Off-target risk analysis method, off-target risk analysis system, program, recording medium

Abstract

The present invention addresses the problem of predicting a potential off-target risk of a DNA-binding tool even when genomic information is unknown or indeterminate. An off-target risk analysis method that comprises: a virtual sequence generation step (S2) for generating a plurality of virtual sequences including a sequence identical to a target sequence and a sequence in which at least one mutation is introduced into the target sequence; a score calculation step (S3) for calculating a score relating to the probability that a DNA-binding tool recognizing the target sequence acts on for each of the plurality of the virtual sequences; and a predicted result output step (S4) for outputting, on the basis of the calculated score, the predicted result indicating the potentiality that the DNA-binding tool acts on a sequence different from the target sequence.

Description

Technical Field

[0001] The present disclosure relates to an off-target risk analysis method for analyzing the risk of off-target effects occurring in genome editing and its derivative technologies, an off-target prediction system, and the like.

Background Art

[0002] Genome editing is a technique for introducing mutations such as deletions, substitutions, and insertions into the sequence of an arbitrary target gene by recognizing the DNA sequence (hereinafter, target sequence) of a target region on the genome sequence and using a DNA-binding tool capable of cleaving the target region. Examples of DNA-binding tools include zinc finger nuclease (ZFN), TALE nuclease (Transcription Activator-Like Effector Nuclease), CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) / Cas (CRISPR-associated protein), and the like.

[0003] In organisms and cells modified by genome editing, in addition to introducing mutations into the target genome sequence of the genome region, there is concern that a phenomenon (so-called off-target effect) in which unexpected mutations are introduced into non-target genome sequences may occur. Since the frequency of off-target effects depends on the DNA sequence recognized by the DNA-binding tool, even when the same DNA-binding tool (for example, CRISPR / Cas9) is used, if the target sequences are different, the frequency of off-target effects will be different. The risk that unexpected mutations will be introduced into other genome sequences that are not the target genome sequence in genome editing is also referred to as off-target risk. In addition, in various technologies applying DNA-binding tools (for example, transcriptional regulation, epigenome editing, chromosome imaging, etc.), zinc fingers, TALEs, and CRISPR / Cas with inactivated nuclease activity are similarly used, so there is a risk of off-target effects depending on the target sequence.

[0004] As described in Non-Patent Documents 1-3, various methods for analyzing off-target risks have been developed to assist in the design and selection of DNA-binding tools used in experiments. [Prior art documents] [Non-patent literature]

[0005] [Non-Patent Document 1] Pawel Sledzinski et al., “Computational Tools and Resources Supporting CRISPR-Cas Experiments”, Cells, Vol. 9, 1288, 2020. [Non-Patent Document 2] X. Robert Bao et al., “Tools for experimental and computational analyzes of off-target editing by programmable nucleases”, Nature Protocol, Vol. 16, pp10-26, 2021. [Non-Patent Document 3] Rongjie Fu et al., “Systematic decomposition of sequence determinants governing CRISPR / CAS9 specificity”, Nature Communications, Vol.13, 474, 2022. [Overview of the project] [Problems that the invention aims to solve]

[0006] Conventionally, the following methods (1) and (2) are known as computer-based methods for predicting off-target risks.

[0007] (1) A method for predicting candidate sequences that may cause off-target effects by performing homology searches, etc., using whole genome sequences (reference genomes) that represent biological species, virus strains, and populations such as races recorded in public databases, before performing genome editing, etc.

[0008] (2) A method for predicting candidate sequences that may cause off-target effects by referring to the whole genome sequence (hereinafter referred to as the unique genome) obtained by whole genome sequencing of the organism, tissue, cell clone, variety, bacterial strain, and virus strain that will actually undergo genome editing, etc., before performing genome editing, etc.

[0009] The off-target risk prediction methods described in (1) and (2) above are all applicable only to species whose whole genome information has been sequenced, or to specific individuals, tissues, cell clones, varieties, bacterial strains, and virus strains. Therefore, it has been difficult to apply them to predict off-target risks when using DNA-binding tools on species (including industrial organisms) whose genome information has not been sequenced or which are difficult to sequence due to the presence of obscure sequence elements such as repeat structures.

[0010] Furthermore, even for organisms whose entire genome has been sequenced, the genomic information is not definitive. This is because spontaneous mutations can occur in each individual and cell. By referring to a reference genome and the sequences of previously sequenced unique genomes, it is not possible to predict the potential off-target effects of DNA-binding tools that may arise when spontaneous mutations occur.

[0011] Thus, in various technologies that utilize DNA-binding tools (for example, medical applications using DNA-binding tools), the potential off-target risks of DNA-binding tools can be significant, but until now, there has been no method to predict these risks in advance. [Means for solving the problem]

[0012] An off-target risk analysis method according to one aspect of the present disclosure includes: a virtual sequence generation step of generating a plurality of virtual sequences including a sequence identical to a target sequence and a sequence in which at least one mutation has been introduced into the target sequence; a score calculation step of calculating a score for each of the plurality of virtual sequences relating to the probability that a DNA binding tool that recognizes the target sequence will act on it; and a prediction result output step of outputting a prediction result indicating the possibility that the DNA binding tool will act on a sequence other than the target sequence, based on the calculated score.

[0013] An off-target risk analysis system according to one aspect of the present disclosure includes: a virtual sequence generation unit that generates a plurality of virtual sequences including a sequence identical to a target sequence and a sequence in which at least one mutation has been introduced into the target sequence; a score calculation unit that calculates a score related to the probability that a DNA binding tool that recognizes the target sequence will act on each of the plurality of virtual sequences; and a prediction result output unit that outputs a prediction result indicating the possibility that the DNA binding tool will act on a sequence other than the target sequence, based on the calculated score.

[0014] Each aspect of the off-target risk analysis system according to this disclosure may be implemented by a computer, in which case the control program for the off-target risk analysis system that implements the off-target risk analysis system by operating the computer as each part (software element) of the off-target risk analysis system, and a computer-readable recording medium on which the control program is recorded, also fall within the scope of this disclosure. [Effects of the Invention]

[0015] This invention provides off-target risk analysis methods and systems that can predict the potential off-target risks of DNA-binding tools, even when genomic information is unknown or uncertain. [Brief explanation of the drawing]

[0016] [Figure 1]It is a block diagram showing an example of the schematic configuration of an off-target risk analysis system. [Figure 2] It is a diagram showing an example of a virtual array. [Figure 3] It is a diagram showing an example of a virtual array. [Figure 4] It is a diagram showing an example of a virtual array. [Figure 5] It is a flowchart showing an example of the processing flow executed by an off-target risk analysis device. [Figure 6] It is a diagram showing an example of the schematic configuration of an off-target risk analysis system. [Figure 7] It is a block diagram showing an example of the schematic configuration of an off-target risk analysis device. [Figure 8] It is a diagram showing an example of the data structure of user information. [Figure 9] It is a diagram showing an example of the data structure of an analysis result log. [Figure 10] It is a graph showing the correlation between the prediction result output using the off-target risk analysis method according to the present disclosure and the off-target occurrence rate obtained as a result of analyzing the off-target action in actual cells. [Figure 11] It is a graph showing the correlation between the prediction result output using the off-target risk analysis method according to the present disclosure and the off-target occurrence rate obtained as a result of analyzing the off-target action in actual cells. [Figure 12] It is a graph showing the correlation between the prediction result output using the off-target risk analysis method according to the present disclosure and the off-target occurrence rate obtained as a result of analyzing the off-target action in actual cells. [Figure 13] It is a scatter diagram showing the correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "CHANGE-seq" group. [Figure 14] It is a scatter diagram showing the correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "TTISS" group. [Figure 15] This scatter plot shows the correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "GUIDE-seq" group. [Figure 16] This is a scatter plot showing the correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "CHANGE-seq" group. [Figure 17] This is a scatter plot showing the correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "TTISS" group. [Figure 18] This is a scatter plot showing the correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "GUIDE-seq" group. [Modes for carrying out the invention]

[0017] [Embodiment 1] One embodiment of this disclosure will be described in detail below.

[0018] (Features of the Off-Target Risk Analysis System 100) The off-target risk analysis system 100 according to Embodiment 1 of this disclosure is a system that outputs a predictive result indicating that a DNA-binding tool that recognizes a target sequence may act on a sequence other than the target sequence.

[0019] DNA-binding tools are tools that specifically bind to a particular genomic region and enable cutting, modification, and editing of that genomic region. The genomic region subjected to cutting, modification, and editing by a DNA-binding tool may be a specific region on the cell's genome, and DNA-binding tools may also be called genome editing tools.

[0020] When a DNA-binding tool that recognizes a target sequence acts on a sequence other than the target sequence, this is called "off-target action." For example, off-target action on genes in a cell's genome can result in activation of oncogenes and inactivation of tumor suppressor genes in that cell. Furthermore, the effects of off-target action can be permanent. Therefore, it is desirable to accurately estimate the risk of off-target action (hereinafter referred to as off-target risk) before actually using a DNA-binding tool that recognizes a target sequence.

[0021] The off-target risk analysis system 100 performs the following processes (a) to (c). (a) Generate a number of virtual sequences, including a sequence identical to the target sequence recognized by the DNA binding tool, and a sequence in which at least one mutation has been introduced into the target sequence. (b) Calculate a score for each of several hypothetical sequences that relates to the probability that the DNA binding tool will act. (c) Based on the score calculated in (b) above, output a predictive result showing the off-target risk of the DNA binding tool.

[0022] In this disclosure, the “DNA-binding tool” may be any of the following (i) to (v). (i) A zinc finger, or a fusion polypeptide combining a zinc finger with a functional domain. (ii) TALE (Transcription Activator-Like Effector), or a fusion polypeptide combining a TALE with a functional domain. (iii) Pentatricopeptide repeat (PPR), or a fusion polypeptide combining a pentatricopeptide repeat with a functional domain. (iv) A fusion polypeptide-nucleic acid complex consisting of wild-type CRISPR / Cas (CRISPR-associated protein) or wild-type CRISPR / Cas combined with a functional domain. (v) A modified CRISPR / Cas (CRISPR-associated protein), or a fusion polypeptide-nucleic acid complex combining a modified CRISPR / Cas with a functional domain.

[0023] If the DNA-binding tool is as described in (i) above, the target sequence is a nucleotide sequence recognized by a domain having a zinc finger protein motif.

[0024] If the DNA binding tool is as described in (ii) above, the target sequence is a nucleotide sequence recognized by the region to which the TAL module (Transcription Activator-Like Module) is attached.

[0025] If the DNA binding tool is as described in (iii) above, the target sequence is a nucleotide sequence recognized by a region of consecutive PPR motifs.

[0026] If the DNA binding tool is either (iv) or (v) above, the target sequence is a nucleotide sequence complementary to the guide RNA (gRNA) that forms a complex with Cas, and a PAM (protospacer adjacent motif) sequence recognized by Cas.

[0027] DNA binding tools that fall under any of the above categories (i) to (v) may misidentify unintended DNA having a base sequence similar to the target sequence.

[0028] The analytical methods described in Non-Patent Documents 1-3 are all capable of analyzing off-target risks when using DNA-binding tools on species whose whole genome information has been sequenced, or on specific individuals, tissues, cell clones, varieties, bacterial strains, and viral strains. However, it is difficult to apply the analytical methods described in Non-Patent Documents 1-3 to predict off-target risks when using DNA-binding tools on species whose genome information has not been sequenced or is difficult to sequence (including industrial organisms).

[0029] In contrast, the off-target risk analysis system 100 according to this disclosure can output predictive results showing the off-target risk of a DNA-binding tool that recognizes a target sequence, provided that the target sequence is given. In other words, the off-target risk analysis system 100 can evaluate the potential risk of off-target effects that a DNA-binding tool that recognizes a target sequence may have, without referring to the genome sequence to which the DNA-binding tool is applied.

[0030] (Outline configuration of off-target risk analysis system 100) The following describes the general configuration of the off-target risk analysis system 100 using Figure 1. Figure 1 is a block diagram showing an example of the general configuration of the off-target risk analysis system 100.

[0031] As shown in Figure 1, the off-target risk analysis system 100 may include an off-target risk analysis device 1 and a display device 4. Figure 1 shows an off-target risk analysis system 100 that includes one off-target risk analysis device 1 and one display device 4. However, the configuration of the off-target risk analysis system 100 is not limited to this. For example, the number of display devices 4 in the off-target risk analysis system 100 may be zero or multiple.

[0032] In the off-target risk analysis system 100, the off-target risk analysis device 1 and the display device 4 are connected to each other in a way that allows them to communicate with one another. The off-target risk analysis device 1 and the display device 4 may be connected directly by wire or wireless connection, or they may be connected via a communication network. The form of the communication network is not limited and may be a local area network (LAN) or the internet.

[0033] The off-target risk analysis device 1 is a device that uses target sequence data to output predictive results indicating the off-target risk of DNA-binding tools. The outputted predictive results may be transmitted from the off-target risk analysis device 1 to the display device 4.

[0034] The display device 4 is typically a computer, smartphone, tablet, or the like used by a user of the off-target risk analysis system 100. Figure 1 shows an off-target risk analysis system 100 in which the display device 4 is separate from the off-target risk analysis device 1. However, the configuration of the off-target risk analysis system 100 is not limited to this. For example, the display device 4 may be an integrated device with the off-target risk analysis device 1, in which case the display device 4 may be a display unit (display, etc.) provided by the off-target risk analysis device 1.

[0035] (Configuration of Off-Target Risk Analysis System 1) Next, the configuration of the off-target risk analysis device 1 will be described. The off-target risk analysis device 1 comprises a control unit 10, a storage unit 20, and an input unit 30.

[0036] In one example, the control unit 10 may be a CPU (Central Processing Unit). The control unit 10 reads the control program, which is software stored in the memory unit 20, and loads it into memory such as RAM (Random Access Memory) to execute various functions. Note that in the memory unit 20 shown in Figure 1, the control program is not shown for the sake of simplicity.

[0037] The control unit 10 includes a target sequence receiving unit 11, a virtual sequence generation unit 12, a score calculation unit 13, and a prediction result output unit 14.

[0038] The target sequence receiving unit 11 receives target sequence data indicating the target sequence input using the input unit 30. The target sequence receiving unit 11 may store the received target sequence data in the storage unit 20.

[0039] The virtual sequence generation unit 12 generates multiple virtual sequences from the target sequence data, including sequences identical to the target sequence and sequences in which at least one mutation has been introduced into the target sequence. In other words, the virtual sequence generation unit 12 virtually generates diverse virtual sequence data, including sequences that may be misrecognized by the DNA binding tool that recognizes the target sequence. The virtual sequence generation unit 12 may also generate virtual sequences based on virtual sequence generation rules 21 stored in the storage unit 20.

[0040] Here, the mutation introduced into the target sequence by the virtual sequence generation unit 12 may be any of substitution, deletion, or insertion.

[0041] The virtual sequences generated by the virtual sequence generation unit 12 will be explained using Figures 2 to 4. Figures 2 to 4 show examples of virtual sequences. The virtual sequences shown in Figures 2 to 4 are generated from a target sequence derived from the human β-globin gene, which consists of 23 bases. However, in this disclosure, the virtual sequence may be shorter or longer than 23 bases. Also, in this disclosure, the target sequence is not limited to that derived from the human β-globin gene. Furthermore, in this disclosure, the virtual sequence does not have to be a continuous sequence; for example, it may be a discontinuous sequence with one or more arbitrary bases (N) sandwiched in one or more places.

[0042] First, a virtual sequence in which a substitution has been introduced into the target sequence will be explained using Figure 2. Figure 2 shows an example of a virtual sequence in which a substitution has been introduced into the target sequence. As shown in Figure 2, the virtual sequence generation unit 12 may generate multiple virtual sequences, including a sequence identical to the target sequence and a sequence in which a substitution has been introduced into the target sequence.

[0043] In Figure 2, sequence M1 (SEQ ID NO: 1) is the same as the target sequence. Sequences M2 to M7 show examples of sequences in which one of the nucleotides in sequence M1 has been substituted. For example, sequence M2 (SEQ ID NO: 2) is a sequence in which the "A" at the 5' end of sequence M1 has been replaced with "T", sequence M3 (SEQ ID NO: 3) is a sequence in which it has been replaced with "G", and sequence M4 (SEQ ID NO: 4) is a sequence in which it has been replaced with "C". Also, sequence M5 (SEQ ID NO: 5) is a sequence in which the second "G" from the 5' end of sequence M1 has been replaced with "A", sequence M6 (SEQ ID NO: 6) is a sequence in which it has been replaced with "T", and sequence M7 (SEQ ID NO: 7) is a sequence in which it has been replaced with "C".

[0044] Figure 2 shows sequences M2 to M7 as examples of virtual sequences in which a substitution is introduced at one nucleotide of the target sequence, but the virtual sequences generated by the virtual sequence generator 12 are not limited to these. The virtual sequence generator 12 may comprehensively generate sequences in which a single nucleotide substitution is introduced at the target sequence. Furthermore, the virtual sequence generator 12 may generate virtual sequences in which substitutions are introduced at multiple nucleotides of the target sequence. For example, the virtual sequence generator 12 may comprehensively generate sequences in which two-nucleotide substitutions, three-nucleotide substitutions, and four-nucleotide substitutions are introduced at the target sequence.

[0045] In other words, if the mutation is a substitution, the multiple virtual sequences generated by the virtual sequence generation unit 12 may include the following sequences. • A sequence obtained by substituting at least one adenine (A) in the target sequence with at least one of thymine (T), cytosine (C), and guanine (G), and / or • A sequence obtained by substituting at least one thymine (T) in the target sequence with at least one of adenine (A), cytosine (C), and guanine (G), and / or • A sequence obtained by substituting at least one cytosine (C) in the target sequence with at least one of adenine (A), thymine (T), and guanine (G), and / or • A sequence in which at least one guanine (G) in the target sequence is replaced with at least one of adenine (A), thymine (T), and cytosine (C).

[0046] Next, a virtual sequence in which a deletion has been introduced into the target sequence will be explained using Figure 3. Figure 3 shows an example of a virtual sequence in which a deletion has been introduced into the target sequence. As shown in Figure 3, the virtual sequence generation unit 12 may generate multiple virtual sequences, including a sequence identical to the target sequence and a sequence in which a deletion has been introduced into the target sequence.

[0047] In Figure 3, sequence M1 (sequence number 1) is the same sequence as the target sequence. Sequences M8 to M10 show examples of sequences in which a deletion has been introduced into one of the nucleotides contained in sequence M1. For example, sequence M8 (sequence number 8) is a sequence in which the "A" at the 5' end of sequence M1 has been deleted, sequence M9 (sequence number 9) is a sequence in which the second "G" from the 5' end of sequence M1 has been deleted, and sequence M10 (sequence number 10) is a sequence in which the third "C" from the 5' end of sequence M1 has been deleted.

[0048] Figure 3 shows sequences M8 to M10 as examples of virtual sequences in which a deletion is introduced into one nucleotide of the target sequence, but the virtual sequences generated by the virtual sequence generator 12 are not limited to these. The virtual sequence generator 12 may comprehensively generate sequences in which a single nucleotide deletion is introduced into the target sequence. Furthermore, the virtual sequence generator 12 may generate virtual sequences in which deletions are introduced into multiple nucleotides of the target sequence. For example, the virtual sequence generator 12 may comprehensively generate sequences in which two-nucleotide deletions, three-nucleotide deletions, and four-nucleotide deletions are introduced into the target sequence.

[0049] Next, a virtual sequence in which an insertion has been introduced into the target sequence will be explained using Figure 4. Figure 4 shows an example of a virtual sequence in which an insertion has been introduced into the target sequence. As shown in Figure 4, the virtual sequence generation unit 12 may generate multiple virtual sequences, including a sequence identical to the target sequence and a sequence in which an insertion has been introduced into the target sequence.

[0050] In Figure 4, sequence M1 (SEQ ID NO: 1) is the same as the target sequence. Sequences M11 to M18 show examples of sequences in which a single base insertion has been introduced into sequence M1. For example, sequence M11 (SEQ ID NO: 11) is a sequence in which "A" is inserted between the 5' end and the second nucleotide "AG" of sequence M1, sequence M12 (SEQ ID NO: 12) is a sequence in which "T" is inserted, sequence M13 (SEQ ID NO: 13) is a sequence in which "G" is inserted, and sequence M14 (SEQ ID NO: 14) is a sequence in which "C" is inserted. Also, sequence M15 (SEQ ID NO: 15) is a sequence in which "A" is inserted between the second and third nucleotides "GC" from the 5' end of sequence M1, sequence M16 (SEQ ID NO: 6) is a sequence in which "T" is inserted, sequence M17 (SEQ ID NO: 17) is a sequence in which "G" is inserted, and sequence M18 (SEQ ID NO: 18) is a sequence in which "C" is inserted.

[0051] Figure 4 shows sequences M11 to M18 as examples of virtual sequences in which an insertion is introduced at one location in the target sequence, but the virtual sequences generated by the virtual sequence generation unit 12 are not limited to these. For example, the virtual sequence generation unit 12 may generate a sequence in which a two-base insertion is introduced at one location in the target sequence. Alternatively, the virtual sequence generation unit 12 may comprehensively generate sequences in which an insertion is introduced at one location in the target sequence. Furthermore, the virtual sequence generation unit 12 may generate virtual sequences in which insertions are introduced at multiple locations in the target sequence. For example, the virtual sequence generation unit 12 may comprehensively generate sequences in which insertions are introduced at two, three, and four locations in the target sequence.

[0052] The virtual sequence generation unit 12 may generate virtual sequences in which two or more mutations are introduced into the target sequence. For example, the virtual sequence generation unit 12 may generate a virtual sequence in which a substitution is introduced into the target sequence, a virtual sequence in which a deletion is introduced into the target sequence, and a virtual sequence in which an insertion is introduced into the target sequence.

[0053] For sequences with low homology to the target sequence recognized by the DNA binding tool, the tool is less likely to be effective. Therefore, each of the multiple virtual sequences only needs to differ from the target sequence by four or fewer nucleotides; there is little need to generate virtual sequences with more than four mutations. This reduces the burden on computational resources while ensuring the accuracy of the prediction results.

[0054] The off-target risk analysis device 1 may be configured not to generate virtual sequences that satisfy specific conditions and to calculate the scores described later. Here, virtual sequences that satisfy specific conditions may be, for example, virtual sequences that are expected from known knowledge to have a low probability of contributing to the occurrence of off-target effects. By adopting such a configuration, it is possible to prevent scores related to virtual sequences that have a low probability of contributing to the occurrence of off-target effects from excessively influencing the prediction results and to improve the accuracy of the output prediction results.

[0055] Returning to Figure 1, the score calculation unit 13 calculates a score for each of the multiple virtual sequences that relates to the probability that a DNA binding tool that recognizes the target sequence will act. The score calculation unit 13 may also calculate the score using a value that indicates the stability when a DNA binding tool is bound to each of the multiple virtual sequences. The virtual sequence generation unit 12 may calculate the score based on the score calculation rule 22 stored in the memory unit 20. The score calculation rule 22 may be a known score calculation rule that applies an in silico analysis method developed according to the type of DNA binding tool. Alternatively, the score calculation rule 22 may be a combination of multiple known calculation rules.

[0056] If the DNA binding tool is CRISPR / Cas9, then the same scoring rules 22 as those for scoring tools to calculate scores such as the MOFF score (see Non-Patent Literature 3), the CRISPR-Net score, and the CFD score can be applied. For example, CRISPR-Net is a different type of scoring tool from the MOFF score (Jiecong Lin et al., “CRISPR-Net: A Recurrent Convolutional Network Quantifies CRISPR Off-Target Activities with Mismatches and Indels”, Advanced Science, Vol 7, 1903562, 2020) (https: / / doi.org / 10.1002 / advs.201903562). If the DNA binding tool is one of the following: zinc finger nuclease (ZFN), TALE nuclease (TALEN), or pentatricopeptide repeat (PPR) nuclease, scoring can be performed similarly by, for example, alignment considering mismatch gaps and Tm value calculation, which can be done using Biophython.

[0057] The prediction result output unit 14 outputs a prediction result indicating the possibility that the DNA binding tool will act on a sequence other than the target sequence, based on the calculated score. The prediction result output unit 14 may also output an evaluation value as the prediction result, which is calculated using all the scores calculated for each of the multiple virtual sequences. This evaluation value may be the sum of all the scores calculated for each of the multiple virtual sequences. Furthermore, this value is not limited to the sum of all scores, but refers to a calculation that can be expressed as an n-variable function f(s1,s2,···,sn) for n virtual sequences. Here, "s" represents the score of each virtual sequence. This n-variable function f may include not only linear transformations but also nonlinear transformations by models generated by machine learning, and may also include a term that represents the error of the computer. Furthermore, the n-variable function f does not have to be a single-valued function, but may be a multi-valued function.

[0058] The system may be configured so that the scores calculated for a virtual sequence into which mutations have been introduced to make it less likely for the DNA binding tool to misrecognize (i.e., a sequence less likely to contribute to the occurrence of off-target effects) do not excessively influence the prediction results. In this case, the prediction result output unit 14 may output as the prediction result a value obtained by summing only the scores calculated for the virtual sequence into which mutations have been introduced to make it a sequence more likely to contribute to the occurrence of off-target effects.

[0059] (Process flow performed by the off-target risk analysis system) Next, the processing flow performed by the off-target risk analysis device will be explained using Figure 5. Figure 5 is a flowchart showing an example of the processing flow performed by the off-target risk analysis device 1. Figure 5 also shows the processing flow performed by the off-target risk analysis system 100, which includes the off-target risk analysis device 1.

[0060] First, the target sequence receiving unit 11 receives input of the target sequence from the user (step S1). The target sequence data indicating the target sequence may be, for example, text data.

[0061] Next, the virtual sequence generation unit 12 generates a number of virtual sequences, including a sequence identical to the target sequence and a sequence in which at least one mutation has been introduced into the target sequence (Step S2: Virtual Sequence Generation Step).

[0062] Next, the score calculation unit 13 calculates a score related to the probability that the DNA binding tool that recognizes the target sequence will act, for each of the multiple virtual sequences generated in step S2 (step S3: score calculation step).

[0063] The prediction result output unit 14 outputs a prediction result indicating the possibility that the DNA binding tool will act on a sequence other than the target sequence, based on the score calculated in step S3 (step S4: prediction result output step).

[0064] According to the above configuration, the off-target risk analysis device 1 generates multiple virtual sequences from target sequences recognized by the DNA binding tool, and calculates a score related to the probability that the DNA binding tool will act on each of the generated virtual sequences. Then, using the calculated scores, the off-target risk analysis device 1 outputs a prediction result indicating the possibility that the DNA binding tool will act on sequences other than the target sequence. This prediction result is, so to speak, information indicating the potential off-target risk of the DNA binding tool.

[0065] Thus, the off-target risk analysis device 1 can predict the potential off-target risks of DNA-binding tools even when the genomic information of the target on which the DNA-binding tool is applied is unknown or uncertain.

[0066] [Embodiment 2] Other embodiments of this disclosure are described below.

[0067] The off-target risk analysis system 100 shown in Figure 1 has an off-target risk analysis device 1 equipped with an input unit 30 that accepts target sequence input from the user and outputs the prediction results to a display device 4, but is not limited to this configuration. For example, as shown in the off-target risk analysis system 100a in Figure 6, the system may include an off-target risk analysis device 1a that is connected to communication devices 5a and 5b used by each user via a communication network 9.

[0068] In the off-target risk analysis system 100a shown in Figure 6, the off-target risk analysis device 1a receives target sequence data indicating the target sequence from communication devices 5a and 5b, respectively. The off-target risk analysis device 1a then transmits the prediction result corresponding to the target sequence received from communication device 5a to communication device 5a, and transmits the prediction result corresponding to the target sequence received from communication device 5b to communication device 5b. Note that Figure 6 shows the off-target risk analysis system 100a including communication devices 5a and 5b and the off-target risk analysis device 1a, but is not limited to this. In the off-target risk analysis system 100a, the off-target risk analysis device 1a may be able to communicate with two or more communication devices.

[0069] (Configuration of Off-Target Risk Analysis System 1a) The configuration of the off-target risk analysis device 1a will be explained with reference to Figure 7. Figure 7 is a functional block diagram showing an example of the configuration of an off-target risk analysis system 100a according to one aspect of this disclosure. For the sake of explanation, components having the same function as those described in the above embodiments will be denoted by the same reference numerals, and their descriptions will not be repeated.

[0070] As shown in Figure 7, the off-target risk analysis device 1a includes a communication unit 16 that functions as a communication interface with communication devices 5a and 5b. The target sequence receiving unit 11 receives target sequence data via the communication unit 16. The prediction result output unit 14 transmits the prediction results to communication devices 5a and 5b via the communication unit 16. The off-target risk analysis device 1a may also generate a web page for each target sequence showing the prediction results for the received target sequence and provide the user who transmitted the target sequence with information on how to access the web page.

[0071] Here, communication devices 5a and 5b may be communication devices used by users who have been pre-registered as users of the off-target risk analysis system 100a. In this case, the storage unit 20a may store user information 23, which includes information about users who have been pre-registered as users of the off-target risk analysis system 100a.

[0072] Figure 8 shows an example of the data structure of user information 23. User information 23 may associate a user ID assigned to each user with the user's name, affiliation, and contact information. In Figure 8, the user assigned user ID "U001" has the name "AA AA", belongs to "XX University School of Medicine", and has contact information (e.g., email address) "AAAA@xxx.xx.xx". The user assigned user ID "U002" has the name "BB BB", belongs to "YY Research Institute", and has contact information "BBBB@yyy.yy.yy". For example, the prediction results regarding the target sequence received from the user with user ID "U001" are sent to "AAAA@xxx.xx.xx".

[0073] The off-target risk analysis device 1a may store the prediction results for each received target sequence in the analysis result log 24 of the storage unit 20a. Figure 9 shows an example of the data structure of the analysis result log 24. The analysis result log 24 may associate the user ID assigned to each user with the target sequence data received from each user, the date and time of receipt, and the prediction result. In Figure 9, the target sequence data received from user ID "U001" on "2022 / 9 / 1" at "PM1:50" and the prediction result for that target sequence are stored in association with each other.

[0074] By adopting this configuration, the off-target risk analysis device 1a can provide each user who is the source of the target sequence data with prediction results obtained by analyzing target sequences received from each of multiple users. For example, the administrator managing the off-target risk analysis device 1a may charge each user (or the organization to which each user belongs) a predetermined fee as compensation for the service of providing prediction results regarding the received target sequences.

[0075] [Examples of implementation using software] The functions of the off-target risk analysis devices 1 and 1a (hereinafter referred to as "devices") are programs that cause a computer to function as the device, and these programs can be realized by programs that cause a computer to function as each control block of the device (particularly each part included in the control units 10 and 10a).

[0076] 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., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.

[0077] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0078] Furthermore, some or all of the functions of each of the above control blocks can also be implemented by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of this disclosure. In addition, it is also possible to implement the functions of each of the above control blocks by, for example, a quantum computer.

[0079] Furthermore, each process described in the above embodiments may be performed by AI (Artificial Intelligence). In this case, the AI ​​may operate on the control device described above, or it may operate on other devices (for example, an edge computer or a cloud server).

[0080] This disclosure is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of this disclosure.

[0081] 〔summary〕 An off-target risk analysis method according to Embodiment 1 of the present disclosure includes: a virtual sequence generation step of generating a plurality of virtual sequences including a sequence identical to a target sequence and a sequence in which at least one mutation has been introduced into the target sequence; a score calculation step of calculating a score for each of the plurality of virtual sequences relating to the probability that a DNA binding tool that recognizes the target sequence will act on it; and a prediction result output step of outputting a prediction result indicating the possibility that the DNA binding tool will act on a sequence other than the target sequence, based on the calculated score.

[0082] The off-target risk analysis method described above generates multiple hypothetical sequences, including sequences identical to the target sequence and sequences with at least one mutation introduced into the target sequence. A score related to the probability that a DNA binding tool recognizing the target sequence will act on each of these hypothetical sequences is calculated. Based on the calculated scores, a predictive result indicating the likelihood that the DNA binding tool will act on sequences other than the target sequence is output.

[0083] Using the off-target risk analysis method described above, given a target sequence, it is possible to output predictive results indicating the possibility that a DNA-binding tool that recognizes the target sequence may act on a sequence other than the target sequence. In other words, this off-target risk analysis method allows for the evaluation of the potential risk of off-target effects that a DNA-binding tool that recognizes a target sequence may have, without referring to the genome sequence to which it is applied.

[0084] The off-target risk analysis method according to aspect 2 of the present disclosure is characterized in that, in aspect 1, the at least one mutation may be introduced into the entire target sequence or a portion of the target sequence in the virtual sequence generation step.

[0085] Some DNA-binding tools are known to have a higher risk of off-target effects occurring in a specific part of the target sequence than in other parts. In such cases, the risk of off-target effects may be evaluated by focusing on the specific part with a higher risk of off-target effects.

[0086] According to the above configuration, in the virtual sequence generation step, at least one mutation is introduced into the entire target sequence or a portion of the target sequence. This allows for efficient assessment of the risk of off-target effects.

[0087] The off-target risk analysis method according to aspect 3 of this disclosure, in aspect 1 or 2 above, the mutation may be any of substitution, deletion, or insertion.

[0088] The off-target risk analysis method according to aspect 4 of the present disclosure, in aspect 3 above, if the mutation is a substitution, the plurality of virtual sequences may include: (1) a sequence in which at least one adenine (A) of the target sequence is substituted with at least one of thymine (T), cytosine (C), and guanine (G); and / or (2) a sequence in which at least one thymine (T) of the target sequence is substituted with at least one of adenine (A), cytosine (C), and guanine (G); and / or (3) a sequence in which at least one cytosine (C) of the target sequence is substituted with at least one of adenine (A), thymine (T), and guanine (G); and / or (4) a sequence in which at least one guanine (G) of the target sequence is substituted with at least one of adenine (A), thymine (T), and cytosine (C).

[0089] The above configuration allows for the comprehensive generation of multiple virtual sequences, including sequences identical to the target sequence and sequences in which at least one substitution has been introduced into the target sequence. This enables an unbiased evaluation of the potential risk of off-target effects that DNA-binding tools that recognize target sequences may have.

[0090] The off-target risk analysis method according to aspect 5 of this disclosure is such that, in any of aspects 1 to 4 above, in the virtual sequence generation step, each of the plurality of virtual sequences may differ from the target sequence by four or fewer nucleotides.

[0091] For sequences with low homology to the target sequence recognized by the DNA binding tool, the DNA binding tool is less likely to be effective. The above configuration allows for reduced computational load while ensuring the accuracy of prediction results.

[0092] The off-target risk analysis method according to aspect 6 of this disclosure may, in any of aspects 1 to 5 above, calculate the score in the score calculation step using a value that indicates the stability when the DNA binding tool is bound to each of the plurality of virtual sequences.

[0093] With the above configuration, the score can be calculated accurately for each of the multiple virtual arrays.

[0094] In any of the embodiments 1 to 6 described above, the off-target risk analysis method according to embodiment 7 of this disclosure may output an evaluation value calculated using all the scores calculated for each of the plurality of virtual sequences as the prediction result in the prediction result output step.

[0095] According to the above configuration, the score calculated for each of the multiple virtual sequences can be used to easily assess the potential risk of off-target effects that a DNA-binding tool that recognizes a target sequence may have.

[0096] The off-target risk analysis method according to aspect 8 of this disclosure may be, in any of aspects 1 to 7 above, the DNA-binding tool being (1) a zinc finger or a fusion polypeptide combining a zinc finger with a functional domain, (2) a TALE (Transcription Activator-Like Effector) or a fusion polypeptide combining a TALE with a functional domain, (3) a pentatricopeptide repeat or a fusion polypeptide combining a pentatricopeptide repeat with a functional domain, (4) a wild-type CRISPR / Cas (CRISPR-associated protein) or a fusion polypeptide-nucleic acid complex combining a wild-type CRISPR / Cas with a functional domain, or (5) a modified CRISPR / Cas (CRISPR-associated protein) or a fusion polypeptide-nucleic acid complex combining a modified CRISPR / Cas with a functional domain.

[0097] The off-target risk analysis system according to aspect 9 of this disclosure comprises: a virtual sequence generation unit that generates a plurality of virtual sequences including a sequence identical to the target sequence and a sequence in which at least one mutation has been introduced into the target sequence; a score calculation unit that calculates a score related to the probability that a DNA binding tool that recognizes the target sequence will act on each of the plurality of virtual sequences; and a prediction result output unit that outputs a prediction result indicating the possibility that the DNA binding tool will act on a sequence other than the target sequence based on the calculated score. The above configuration provides the same effects as aspect 1.

[0098] A program according to aspect 10 of the present disclosure causes a computer to perform the following steps: a virtual sequence generation step of generating a plurality of virtual sequences including a sequence identical to a target sequence and a sequence in which at least one mutation has been introduced into the target sequence; a score calculation step of calculating a score for each of the plurality of virtual sequences relating to the probability that a DNA binding tool that recognizes the target sequence will act on it; and a prediction result output step of outputting a prediction result that indicates the possibility that the DNA binding tool will act on a sequence other than the target sequence, based on the calculated score.

[0099] The recording medium relating to aspect 11 of this disclosure is a computer-readable recording medium on which the program described in aspect 10 is recorded. [Examples]

[0100] (Example 1) Hereinafter, Embodiment 1 of this disclosure will be described with reference to Figures 10 to 12.

[0101] Several studies have reported results from searching the entire human genome for off-target candidate sequences for CRISPR / Cas9, targeting sequences derived from the human genome, using experimental methods (e.g., Lamsfus-Celle, A. et al., Scientific Reports, 10, 10133.2020). Using the results of these experimental searches, the CRISPR / Cas9 off-target incidence rate can be calculated as the proportion of off-target reads to the total number of reads.

[0102] Therefore, in this embodiment, the correlation between the prediction results output by the off-target risk analysis method according to one aspect of this disclosure and the off-target incidence rate calculated as described above was investigated for 14 types of target sequences. In order to output the prediction results, virtual sequences were comprehensively generated by introducing a 4-nucleotide substitution into the target sequence. Furthermore, the MOFF score described in Non-Patent Document 3 was used as the score calculation tool to output the prediction results.

[0103] Figures 10 to 12 are graphs showing the correlation between the prediction results output using the off-target risk analysis method described herein and the off-target incidence rate obtained from the analysis of actual off-target effects within cells. The prediction results shown in Figure 10 are those output using a score calculated for a hypothetical sequence in which mutations have been comprehensively introduced throughout the target sequence. As shown in Figure 10, the correlation between the prediction results and the off-target incidence rate is high (R 2 (0.5675) It was demonstrated that the prediction accuracy of the prediction results output by the off-target risk analysis method according to one aspect of this disclosure is high.

[0104] The prediction results shown in Figure 11 use only the scores calculated for a hypothetical sequence in which a mutation is introduced in a non-seed region with low specificity (the 8 nucleotides immediately preceding the PAM) within the target sequence. As shown in Figure 11, it was found that the prediction accuracy improves by outputting the prediction results using only the scores calculated for a hypothetical sequence in which a mutation is introduced in a non-seed region with low specificity (R2 (0.57465).

[0105] The prediction results shown in Figure 12 are generated by comprehensively creating virtual sequences by introducing two nucleotide substitutions into the target sequence, calculating a score for each virtual sequence, and using the calculated scores to generate the prediction results. As shown in Figure 12, even when the number of substitutions introduced to generate the virtual sequences is reduced from 4 to 2, a decrease in the correlation between the prediction results and the off-target incidence rate is observed (R 2 The prediction accuracy remained high (0.5351).

[0106] (Example 2) Hereinafter, Embodiment 2 of this disclosure will be described with reference to Figures 13 to 18.

[0107] [Target sequence] The virtual sequence used in this embodiment is a nucleotide sequence in which a mismatch of up to one base pair is introduced to the entire target sequence. The target sequence was one that was used in off-target analysis experiments in Non-Patent Literature 3 and has a proven track record of evaluation with the scoring tool CRISPR-Net.

[0108] Specifically, we used 108 gRNA sequences used in the off-target analysis experiment "CHANGE-seq," 59 sequences used in the off-target analysis experiment "TTISS," and 10 sequences used in the off-target analysis experiment "GUIDE-seq," as described in Non-Patent Literature 3. Specifically, we selected and used guide RNA (gRNA) sequences that met the following condition I from these gRNA sequences.

[0109] Condition I: Genomic DNA sequences including PAM sequences can be extracted using the program "ExtendSeq.py," which is publicly available at https: / / github.com / KazukiNakamae / Frame_Editor_sgRNA_selection.

[0110] [Virtual array] The selected gRNA sequences were 102 used in the off-target analysis experiment "CHANGE-seq," 54 used in the off-target analysis experiment "TTISS," and 8 used in the off-target analysis experiment "GUIDE-seq." A virtual sequence was generated by introducing a mismatch of up to one base pair into the target sequence, which is a base sequence complementary to the selected gRNA sequence.

[0111] [Score calculation using the scoring tool CRISPR-Net] The scoring using the CRISPR-Net scoring tool was performed on the data analysis platform "Code Ocean" (https: / / codeocean.com). Specifically, a copy of the script from the CRISPR-Net execution space (https: / / codeocean.com / capsule / 9553651 / tree / v1) was created, and the execution scripts "run.sh" and "CRISPR_Net.py" were modified to automatically calculate the CRISPR-Net score for the generated virtual array.

[0112] [Specificity prediction score using CRISPR-Net score (prediction result)] Based on which of the off-target analysis experiments "CHANGE-seq," "TTISS," and "GUIDE-seq" described in Non-Patent Document 3 used the target sequence used to generate the virtual sequences described above, the virtual sequences were classified into three groups. Specifically, each virtual sequence was classified into the "CHANGE-seq" group, the "TTISS" group, and the "GUIDE-seq" group.

[0113] Next, for each of the hypothetical sequences in each group, a prediction result indicating the likelihood that the DNA binding tool will act on a sequence other than the target sequence was calculated by multiplying the sum of the CRISPR-Net scores by -1. That is, the prediction result is (-1) × Σ(CRISPR-Net score). Furthermore, for each of the hypothetical sequences in each group, a prediction result indicating the likelihood that the DNA binding tool will act on a sequence matching the target sequence and a sequence different from it was calculated by multiplying the sum of the MOFF scores by -1. That is, the prediction result is (-1) × Σ(MOFF score). Hereinafter, these prediction results output using the off-target risk analysis method described herein will be referred to as "specificity prediction scores".

[0114] [Comparison of specificity prediction score with experimental data] The specificity prediction score calculated as described above was compared with the actual off-target incidence rate. Here, the results of the off-target analysis experiments "CHANGE-seq," "TTISS," and "GUIDE-seq" described in Non-Patent Literature 3 were used as the actual off-target incidence rate. Specifically, in this embodiment, the "Off-on-ratio" score for each of the "CHANGE-seq" group, "TTISS" group, and "GUIDE-seq" group described in Non-Patent Literature 3 was used.

[0115] Scatter plots were created for the "Off-on-ratio" score and specificity prediction score for each of the "CHANGE-seq" group, the "TTISS" group, and the "GUIDE-seq" group. The correlation between the "Off-on-ratio" score and the specificity prediction score was evaluated using Spearman's correlation coefficient.

[0116] Figures 13 and 16 are scatter plots for the "CHANGE-seq" group, Figures 14 and 17 are scatter plots for the "TTISS" group, and Figures 15 and 18 are scatter plots for the "GUIDE-seq" group.

[0117] <Result> [Correlation between specificity prediction score calculated using CRISPR-Net score and "Off-on-ratio" score] Figure 13 is a scatter plot showing the correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "CHANGE-seq" group. As shown in Figure 13, the Spearman correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "CHANGE-seq" group was -0.231839. A significant correlation (Spearman rank correlation coefficient test for no correlation, p-value < 0.05) was found between the "Off-on-ratio" score and the specificity prediction score using the CRISPR-Net score.

[0118] Figure 14 is a scatter plot showing the correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "TTISS" group. As shown in Figure 14, the Spearman correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "TTISS" group was -0.495973. A significant correlation (Spearman rank correlation coefficient test of no correlation, p-value < 0.05) was found between the "Off-on-ratio" score and the specificity prediction score using the CRISPR-Net score.

[0119] Figure 15 is a scatter plot showing the correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "GUIDE-seq" group. As shown in Figure 15, the Spearman correlation between the specificity prediction score calculated using the CRISPR-Net score and the "Off-on-ratio" score in the "GUIDE-seq" group was -0.380952.

[0120] [Correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score] Figure 16 is a scatter plot showing the correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "CHANGE-seq" group. As shown in Figure 16, the Spearman correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "CHANGE-seq" group was -0.566788.

[0121] Figure 17 is a scatter plot showing the correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "TTISS" group. As shown in Figure 17, the Spearman correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "TTISS" group was -0.655498.

[0122] Figure 18 is a scatter plot showing the correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "GUIDE-seq" group. As shown in Figure 18, the Spearman correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score in the "GUIDE-seq" group was -0.761904.

[0123] [Correlation between the specificity prediction score calculated using the MOFF score and the "Off-on-ratio" score] In all three groups—the "CHANGE-seq" group, the "TTISS" group, and the "GUIDE-seq" group—a significant correlation (Spearman rank correlation coefficient test for no correlation, p-value < 0.05) was observed between the "Off-on-ratio" score and the specificity prediction score using the MOFF score.

[0124] Furthermore, among the "CHANGE-seq" group, "TTISS" group, and "GUIDE-seq" group, a significant correlation was observed between the "Off-on-ratio" score and the specificity prediction score using the CRISPR-Net score in the "CHANGE-seq" group and the "TTISS" group, excluding the "GUIDE-seq" group which had fewer gRNA sequences used in off-target analysis experiments.

[0125] These results strongly suggest that the MOFF score and the CRISPR-Net score can be applied as score calculation rule 22 in the off-target risk analysis method relating to this disclosure. [Explanation of Symbols]

[0126] 11 Target sequence receiving section 12 Virtual Array Generation Unit 13. Score Calculation Section 14. Prediction Result Output Unit 100, 100a Off-Target Risk Analysis System S2 Virtual Array Generation Step S3 Score Calculation Steps S4 Prediction result output step

Claims

1. A virtual sequence generation step in which a computer generates a plurality of virtual sequences, including a sequence identical to a target sequence and a sequence in which at least one mutation has been introduced into the target sequence, The computer calculates a score for each of the plurality of virtual sequences, relating to the probability that a DNA binding tool that recognizes the target sequence will act; The computer outputs a prediction result based on the calculated score, indicating that the DNA binding tool is likely to act on a sequence other than the target sequence. Includes, The aforementioned DNA-binding tool is Zinc fingers, or fusion polypeptides combining zinc fingers with functional domains. TALE (Transcription Activator-Like Effector), or a fusion polypeptide combining TALE with a functional domain. Pentatricopeptide repeat, or a fusion polypeptide combining a pentatricopeptide repeat with a functional domain. Wild-type CRISPR / Cas (CRISPR-associated protein), or a fusion polypeptide-nucleic acid complex combining wild-type CRISPR / Cas with a functional domain, or It is a modified CRISPR / Cas (CRISPR-associated protein), or a fusion polypeptide-nucleic acid complex combining a modified CRISPR / Cas with a functional domain. Off-target risk analysis methods.

2. In the virtual sequence generation step, the at least one mutation is introduced into the entire target sequence or a portion of the target sequence. The off-target risk analysis method according to claim 1.

3. The aforementioned mutation is one of substitution, deletion, or insertion. The off-target risk analysis method according to claim 1.

4. If the mutation is a substitution, the plurality of virtual sequences are A sequence obtained by substituting at least one adenine (A) in the target sequence with at least one of thymine (T), cytosine (C), and guanine (G), and / or A sequence obtained by substituting at least one thymine (T) in the target sequence with at least one of adenine (A), cytosine (C), and guanine (G), and / or A sequence obtained by substituting at least one cytosine (C) in the target sequence with at least one of adenine (A), thymine (T), and guanine (G), and / or A sequence obtained by substituting at least one guanine (G) in the target sequence with at least one of adenine (A), thymine (T), and cytosine (C), including, The off-target risk analysis method according to claim 3.

5. In the virtual sequence generation step, each of the plurality of virtual sequences differs from the target sequence by four or fewer nucleotides. The off-target risk analysis method according to claim 1.

6. In the score calculation step, the score is calculated using a value that indicates the stability when the DNA-binding tool is attached to each of the plurality of virtual sequences. The off-target risk analysis method according to claim 1.

7. In the prediction result output step, the computer outputs an evaluation value calculated using all the scores calculated for each of the plurality of virtual arrays as the prediction result. The off-target risk analysis method according to claim 1.

8. A virtual sequence generation unit that generates multiple virtual sequences, including a sequence identical to the target sequence and a sequence in which at least one mutation has been introduced into the target sequence, A score calculation unit calculates a score for each of the plurality of virtual sequences that relates to the probability that the DNA binding tool that recognizes the target sequence will act, Based on the calculated score, a prediction result output unit outputs a prediction result indicating the possibility that the DNA binding tool will act on a sequence different from the target sequence, Equipped with, The aforementioned DNA-binding tool is Zinc fingers, or fusion polypeptides combining zinc fingers with functional domains. TALE (Transcription Activator-Like Effector), or a fusion polypeptide combining TALE with a functional domain. Pentatricopeptide repeat, or a fusion polypeptide combining a pentatricopeptide repeat with a functional domain. Wild-type CRISPR / Cas (CRISPR-associated protein), or a fusion polypeptide-nucleic acid complex combining wild-type CRISPR / Cas with a functional domain, or An off-target risk analysis system consisting of a modified CRISPR / Cas (CRISPR-associated protein), or a fusion polypeptide-nucleic acid complex combining a modified CRISPR / Cas with a functional domain.

9. On the computer, A virtual sequence generation step that generates multiple virtual sequences, including a sequence identical to the target sequence and a sequence in which at least one mutation has been introduced into the target sequence, A score calculation step of calculating a score related to the probability that the DNA binding tool that recognizes the target sequence will act, for each of the plurality of virtual sequences, A prediction result output step that outputs a prediction result indicating the possibility that the DNA binding tool will act on a sequence different from the target sequence, based on the calculated score, A program to execute, The aforementioned DNA-binding tool is Zinc fingers, or fusion polypeptides combining zinc fingers with functional domains. TALE (Transcription Activator-Like Effector), or a fusion polypeptide combining TALE with a functional domain. Pentatricopeptide repeat, or a fusion polypeptide combining a pentatricopeptide repeat with a functional domain. Wild-type CRISPR / Cas (CRISPR-associated protein), or a fusion polypeptide-nucleic acid complex combining wild-type CRISPR / Cas with a functional domain, or A program that is a modified CRISPR / Cas (CRISPR-associated protein), or a fusion polypeptide-nucleic acid complex combining a modified CRISPR / Cas with a functional domain.

10. A computer-readable recording medium on which the program described in claim 9 is recorded.

Images