Method and system for predicting crisper off-target effects
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NANKAI UNIV
- Filing Date
- 2024-03-05
- Publication Date
- 2026-08-07
AI Technical Summary
然而，它们都使用或运算来人为地压缩目标上和目标外序列的编码，在一定程度上限制了输入特征的表示空间
[0034]与其他可应用于indel的工作相比，本发明将中靶和脱靶序列对、中靶序列和脱靶序列，进行编码后输入到脱靶效应预测模型中，编码方式的目的是将特征从人工编码转换成脱靶效应预测模型中国的可自适应的词向量编码，从而减少人工编码造成的信息损失。
Smart Images
Figure CN118098342B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of synthetic biology gene editing technology, and in particular to a method and system for predicting CRISPR off-target effects. Background Technology
[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.
[0003] The Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-CRISPR-associated protein 9 (Cas9) system is an advanced technology applicable to genome engineering. It is a two-component system in which the Cas9 endonuclease is guided to the DNA target sequence upstream of the PAM and complemented by sgRNA, allowing the bases of the target sequence to be edited. It holds promise for applications in gene therapy and agricultural productivity. However, the CRISPR-Cas9 system suffers from significant off-target problems. Various issues, such as mismatched sequences, cannot guarantee the specificity of gene editing.
[0004] Quantifying off-target effects using sequencing technologies such as GUIDE-SEQ, SITE-SEQ, Circle-SEQ, and Digenome-SEQ can help optimize CRISPR-Cas9 systems. While these technologies perform well in terms of accuracy, considering time, money, and labor costs, introducing computational methods to quantify off-target effects may be more cost-effective. Furthermore, the ability of machine learning methods to capture latent features can help researchers improve the efficiency and specificity of CRISPR-Cas9 systems.
[0005] The above research has made encouraging progress using handcrafted features and machine learning models. Furthermore, some studies have further developed the ability to predict off-target efficiency in two aspects: (1) The input of the above machine learning methods is mainly handcrafted features, which effectively promotes the prediction of off-target efficiency. However, these handcrafted features may increase specialization and heterogeneity, resulting in weaker generalization ability of the machine learning models. Several studies have shown good off-target effect prediction performance using only sequence features; (2) The above machine learning models still have limited ability to mine data features and make predictions, and some studies have shown that deep learning models have stronger off-target effect prediction capabilities. However, these methods only consider mismatches in the off-target problem, ignoring insertions and deletions between the target DNA and guide RNA sequences, which also affects the off-target effect problem.
[0006] Several deep learning models applicable to INDELs have demonstrated their capabilities in their own experiments. However, they all use OR operations to artificially compress the encoding of on-target and off-target sequences, which to some extent limits the representation space of the input features. Furthermore, the relatively small architecture of these models limits their generalization ability and their applicability to datasets with diverse features. In conclusion, existing off-target effect prediction models still have room for improvement. Summary of the Invention
[0007] To address the technical problems existing in the background art, the present invention provides a CRISPR off-target effect prediction method and system. The present invention designs an off-target effect prediction model, aiming to improve the off-target effect prediction capability through improvements at the computer technology level.
[0008] To achieve the above objectives, the present invention adopts the following technical solution:
[0009] The first aspect of the present invention provides a method for predicting CRISPR off-target effects.
[0010] CRISPR off-target effect prediction methods include:
[0011] Obtain the target hit and off-target sequence pairs, the target hit sequence, and the off-target sequence;
[0012] Word embedding and positional encoding are performed on the target-target and off-target sequence pairs, the target sequence, and the off-target sequence to obtain the target-target and off-target sequence pair features, the target sequence features, and the off-target sequence features.
[0013] The features of the hit-target and off-target sequence pairs, the features of the hit-target sequence, and the features of the off-target sequence are respectively input into three network branches for feature extraction, resulting in the first feature, the second feature, and the third feature.
[0014] The first, second, and third features are fused and passed through a fully connected layer to obtain the predicted value.
[0015] Furthermore, the process of performing the word embedding encoding includes: setting a dictionary, converting bases and base pairs into word index vectors required for word embedding encoding, and encoding the word index vectors through an embedding layer to obtain a word vector matrix.
[0016] Furthermore, the word embedding encoding process includes: adding one more bit on top of the 20-bit target sequence and the 3-bit PAM sequence to form a 24-bit base sequence or base pair sequence; encoding the 24-bit base sequence or base pair sequence into a word index vector using a dictionary; and encoding the word index vector through an embedding layer to obtain a word vector matrix.
[0017] Furthermore, the embedding layer is encoded using the following formula:
[0018] V e =V i ×M
[0019] Among them, V i This represents the word index vector encoded by the one-hot code corresponding to the ordinal number in the dictionary, assuming the dictionary size is s. a Then V i The shape is [24, s a Assuming the word embedding length is l b Then the shape of the embedding layer parameter matrix M is [s a ,l b The word vector matrix V output by the embedding layer is... e The shape is [24, l b ].
[0020] Furthermore, the process of performing the positional encoding includes: performing positional encoding on the word vector matrix to obtain an encoding that includes positional information for the distance between adaptive discrete information.
[0021] Furthermore, each of the network branches includes a self-attention layer and a convolutional layer. The self-attention layer enhances the target sequence pair features, target sequence features, and off-target sequence features, while the convolutional layer extracts the first feature, the second feature, and the third feature.
[0022] A second aspect of the present invention provides a CRISPR off-target effect prediction system.
[0023] The CRISPR off-target effect prediction system includes:
[0024] The data acquisition module is configured to acquire target hit and off-target sequence pairs, target hit sequence, and off-target sequence.
[0025] The encoding module is configured to perform word embedding encoding and position encoding on the target-out-target sequence pairs, the target sequence, and the out-target sequence to obtain the target-out-target sequence pair features, the target sequence features, and the out-target sequence features.
[0026] The feature extraction module is configured to input the target-target and off-target sequence pair features, the target-target sequence features, and the off-target sequence features into three network branches for feature extraction, thereby obtaining the first feature, the second feature, and the third feature.
[0027] The prediction module is configured to fuse the first feature, the second feature, and the third feature, and then pass them through a fully connected layer to obtain the predicted value.
[0028] A third aspect of the present invention provides a computer-readable storage medium.
[0029] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the CRISPR off-target effect prediction method as described in the first aspect above.
[0030] A fourth aspect of the present invention provides a computer device.
[0031] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the steps in the CRISPR off-target effect prediction method as described in the first aspect above.
[0032] A fifth aspect of the present invention provides a computer program product comprising a computer program / instructions that, when executed by a processor, implement the steps in the CRISPR off-target effect prediction method as described in the first aspect.
[0033] Compared with the prior art, the beneficial effects of the present invention are:
[0034] Compared to other works applicable to indel, this invention encodes the target and off-target sequence pairs, the target sequence and the off-target sequence, and then inputs them into the off-target effect prediction model. The purpose of the encoding method is to transform the features from manual encoding into adaptive word vector encoding in the off-target effect prediction model, thereby reducing the information loss caused by manual encoding.
[0035] The off-target effect prediction model (CRISPR-M) used in this invention has three branches corresponding to three input features: off-target sequence pair features, target sequence features, and off-target sequence features. Each branch processes its corresponding input features based on convolutional neural networks and recurrent neural networks. The outputs of the three branches are concatenated and processed through several fully connected layers to form the final output.
[0036] This invention collected test data for multiple types of mismatch and indel, and performed cross-validation on test sets with only indel, only mismatch, and both mismatch and indel. It showed competitive performance in terms of ROC, PRC, Spearson coefficient, and Fscore, and even significantly outperformed several recent models.
[0037] This invention visualizes and analyzes the features captured by the off-target effect prediction model, statistically analyzes the impact of base substitutions at different positions on the model's prediction score, reveals the influence of mismatches and indels on off-target effects, and verifies the effectiveness of the model. Attached Figure Description
[0038] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.
[0039] Figure 1 This is a framework diagram of the CRISPR off-target effect prediction method shown in this invention. Detailed Implementation
[0040] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0041] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0042] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.
[0043] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and systems according to various embodiments of this disclosure. It should be noted that each block in a flowchart or block diagram may represent a module, segment, or portion of code, which may include one or more executable instructions for implementing the logical functions specified in the various embodiments. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutively represented blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, may be implemented using a dedicated hardware-based system that performs the specified functions or operations, or using a combination of dedicated hardware and computer instructions.
[0044] Example 1
[0045] This embodiment provides a method for predicting CRISPR off-target effects. This embodiment uses the application of this method to a server as an example for illustration. It is understood that this method can also be applied to terminals, and can also be applied to systems including terminals, servers, and other components, and can be implemented through interaction between the terminal and the server. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network servers, cloud communication, middleware services, domain name services, CDN security services, and big data and artificial intelligence platforms. The terminal can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. The terminal and server can be directly or indirectly connected via wired or wireless communication, which is not limited herein. In this embodiment, the method includes the following steps:
[0046] Obtain the target hit and off-target sequence pairs, the target hit sequence, and the off-target sequence;
[0047] Word embedding and positional encoding are performed on the target-target and off-target sequence pairs, the target sequence, and the off-target sequence to obtain the features of the target-target and off-target sequence pairs, the features of the target sequence, and the features of the off-target sequence.
[0048] The features of the hit-target and off-target sequence pairs, the features of the hit-target sequence, and the features of the off-target sequence are respectively input into three network branches for feature extraction, resulting in the first feature, the second feature, and the third feature.
[0049] The first, second, and third features are fused and passed through a fully connected layer to obtain the predicted value.
[0050] The following is a detailed description of this embodiment with reference to the accompanying drawings, such as... Figure 1 As shown:
[0051] Figure 1 This embodiment demonstrates the main flow of the solution, which includes three parts: input, word embedding dictionary, and model architecture. A detailed explanation follows:
[0052] (1) Multi-view input
[0053] like Figure 1 The input module shown in this invention uses the target-miss sequence pair, the target sequence, and the miss sequence as three input features. These three features represent three perspectives: the correlation between the target and miss sequences, the features of the target sequence itself, and the features of the miss sequence itself. These three perspectives are somewhat coupled, but this invention aims to increase comprehensiveness and coverage by isolating the sequence features and sequence pair features, thereby reducing information loss during data encoding.
[0054] (2) Encoding Scheme
[0055] ONE-HOT is a discretization encoding method that maps discretized information to Euclidean space, ensuring that different types of information maintain equal distances. Since the four bases are discrete information, using an ONE-HOT-based encoding method is logical. However, this is equivalent to artificially setting equal distances between the four bases, and such an artificial setting may not conform to actual natural laws.
[0056] This invention employs a word embedding encoding method. Compared to one-hot encoding, word embedding encoding can also encode discrete information in a manner similar to distributed representation, and it can also adaptively handle the distance between discrete information, such as the distance between the four bases in Euclidean space mentioned earlier. Specifically, by setting a dictionary, such as... Figure 1 The word embedding module converts bases and base pairs into word indices required for word embedding encoding. To accommodate the indel, an additional bit is added above the 20-bit target sequence and the 3-bit PAM sequence, forming a 24-bit base sequence or base pair sequence. Figure 1 As shown, taking a 24-bit base pair sequence as an example, after encoding it as a word index using a dictionary, it is input into the embedding layer used for word embedding encoding. The encoding formula in the embedding layer is as follows:
[0057] V e =V i ×M (1)
[0058] Among them, V i This represents the word index vector encoded by the one-hot code corresponding to the ordinal number in the dictionary, assuming the dictionary size is s. a Then V i The shape is [24, s a [Assuming the word embedding length is l] b Then the shape of the embedding layer parameter matrix M is [s a ,l b The word vector matrix V output by the embedding layer is... e The shape is [24, l b Furthermore, since different target locations have varying impacts on miss efficiency, position encoding was also incorporated. The formula used for position encoding is as follows:
[0059]
[0060] Where pos represents the position of a base or base pair in the target sequence, i represents the dimension of the base or base pair word vector, and d represents the maximum dimension of the word vector. For example, for a word vector of length l b A sequence of 24 bases in length has 24 possible values for pos, while i has 1...b If d takes any value, then d equals l. b This formula assigns a unique position to each value in the word vector matrix. Let the word vector V... e The input is a positional encoding layer; after adding positional encoding, the output completes the entire encoding process. The encoding formula is as follows:
[0061] V = V e +V p (3)
[0062] Among them, V p It is a position-encoded matrix, where each element value is based on PE. (pos,i) Calculate; V p The result is obtained by calculating V using formula (2). e Position matrices of the same shape. Place the position matrix V... p With word vector matrix V e Adding them together yields the V matrix, which completes the encoding. Continuing to use the values from the example above, V here... e and V p The shapes are all [24, l b Through these processes, an encoding containing location information is obtained that can adaptively determine the distance between discrete information. In Experiment 2, compared with the encoding based on empirical settings, it can be found that the encoding method used in this invention does indeed have a performance improvement.
[0063] (3) Model Architecture
[0064] This invention proposes a deep learning model called CRISPR-M for predicting off-target effects of sgRNA, which includes convolutional layers, recurrent layers, attention layers, fully connected layers, some regularization strategies, and the aforementioned embedding and positional encoding layers for encoding. Figure 1 The model architecture module in the image demonstrates a simplified structure of CRISPR-M. Following the order from input to output, CRISPR-M can be divided into four parts: (a) embedding module; (b) convolutional module; (c) recurrent layer module; and (d) fully connected module.
[0065] First, the target sequence pair, the mid-target sequence, and the off-target sequence, encoded with word embeddings, are input into the embedding module as three inputs. These three inputs are encoded in the word embedding layer and the positional encoding layer, respectively, according to the rules described in the encoding scheme. Then, for each input, a branch is arranged for separate processing. At the beginning of each branch, a self-attention layer is used to enhance the features of each input, facilitating feature extraction in subsequent convolutional layers. The self-attention layer employs multi-head attention.
[0066] Secondly, a convolutional layer is placed after the attention layer for feature extraction. Note that, consistent with the encoding scheme, the vector size output by the attention layer here should be [24, l] for each sample. b The output of the attention layer is transformed into [24, l] before the convolutional layer. b [24, 1, fn], where fn represents the number of filters, and fn is initially set to 1. In each branch, one or two convolutional layers output a tensor of shape [24, 1, fn] after extracting features. This tensor is then transformed into [24, fn] and fed into a bidirectional recurrent layer with 32 units. Figure 1 As shown, the outputs of the three branches of the recurrent layer are flattened and connected together, and finally output a single predicted value through three fully connected layers with unit numbers of 256, 64 and 1.
[0067] The range of this single predicted value is [0,1]. A value close to 1 indicates that the sgRNA of the CRISPR system is likely to bind to this genomic site, while a value close to 0 indicates that it will not bind to this genomic site.
[0068] During training, the optimizer is set to Adam, and the model evaluation metrics are accuracy, AUROC, and AURPC. The loss function is the binary cross-entropy loss function. To achieve better training results, this invention also sets up callback functions: (a) if the output of the loss function no longer decreases in ten epochs, training is stopped early; (b) if the validation AURPC of the model in the current epoch is larger, the current model is saved; (c) if the validation loss does not change in four epochs, the learning rate is reduced.
[0069] Example 2
[0070] This embodiment provides a CRISPR off-target effect prediction system.
[0071] The CRISPR off-target effect prediction system includes:
[0072] The data acquisition module is configured to acquire target hit and off-target sequence pairs, target hit sequence, and off-target sequence.
[0073] The encoding module is configured to perform word embedding encoding and position encoding on the target-out-target sequence pairs, the target sequence, and the out-target sequence to obtain the target-out-target sequence pair features, the target sequence features, and the out-target sequence features.
[0074] The feature extraction module is configured to input the target-target and off-target sequence pair features, the target-target sequence features, and the off-target sequence features into three network branches for feature extraction, thereby obtaining the first feature, the second feature, and the third feature.
[0075] The prediction module is configured to fuse the first feature, the second feature, and the third feature, and then pass them through a fully connected layer to obtain the predicted value.
[0076] It should be noted that the data acquisition module, encoding module, feature extraction module, and prediction module described above are the same examples and application scenarios implemented in Embodiment 1, but are not limited to the content disclosed in Embodiment 1. It should also be noted that these modules, as part of a system, can be executed in a computer system, such as a set of computer-executable instructions.
[0077] Example 3
[0078] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the CRISPR off-target effect prediction method as described in Embodiment 1 above.
[0079] Example 4
[0080] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the CRISPR off-target effect prediction method as described in Embodiment 1 above.
[0081] Example 5
[0082] This embodiment provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps in the CRISPR off-target effect prediction method as described in Embodiment 1.
[0083] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.
[0084] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0085] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0086] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0087] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.
[0088] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A CRISPR off-target effect prediction method, characterized in that, include: Obtain the target hit and off-target sequence pairs, the target hit sequence, and the off-target sequence; Word embedding and positional encoding are performed on the target-target and off-target sequence pairs, the target sequence, and the off-target sequence to obtain the features of the target-target and off-target sequence pairs, the features of the target sequence, and the features of the off-target sequence. The features of the hit-target and off-target sequence pairs, the features of the hit-target sequence, and the features of the off-target sequence are respectively input into three network branches for feature extraction, resulting in the first feature, the second feature, and the third feature. The first, second, and third features are fused and passed through a fully connected layer to obtain the predicted value; The process of performing the word embedding encoding includes: setting a dictionary, converting bases and base pairs into word index vectors required for word embedding encoding, encoding the word index vectors through an embedding layer to obtain a word vector matrix; The word embedding encoding process includes: adding one bit above the 20-bit target sequence and the 3-bit PAM sequence to form a 24-bit base sequence or base pair sequence; encoding the 24-bit base sequence or base pair sequence into a word index vector using a dictionary; and encoding the word index vector through an embedding layer to obtain a word vector matrix. The embedding layer is encoded using the following formula: Among them, V i This represents the word index vector encoded by the one-hot code corresponding to the ordinal number in the dictionary, assuming the dictionary size is s. a Then V i The shape is [24, s a Assuming the word embedding length is l b Then the shape of the embedding layer parameter matrix M is [s a , l b The word vector matrix V output by the embedding layer is... e The shape is [24, l b ].
2. The CRISPR off-target effect prediction method according to claim 1, characterized in that, The process of performing the positional encoding includes: performing positional encoding on the word vector matrix to obtain an encoding that includes positional information for the distance between adaptive discrete information.
3. The CRISPR off-target effect prediction method according to claim 1, characterized in that, Each of the network branches includes a self-attention layer and a convolutional layer. The self-attention layer enhances the features of the target sequence and the off-target sequence, as well as the features of the target sequence and the off-target sequence. The convolutional layer extracts the first feature, the second feature, and the third feature.
4. A CRISPR off-target effect prediction system, characterized in that, include: The data acquisition module is configured to acquire target hit and off-target sequence pairs, target hit sequence, and off-target sequence. The encoding module is configured to perform word embedding encoding and position encoding on the target-out-target sequence pairs, the target sequence, and the out-target sequence to obtain the target-out-target sequence pair features, the target sequence features, and the out-target sequence features. The feature extraction module is configured to input the target-target and off-target sequence pair features, the target-target sequence features, and the off-target sequence features into three network branches for feature extraction, thereby obtaining the first feature, the second feature, and the third feature. The prediction module is configured to fuse the first feature, the second feature, and the third feature, and then pass them through a fully connected layer to obtain the predicted value. The process of performing the word embedding encoding includes: setting a dictionary, converting bases and base pairs into word index vectors required for word embedding encoding, encoding the word index vectors through an embedding layer to obtain a word vector matrix; The word embedding encoding process includes: adding one bit above the 20-bit target sequence and the 3-bit PAM sequence to form a 24-bit base sequence or base pair sequence; encoding the 24-bit base sequence or base pair sequence into a word index vector using a dictionary; and encoding the word index vector through an embedding layer to obtain a word vector matrix. The embedding layer is encoded using the following formula: Among them, V i This represents the word index vector encoded by the one-hot code corresponding to the ordinal number in the dictionary, assuming the dictionary size is s. a Then V i The shape is [24, s a Assuming the word embedding length is l b Then the shape of the embedding layer parameter matrix M is [s a , l b The word vector matrix V output by the embedding layer is... e The shape is [24, l b ].
5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the CRISPR off-target effect prediction method as described in any one of claims 1-3.
6. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the CRISPR off-target effect prediction method as described in any one of claims 1-3.
7. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps in the CRISPR off-target effect prediction method as described in any one of claims 1-3.