Interactive CRISPR / Cas9 off-target and targeting analysis system and analysis method

Abstract

The invention discloses an interactive CRISPR / Cas9 off-target and targeting analysis system and analysis method, and relates to the technical field of gene editing.The interactive CRISPR / Cas9 off-target and targeting analysis system comprises a user login management module, a data management module, a resource sharing module, a self-adaptive training module and a dynamic interaction analysis module, the user login management module has a task viewing function and a data uploading record viewing function, wherein the data uploading record viewing function allows a user to view historical records of uploaded data in a system background. By integrating various bioinformatics analysis tools and machine learning algorithms, a user-friendly interaction platform is provided, off-target and targeting analysis results are obtained in real time, non-professionals can also easily use the platform, researchers can better control the gene editing process by integrating targeting and off-target analysis, and the target editing efficiency is improved. Unexpected side effects are reduced, and development and application of a gene editing technology are promoted.

Description

Technical Field

[0001] The present invention relates to the field of gene editing technology, and in particular to an interactive CRISPR / Cas9 off-target and on-target analysis system and analysis method. Background Art

[0002] In the field of gene editing, CRISPR-Cas9 technology has attracted widespread attention due to its high efficiency, precision and ease of operation. This technology uses a designed guide RNA (sgRNA) and a nuclease Cas9 to locate and cut a specific gene sequence, thereby achieving gene knockout or modification. However, CRISPR-Cas9 technology also faces the challenge of off-target effects, that is, Cas9 may cut DNA at non-target sites, which may lead to unexpected gene mutations.

[0003] In practical applications, researchers usually verify the specificity and efficiency of CRISPR-Cas9 through a variety of experimental methods. This includes using PCR and sequencing to detect mutations at the target site, and evaluating potential off-target sites through whole genome sequencing (WGS). In addition, researchers have developed a variety of detection technologies, such as Digenome-seq and CIRCLE-seq, which can evaluate the off-target effects of CRISPR-Cas9 across the entire genome.

[0004] By integrating on-target and off-target analysis, researchers can better control the gene editing process, reduce unintended side effects, and promote the development and application of gene editing technology.

[0005] The existing technology has the following problems:

[0006] With the widespread application of CRISPR / Cas9 technology in the field of gene editing, the evaluation of its off-target effects and on-target efficiency has become crucial. Although existing gene editing tools have made some progress in improving editing specificity and efficiency, there is a lack of an integrated interactive system that can simultaneously analyze off-target and on-target efficiently and accurately. Most of the current tools focus on a single analysis and fail to provide a comprehensive perspective to comprehensively evaluate the editing effect of CRISPR / Cas9. In addition, existing tools have deficiencies in user interactivity, real-time data analysis, and personalized feedback, which limits researchers' in-depth understanding and rapid optimization of editing results. Therefore, an interactive CRISPR / Cas9 off-target and on-target analysis system and analysis method are proposed. Summary of the invention

[0007] The purpose of the present invention is to provide an interactive CRISPR / Cas9 off-target and on-target analysis system and analysis method, aiming to make up for the shortcomings of existing tools. The system integrates multiple bioinformatics analysis tools and machine learning algorithms to provide a user-friendly interactive platform to obtain off-target and on-target analysis results in real time, and can be easily used by non-professionals to solve the problems raised in the above background technology.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0009] In a first aspect, the present invention provides an interactive CRISPR / Cas9 off-target and on-target analysis system, the interactive CRISPR / Cas9 off-target and on-target analysis system comprising: a user login management module, a data management module, a resource sharing module, an adaptive training module and a dynamic interactive analysis module;

[0010] The user login management module includes a task viewing function and a data upload record viewing function, wherein the data upload record viewing function allows the user to view the historical records of uploaded data in the system background;

[0011] The data management module is used to allow users to upload data, receive uploaded genome data, and provide real-time upload progress feedback during the upload process.

[0012] A further improvement of the technical solution of the present invention is that: the resource sharing module provides sharable data and allows users to set their own data to be shared, and the data format is .fasta.

[0013] A further improvement of the technical solution of the present invention is that the adaptive training module is automatically adjusted according to real-time feedback during the model training process to improve training efficiency and model performance, including the use of validation set performance adjustment regularization, adaptive data weighting and adaptive learning rate algorithms AdaGrad, RMSprop, Adam.

[0014] A further improvement of the technical solution of the present invention is that: the AdaGrad is an adaptive learning rate algorithm that adjusts the learning rate of each parameter by accumulating historical gradient information;

[0015]

[0016] where g t is the gradient of the tth iteration, r t is the cumulative sum of squares of the gradients, and ε is a small number to prevent division by zero.

[0017] A further improvement of the technical solution of the present invention is that: the RMSProp is an improved version of AdaGrad, which solves the problem that the AdaGrad learning rate may become too small by using an exponentially weighted moving average to calculate the square gradient;

[0018]

[0019]

[0020] Where β is a constant between 0 and 1 that controls the decay rate of the moving average.

[0021] A further improvement of the technical solution of the present invention is that: the Adam combines the ideas of RMSProp and the momentum method (Momentum), and dynamically adjusts the learning rate of each parameter by calculating the first-order moment estimate (mean) and the second-order moment estimate (uncentered variance) of the gradient;

[0022] m t =β1×m t-1 +(1-β1)×g t

[0023]

[0024]

[0025] Where m t and v t are the estimates of the first-order moment and the second-order moment, respectively, and β1 and β2 are the exponential decay factors for estimating these two moments.

[0026] A further improvement of the technical solution of the present invention is that the dynamic interactive analysis module is the core component of the platform. By integrating a variety of bioinformatics analysis tools and online interactive analysis tools, it provides users with a powerful and user-friendly analysis environment, mainly including off-target effect analysis models, on-target effect analysis models, and data visualization and result display.

[0027] In a second aspect, the present invention further provides an interactive CRISPR / Cas9 off-target and on-target analysis method, the interactive CRISPR / Cas9 off-target and on-target analysis method comprising the following steps:

[0028] Step 1: The user logs in to the system platform and obtains the user's login credential information;

[0029] Step 2: Select the platform data set or user uploaded data. If the user uploads data, you can choose whether to share the data;

[0030] Step 3: Provide users with adaptive training options, personalized parameter settings and optimization services;

[0031] Step 4: Allow the user to select a suitable bioinformatics analysis tool from the system;

[0032] Step 5: Provide a user interface to achieve dynamic interactive analysis of data, through which users can interact and analyze data with the system in real time;

[0033] Step 6. Users can view the analysis results, and these results can be saved and integrated into the user's personal analysis report, which will show detailed analysis results.

[0034] Due to the adoption of the above technical solution, the present invention has the following technical advances compared with the prior art:

[0035] 1. The present invention provides an interactive CRISPR / Cas9 off-target and on-target analysis system and analysis method, aiming to make up for the shortcomings of existing tools. The system integrates multiple bioinformatics analysis tools and machine learning algorithms to provide a user-friendly interactive platform to obtain off-target and on-target analysis results in real time, and can be easily used by non-professionals.

[0036] 2. The present invention provides an interactive CRISPR / Cas9 off-target and on-target analysis system and analysis method. By integrating on-target and off-target analysis, researchers can better control the gene editing process, reduce unexpected side effects, and promote the development and application of gene editing technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the analysis system module of the present invention;

[0038] Figure 2 Schematic diagram of the analytical method of the present invention. DETAILED DESCRIPTION

[0039] The present invention is further described in detail below in conjunction with embodiments:

[0040] Example 1

[0041] First, as Figure 1 As shown, the present invention provides an interactive CRISPR / Cas9 off-target and on-target analysis system, which includes: a user login management module, a data management module, a resource sharing module, an adaptive training module and a dynamic interactive analysis module;

[0042] The user login management module includes a task viewing function and a data upload record viewing function. The data upload record viewing function allows the user to view the historical records of uploaded data in the system background.

[0043] The data management module is used to allow users to upload data, receive uploaded genomic data, and provide real-time upload progress feedback during the upload process. This platform will provide existing public datasets by default, including

[0044] Off-target data sets include Doench et al. (Protein knockout detection), Haeussler et al. (PCR, Digenome-Seq and HTGTS), Cameron et al. (SITE-Seq), Tasi et al. (GUIDE-seq), Kleinstiver et al (GUIDE-seq), Kleinstiver et al (GUIDE-seq), Chuai et al. al.(GUIDE-Seq,Digenome-Seq,HTGTS,BLESS and IDLVs),Chuai et al.(GUIDE-Seq,Digenome-Seq,HTGTS,BLESS and IDLVs),CIRCLE_seq_10gRNA_wholeDataset,elevation_6gRNA_wholeDataset

[0045] The targeting data sets include HCT116, HEK293T, HELA, HL60, K562, A549, NB4, eSpCas9, SpCas9-HF1, WT-SpCas9, Sniper-Cas9, SpCas9-NG, and xCas9.

[0046] The resource sharing module provides sharable data and allows users to set their own data for sharing. The data format is .fasta.

[0047] The adaptive training module is automatically adjusted according to real-time feedback during model training to improve training efficiency and model performance, including regularization using validation set performance, adaptive data weighting, and adaptive learning rate algorithms AdaGrad, RMSprop, and Adam.

[0048] (1) Using validation set performance to adjust regularization is to dynamically adjust the regularization parameters based on the performance feedback on the validation set. For example, if the error on the validation set starts to increase, the regularization strength can be increased to suppress overfitting.

[0049] (2) Adaptive data weighting is used to process unbalanced data sets. Different weights can be assigned to samples of different categories. These weights can be adaptively adjusted based on the performance of the model during training.

[0050] (3) The adaptive learning rate algorithm is mainly used to dynamically adjust the learning rate through real-time feedback during model training. It not only speeds up the convergence of the model in the early stage, but also can refine the adjustment of the model weights by reducing the learning rate in the later stage of training, thereby avoiding excessive oscillation or falling into the local minimum. In addition, by customizing the learning rate for each parameter, the algorithm enables the model to more effectively handle the update requirements of different parameters, especially when dealing with sparse data or non-stationary gradients, reducing the need to manually adjust the learning rate, making the model training process more stable and reliable.

[0051] a.AdaGrad is an adaptive learning rate algorithm that adjusts the learning rate of each parameter by accumulating historical gradient information;

[0052]

[0053]

[0054] where g t is the gradient of the tth iteration, r t is the cumulative sum of squares of the gradient, ε is a small number to prevent division by zero;

[0055] b.RMSProp is an improved version of AdaGrad. It solves the problem that the learning rate of AdaGrad may become too small by using exponentially weighted moving average to calculate the square gradient.

[0056]

[0057]

[0058] Where β is a constant between 0 and 1, which is used to control the decay rate of the moving average;

[0059] c.Adam combines the ideas of RMSProp and Momentum to dynamically adjust the learning rate of each parameter by calculating the first-order moment estimate (mean) and second-order moment estimate (uncentered variance) of the gradient;

[0060] m t =β1×m t-1 +(1-β1)×g t

[0061]

[0062]

[0063] Where m t and v tare the estimates of the first-order moment and the second-order moment, respectively, and β1 and β2 are the exponential decay factors for estimating these two moments.

[0064] The dynamic interactive analysis module is the core component of the platform. By integrating multiple bioinformatics analysis tools and online interactive analysis tools, it provides users with a powerful and user-friendly analysis environment. This module provides a series of professional bioinformatics analysis tools to help researchers accurately evaluate the specificity and efficiency of gene editing technology. It mainly includes off-target effect analysis models, on-target effect analysis models, data visualization and result display. The following are the main applications of this module in gene off-target and on-target analysis:

[0065] (1) Off-target effect analysis model

[0066] 1. CNN

[0067] DeepCRISPR: uses a CNN architecture to automatically identify sequence and epigenetic features that affect sgRNA effects, improving the prediction accuracy and design optimization of CRISPR gene editing;

[0068] DL-CRISPR: Use CNN and data augmentation techniques to improve the accuracy and generalization of off-target activity prediction in the CRISPR / Cas9 system;

[0069] 2.CNN+Transformer

[0070] AttnToMismatch_CNN: Combining CNN and Transformer to Improve Prediction Accuracy of Non-target Specificity and Cell-specific Adaptability of CRISPR-Cas9 System Recurrent Neural Network;

[0071] 3. RNN

[0072] dl-offtarget: Utilizes the sequence processing capability of RNN to improve the prediction accuracy of off-target effects in CRISPR-Cas9 gene editing;

[0073] 4.CNN+RNN

[0074] CnnCrispr: Combining CNN and RNN, and introducing the GloVe model for feature extraction, significantly improved the prediction accuracy of sgRNA non-targeting effects in the CRISPR-Cas9 system;

[0075] 5. Transformer

[0076] CrisprDNT: A novel Transformer-based model that effectively improves the prediction accuracy of non-target effects in CRISPR-Cas9 gene editing through innovative sgRNA-DNA sequence encoding technology and anti-noise strategy;

[0077] 6. BERT

[0078] CRISPR-BERT: Leveraging the contextual text understanding capabilities of the BERT model, natural language processing techniques are applied to CRISPR / Cas9 gene editing to improve the prediction accuracy and interpretation of off-target effects.

[0079] (2) Targeting efficiency analysis model

[0080] 1. CNN

[0081] DeepCRISPR: CNN is used to analyze sgRNA sequence features, thereby improving the accuracy of target prediction;

[0082] DeepCas9: Uses CNN to automatically learn sgRNA sequence features, visualizes convolution kernels, and displays the matching of identified sequence features with known nucleotide preferences, thereby accurately predicting the targeting efficiency of CRISPR-Cas9;

[0083] 2. CNN+SVR

[0084] CNN-SVR: The combined CNN as the front-end to extract gRNA features and SVR as the back-end to regress and predict gRNA cleavage efficiency improves the prediction accuracy and generalization ability of RNA targeting efficacy;

[0085] 3.CNN+Transform

[0086] AttnToCrispr_CNN: combines the feature extraction capability of CNN with the sequence feature emphasis of the attention mechanism to improve the performance and accuracy of CRISPR / Cas9 sgRNA activity prediction;

[0087] 4.CNN+Attention

[0088] CRISPR-ONT: Incorporating attention mechanisms into CNN to enhance feature recognition for CRISPR gRNA activity prediction, thereby improving the prediction accuracy and interpretability of the model;

[0089] 5.CNN+XGBoost

[0090] CNN-XG: By integrating the image feature extraction capabilities of convolutional neural networks and the efficient tree model prediction performance of XGBoost, more accurate and efficient predictions of complex data sets are achieved;

[0091] 6.CNN+RNN

[0092] DeepHF: Use CNN+RNN to predict the gRNA activity of CRISPR / Cas9 system and improve the prediction accuracy by integrating bioinformatics features;

[0093] AttCRISPR: Combines CNN and bidirectional gated recurrent unit network (BGRU) to predict the sgRNA activity of the CRISPR / Cas9 system, and further improves the prediction performance of the model through transfer learning methods.

[0094] (3) Data visualization and result presentation

[0095] 1. Provide an intuitive interface to display the results of gene editing analysis, where the tabular data includes sequence pair information, GC content, number of mismatches, on-target score, off-target ratio, etc.; the visualization tool Circos diagram displays detailed target sites and off-target sites, and the thickness of the line or the depth of the color indicates the frequency of off-target events or editing efficiency; the bar chart / line chart displays the editing efficiency of the target site, the length of the bar indicates the percentage of editing or the number of editing events, and the discount shows the trend of on-target efficiency at different time points or under different experimental conditions;

[0096] 2. Supports exporting visualization results as image files or data files for user convenience.

[0097] Second, as Figure 2 As shown, the present invention also provides an interactive CRISPR / Cas9 off-target and on-target analysis method, the interactive CRISPR / Cas9 off-target and on-target analysis method comprises the following steps:

[0098] Step 1: The user logs in to the system platform and obtains the user's login credential information;

[0099] Step 2: Select the platform data set or user uploaded data. If the user uploads data, you can choose whether to share the data;

[0100] Step 3: Provide users with adaptive training options, personalized parameter settings and optimization services;

[0101] Step 4: Allow the user to select a suitable bioinformatics analysis tool from the system;

[0102] Step 5: Provide a user interface to achieve dynamic interactive analysis of data, through which users can interact and analyze data with the system in real time;

[0103] Step 6. Users can view the analysis results, and these results can be saved and integrated into the user's personal analysis report, which will show detailed analysis results.

[0104] In summary, the present invention provides an interactive CRISPR / Cas9 off-target and on-target analysis system and analysis method, aiming to make up for the shortcomings of existing tools. The system integrates a variety of bioinformatics analysis tools and machine learning algorithms to provide a user-friendly interactive platform, obtain off-target and on-target analysis results in real time, and can be easily used by non-professionals. By integrating on-target and off-target analysis, researchers can better control the gene editing process, reduce unexpected side effects, and promote the development and application of gene editing technology.

[0105] The above generally describes the present invention in detail, but it is obvious to a person skilled in the art that some modifications or improvements can be made to the present invention. Therefore, modifications or improvements that do not depart from the spirit of the present invention are within the scope of protection of the present invention.

Claims

1. An interactive CRISPR / Cas9 off-target and on-target analysis system, characterized by: The interactive CRISPR / Cas9 off-target and on-target analysis system includes: a user login management module, a data management module, a resource sharing module, an adaptive training module and a dynamic interactive analysis module; The user login management module includes a task viewing function and a data upload record viewing function, wherein the data upload record viewing function allows the user to view the historical records of uploaded data in the system background; The data management module is used to allow users to upload data, receive uploaded genome data, and provide real-time upload progress feedback during the upload process.

2. An interactive CRISPR / Cas9 off-target and on-target analysis system according to claim 1, characterized in that: The resource sharing module provides sharable data and allows users to set their own data to be shared. The data format is .fasta.

3. An interactive CRISPR / Cas9 off-target and on-target analysis system according to claim 1, characterized in that: The adaptive training module is automatically adjusted according to real-time feedback during model training to improve training efficiency and model performance, including using validation set performance to adjust regularization, adaptive data weighting, and adaptive learning rate algorithms AdaGrad, RMSprop, and Adam.

4. An interactive CRISPR / Cas9 off-target and on-target analysis system according to claim 3, characterized in that: The AdaGrad is an adaptive learning rate algorithm that adjusts the learning rate of each parameter by accumulating historical gradient information; where g t is the gradient of the tth iteration, r t is the cumulative sum of squares of the gradients, and ε is a small number to prevent division by zero.

5. An interactive CRISPR / Cas9 off-target and on-target analysis system according to claim 3, characterized in that: The RMSProp is an improved version of AdaGrad, which solves the problem that the AdaGrad learning rate may become too small by using exponentially weighted moving average to calculate the square gradient; Where β is a constant between 0 and 1 that controls the decay rate of the moving average.

6. An interactive CRISPR / Cas9 off-target and on-target analysis system according to claim 3, characterized in that: The Adam combines the ideas of RMSProp and Momentum to dynamically adjust the learning rate of each parameter by calculating the first-order moment estimate (mean) and second-order moment estimate (uncentered variance) of the gradient. m t =β1×m t-1 +(1-β1)×g t Where m t and v t are the estimates of the first-order moment and the second-order moment, respectively, and β1 and β2 are the exponential decay factors for estimating these two moments.

7. An interactive CRISPR / Cas9 off-target and on-target analysis system according to claim 1, characterized in that: The dynamic interactive analysis module is the core component of the platform. By integrating multiple bioinformatics analysis tools and online interactive analysis tools, it provides users with a powerful and user-friendly analysis environment, mainly including off-target effect analysis models, on-target effect analysis models, and data visualization and result display.

8. An interactive CRISPR / Cas9 off-target and on-target analysis method, based on an interactive CRISPR / Cas9 off-target and on-target analysis system according to any one of claims 1 to 7, characterized in that: The interactive CRISPR / Cas9 off-target and on-target analysis method comprises the following steps: Step 1: The user logs in to the system platform and obtains the user's login credential information; Step 2: Select the platform data set or user uploaded data. If the user uploads data, you can choose whether to share the data; Step 3: Provide users with adaptive training options, personalized parameter settings and optimization services; Step 4: Allow the user to select a suitable bioinformatics analysis tool from the system; Step 5: Provide a user interface to achieve dynamic interactive analysis of data, through which users can interact and analyze data with the system in real time; Step 6. Users can view the analysis results, and these results can be saved and integrated into the user's personal analysis report, which will show detailed analysis results.

Images