequency of the mutation included in sequence analysis information in the target genome region of nucleic acid obtained from the population of cells into which the mutation was introduced, using a combination of the virtual sequences that is a part of the generated virtual sequences, and repeats a process of calculating an accuracy evaluation index and a completeness evaluation index of the regression model a predetermined number of times using a different combination each time; and an estimation unit that estimates the efficiency of mutation introduction using a regression model whose evaluation indicated by the completeness evaluation index was higher than a predetermined threshold among the multiple regression models generated, the regression models whose evaluation indicated by the accuracy evaluation index was higher than a predetermined threshold.

[0109] A control program according to a sixth aspect of the present invention is a control program for controlling a computer that estimates the efficiency of mutation introduction by a genome editing tool, and causes the computer to execute the following steps: a virtual sequence generation step of generating, for each mutation, a virtual sequence including the mutation that may occur in the target genome region by the introduction of the mutation, from control sequence information in a target genome region where the mutation is to be introduced of a cell to be subjected to the mutation and recognition sequence information of at least one location in the target genome region recognized by the genome editing tool; a model generation step of generating a regression model capable of estimating the type and frequency of the mutation included in sequence analysis information in the target genome region of nucleic acid obtained from the population of cells into which the mutation was introduced, using a combination of the virtual sequences that is a part of the generated virtual sequences, and repeating the process of calculating an accuracy evaluation index and a completeness evaluation index of the regression model a predetermined number of times using a different combination each time; and an estimation step of estimating the efficiency of mutation introduction using the regression model whose evaluation indicated by the completeness evaluation index was higher than a predetermined threshold among the multiple regression models generated, the regression models whose evaluation indicated by the accuracy evaluation index was higher than a predetermined threshold.

[0110] A recording medium according to a seventh aspect of the present invention is a computer-readable recording medium having the control program according to the sixth aspect recorded thereon. EXAMPLES

[0111] An embodiment of the present invention will now be described with reference to FIGS.

[0112] [Introducing mutations using genome editing tools] We used the CRISPR / Cas9 system as a genome editing tool to perform genome editing on mouse embryos.

[0113] (Guide RNA selection) From the 4,221,722 guide RNAs provided by dbGuide (https: / / frederick.cancer.gov / resources / repositories / dbguide), we narrowed down the guide RNAs that target genes that do not exhibit embryonic lethality by knockout in IMPC (https: / / www.mousephenotype.org). Furthermore, we narrowed down the guide RNAs to those containing genomic regions that do not have differences within ±40 base pairs in all strains of C57BL / 6J (mm10), DBA / 2J and C57BL / 6NJ. Next, we used GGGenome (https: / / gggenome.dbcls.jp / ja / ) to obtain 3,210 guide RNA sequences as guide RNAs that do not have off-target sites of 3 base pairs or less.

[0114] Next, predictions of predicted cleavage activity and mutation patterns were performed using CRISPRscan (https: / / www.crisprscan.org) and InDelphi (https: / / github.com / maxwshen / inDelphi-model), and categorization was performed based on the results. 40 types of guide RNAs were selected from each category so that they were as equal as possible. In this example, the results are shown using a guide RNA that recognizes a locus (Pear1) that was found to be particularly prone to deletions of 60 base pairs or more due to mutation introduction, among the 40 types of guide RNAs.

[0115] (Introduction of CRISPR / Cas9 RNP into mouse embryos) For each guide RNA, PCR was performed using a forward primer with a T7 promoter sequence and each target sequence added, a common reverse primer, and a px330 plasmid to amplify DNA for guide RNA synthesis. Using this template DNA, sgRNA was synthesized by T7 in vitro transcription, and then the synthesized sgRNA was purified. The obtained synthetic sgRNA was frozen once, and then the synthetic sgRNA and Cas9 protein were associated in a buffer to form RNP, and the RNP was introduced into mouse fertilized eggs by electroporation. The electroporated fertilized eggs were cultured in KSOM-AA medium and developed into blastocysts, after which samples were taken one by one and genome extraction was performed by alkaline lysis method.

[0116] (Estimation of mutation introduction efficiency) For each of the 10 alkali-treated blastocyst samples, a region containing about ±200 base pairs centered on the mutation site was amplified by PCR. The amplification product of about 540 bp was subjected to Sanger sequencing to obtain Sanger data containing waveform information. This Sanger data and the target guide RNA sequence were input into a device that executes the conventional method ICE (see Non-Patent Document 1) and an estimation device 1 according to one embodiment of the present invention, respectively, to estimate the efficiency of mutation introduction. The Sanger data corresponds to control sequence information and sequence analysis information, and the target guide RNA sequence corresponds to recognition sequence information, respectively.

[0117] (Data Analysis) Analysis by the conventional ICE method was performed using a script and condition settings equivalent to those of the command line version of ICE (conventional example). Analysis by the estimation device 1 according to the present invention was performed under two conditions: with a loop (present invention example) and without a loop (reference example). The reference example is an example in which the condition settings for the conventional ICE method were modified from the conventional example so that deletions of more than 0 base pairs and up to 100 base pairs were considered, instead of deletions of more than 0 base pairs and up to 40 base pairs. Furthermore, in the present invention example as well, the conditions were set so that deletions of more than 0 base pairs and up to 100 base pairs were considered in the virtual sequence generation step.

[0118] To compare the results, t-tests were performed on the mean values ​​of the coefficient of determination (R^2 value) and BIC value of each transgenic embryo sample for each locus, and the results were plotted as box plots as shown in Figures 5 and 6.

[0119] [Generating and Evaluating Regression Models] Fig. 5 is a box plot showing the coefficient of determination and the BIC value of the regression models obtained in the conventional example and the reference example, and Fig. 6 is a box plot showing the coefficient of determination and the BIC value of the regression models obtained in the conventional example, the reference example, and the example of the present invention.

[0120] As shown in Figure 5, the reference example, which takes into account deletions of more than 0 base pairs and up to 100 base pairs, has an improved coefficient of determination compared to the conventional example, which takes into account deletions of more than 0 base pairs and up to 40 base pairs. On the other hand, the reference example showed a very high BIC value, indicating that the degree of completion of the regression model was worse than that of the conventional example.

[0121] As shown in Figure 6, according to the present invention, in which multiple regression models are generated by loop processing and a regression model selected based on the coefficient of determination and BIC value is used, the coefficient of determination is equivalent to that of the conventional example, and the BIC value is also improved compared to the reference example. This result shows that the regression model generated in the present invention achieves both high accuracy and excellent completeness. Therefore, it is considered that the regression model generated by the present invention can estimate the mutation introduction efficiency with higher accuracy than the conventional one.

[0122] [Additional Notes] The present invention is not limited to the above-described embodiments and examples, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in the different embodiments or examples are also included in the technical scope of the present invention. [Explanation of symbols]

[0123] 12 Virtual Array Generation Unit 13 Model Generation Unit 14 Estimation part 100, 100a Estimation System

Claims

1. A method for estimating efficiency of mutation introduction by a genome editing tool, comprising: A virtual sequence generation step of generating a virtual sequence including the mutation that may occur in the target genome region by the introduction of the mutation for each mutation from control sequence information in the target genome region into which the mutation is to be introduced and at least one recognition sequence information in the target genome region recognized by the genome editing tool; a model generation step of generating a regression model capable of estimating the type and frequency of the mutation contained in sequence analysis information in the target genome region of the nucleic acid obtained from the population of cells into which the mutation was introduced, using a combination of the virtual sequences that is a part of the generated virtual sequences, and calculating an accuracy evaluation index and a completeness evaluation index of the regression model, the step being repeated a predetermined number of times using a different combination each time; and an estimation step of estimating the efficiency of mutation introduction using a regression model whose evaluation indicated by the completeness evaluation index is higher than a predetermined threshold among the multiple regression models generated and whose evaluation indicated by the accuracy evaluation index is higher than a predetermined threshold.

2. In the virtual array generation step, The method according to claim 1 , further comprising the step of generating said virtual sequence using template sequence information for introducing said mutation in addition to said control sequence information and said recognition sequence information.

3. In the model generation step, 2. The estimation method according to claim 1, wherein, if the evaluation indicated by the completeness evaluation index of the regression model generated in the second or subsequent iteration is lower than the evaluation indicated by the completeness evaluation index of any of the regression models generated previously, the combination of the virtual arrays used in the next or subsequent iterations is made part of the virtual array used this time.

4. In the virtual array generation step, The method according to claim 1 , wherein a plurality of virtual sequences are generated, the virtual sequences including at least one deletion of more than 40 base pairs and not more than 100 base pairs as the possible mutation.

5. An estimation system for estimating the efficiency of mutation introduction by a genome editing tool, A virtual sequence generation unit that generates a virtual sequence including the mutation that may occur in the target genome region by the introduction of the mutation, for each mutation, from control sequence information in the target genome region into which the mutation is to be introduced and recognition sequence information at at least one location in the target genome region recognized by the genome editing tool; a model generation unit that generates a regression model capable of estimating the type and frequency of the mutation contained in sequence analysis information in the target genome region of the nucleic acid obtained from the population of cells into which the mutation has been introduced, using a combination of the virtual sequences that is a part of the generated virtual sequences, and repeats a process of calculating an accuracy evaluation index and a completeness evaluation index of the regression model a predetermined number of times using a different combination each time; an estimation unit that estimates the efficiency of mutation introduction using a regression model whose evaluation indicated by the completeness evaluation index is higher than a predetermined threshold among the multiple regression models generated and whose evaluation indicated by the accuracy evaluation index is higher than a predetermined threshold.

6. A control program for controlling a computer that estimates the efficiency of mutation introduction by a genome editing tool, The computer includes: A virtual sequence generation step of generating a virtual sequence including the mutation that may occur in the target genome region by the introduction of the mutation for each mutation from control sequence information in the target genome region into which the mutation is to be introduced and at least one recognition sequence information in the target genome region recognized by the genome editing tool; a model generation step of generating a regression model capable of estimating the type and frequency of the mutation contained in sequence analysis information in the target genome region of the nucleic acid obtained from the population of cells into which the mutation was introduced, using a combination of the virtual sequences that is a part of the generated virtual sequences, and calculating an accuracy evaluation index and a completeness evaluation index of the regression model, the process of repeating this process a predetermined number of times using a different combination each time; and an estimation step of estimating the efficiency of mutation introduction using a regression model whose evaluation indicated by the completeness evaluation index is higher than a predetermined threshold among the multiple regression models generated and whose evaluation indicated by the accuracy evaluation index is higher than a predetermined threshold.

7. A computer-readable recording medium on which the control program according to claim 6 is recorded.

Images