Artificial intelligence-based guided editing optimal design method and system

Abstract

The present application relates to an artificial intelligence-based guided editing optimal design method and system, the method comprising: obtaining all candidate pegRNAs; predicting the editing efficiency of all candidate pegRNAs using a trained editing efficiency artificial intelligence prediction system, and ranking them from high to low according to the editing efficiency, and selecting the top N pegRNAs with the highest editing efficiency as the optimal pegRNAs; finding all sgRNAs paired with the optimal pegRNAs according to the preset conditions, and determining the optimal sgRNA according to the sgRNA priority order. The artificial intelligence model trained by the present application can automatically design the optimal pegRNA and sgRNA from among the countless possible structures based on the accurate prediction of the guided editing efficiency using the latest artificial intelligence technology, thereby greatly improving the editing efficiency and achieving fast, cheap and accurate pegRNA and sgRNA design.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of gene editing, and particularly relates to a guide editing optimal design method and system based on artificial intelligence. BACKGROUND

[0002] Prime editing (PE) is a new precise gene editing technology proposed in 2019. With the help of a designed prime editing guide RNA (pegRNA), PE can randomly realize base substitution, base insertion and base deletion, so PE is a versatile and precise gene editing technology. The technology uses a catalytically impaired Cas9 endonuclease fused to an engineered reverse transcriptase, and pegRNA obtained by modification of sgRNA is used for programming to directly write new genetic information into a specific DNA site. The pegRNA specifies the target site of the DNA and encodes the required gene editing. The pegRNA is composed of three parts: a spacer, a primer binding site (PBS) and a reverse transcription template (RTT). The spacer specifically recognizes and binds to the target sequence of the non-editing strand of the DNA by completely matching the DNA protospacer, the PBS is complementary to the editing strand of the DNA, and the RTT encodes the embedded arbitrary editing that people want to realize, including base substitution, base insertion and base deletion. PE includes PE1, PE2 and PE3. PE3 is introduced on the basis of PE2 to cut the non-editing strand of DNA with a second sgRNA to improve the editing efficiency. PE does not require a double-stranded break or a donor DNA template, and can realize arbitrary substitution of 4 bases with higher efficiency, fewer byproducts and off-target effects, making up for the shortcomings of single-base editing. In addition, PE can also realize the precise insertion and deletion of the base at the target site, greatly expanding the range and ability of gene editing.

[0003] Since point mutations, insertion mutations and deletion mutations of bases cover most of human pathogenic genetic variations, prime editing has great potential in clinical gene therapy research, and in principle can correct up to 89% of known human disease-related gene mutations. However, the research of prime editing is still in its infancy, the influencing factors of editing efficiency are unknown, the editing efficiency depends on pegRNA and sgRNA, and it is difficult to effectively predict by computational model, and the design of pegRNA and sgRNA still relies on experimental attempts, which is extremely time-consuming and laborious, and it is difficult to find the optimal pegRNA and sgRNA through experimental methods to make the efficiency of prime editing the highest, which greatly limits the application of prime editing technology. In addition, the design of pegRNA must be customized for different types and positions of prime editing, which is a complex and time-consuming task. Recently, two web tools named pegFinder and PrimeDesign have been developed by different research groups to provide suggestions for the design of pegRNA. However, since they are both based on artificial prior knowledge and the rules are flexible, the incomplete understanding of prime editing leads to low efficiency of pegRNA selection and design. SUMMARY

[0004] In view of the above problems, the purpose of the present application is to provide an artificial intelligence-based prime editing optimal design method and system, which uses the latest artificial intelligence technology to accurately predict the efficiency of prime editing and construct an optimal prime editing design model, and can automatically design the optimal pegRNA and sgRNA from the countless possible structures, thereby greatly improving the editing efficiency and realizing fast, cheap and accurate pegRNA and sgRNA design, which is of great significance for the application of future prime editing, and also helps the development of multiple treatment fields including human genetic diseases, tumors and AIDS, provides a reference for the gene therapy and drug development of diseases, and will also promote the rapid development of individualized precision medicine.

[0005] To achieve the above purpose, the present application adopts the following technical solutions: the present application discloses an artificial intelligence-based prime editing optimal design method, which comprises the following steps:

[0006] The trained editing efficiency artificial intelligence prediction system is used to predict the editing efficiency of all candidate pegRNAs, and the pegRNAs are sorted from high to low according to the editing efficiency, and the top N pegRNAs with the highest editing efficiency are selected as the optimal pegRNAs;

[0007] According to the preset conditions, all sgRNAs paired with the optimal pegRNAs are found, and the optimal sgRNA is determined according to the sgRNA priority order.

[0008] Preferably, the finding all sgRNAs paired with the optimal pegRNA according to the preset condition and determining the optimal sgRNA according to the sgRNA priority order specifically comprises:

[0009] According to the distance from the sgRNA cutting site to the pegRNA cutting site being 0-100 bases, all sgRNAs paired with the optimal pegRNA are found, and the optimal sgRNA is determined according to the sgRNA priority order.

[0010] Preferably, the obtaining all candidate pegRNAs specifically comprises:

[0011] According to condition one and condition two, all possible candidate PAM sequences are scanned, and then a sequence with a length of 20 bases upstream adjacent to the PAM sequence is taken as the Spacer.

[0012] According to condition three and condition four, the 3' extension sequence PBS sequence and RTT sequence of all candidate pegRNAs meeting the conditions are designed, so as to obtain all candidate pegRNAs capable of realizing the intended editing.

[0013] Preferably, the condition one specifically comprises:

[0014] The distance from the pegRNA cutting site to the target editing site is set to be no more than 50 bases.

[0015] Preferably, the condition two specifically comprises:

[0016] The type of the PAM sequence is NGG or NG.

[0017] Preferably, the condition three specifically comprises:

[0018] The 3' end of the RTT sequence is located at a position at least 5 bases downstream of the 3' end of the target editing sequence.

[0019] Preferably, the condition four specifically comprises:

[0020] Since the lengths of the PBS and RTT sequences are variable, the minimum lengths of the PBS and RTT are both set to be 8 bases according to the experiment setting, and the maximum lengths of the PBS and RTT are respectively set to be 18 and 68 bases.

[0021] The second object of the application can be achieved by adopting the following technical scheme: an artificial intelligence-based guided editing optimal design system, comprising: an obtaining module, which obtains all candidate pegRNAs;

[0022] A prediction module is configured to predict the editing efficiency of all candidate pegRNAs by using a trained editing efficiency artificial intelligence prediction system, and sort the pegRNAs according to the editing efficiency from high to low, and select the top N pegRNAs with the highest editing efficiency as the optimal pegRNAs.

[0023] A sorting module is configured to find all sgRNAs paired with the optimal pegRNAs according to a preset condition, and determine the optimal sgRNA according to the sgRNA priority order.

[0024] The third object of the present application can be achieved by adopting the following technical solution:

[0025] A computer device comprises a processor and a memory for storing programs executable by the processor, and the processor implements the above-mentioned artificial intelligence-based guided editing optimal design method when executing the programs stored in the memory.

[0026] The fourth object of the present application can be achieved by adopting the following technical solution:

[0027] A storage medium stores a program, and the program is executed by a processor to implement the above-mentioned artificial intelligence-based guided editing optimal design method.

[0028] The present application obtains all candidate pegRNAs, predicts the editing efficiency of all candidate pegRNAs by using a trained editing efficiency artificial intelligence prediction system, sorts the pegRNAs according to the editing efficiency from high to low, selects the top N pegRNAs with the highest editing efficiency as the optimal pegRNAs, finds all sgRNAs paired with the optimal pegRNAs according to a preset condition, and determines the optimal sgRNA according to the sgRNA priority order, establishes and develops a pegRNA editing efficiency prediction algorithm based on deep learning, and the trained artificial intelligence model can predict the efficiency of any editing without bias and accurately, and has certain interpretability for the results, thereby effectively predicting the editing efficiency of different pegRNAs, and based on the artificial intelligence prediction system, further developing a system capable of automatically designing the optimal pegRNA, thereby quickly and accurately obtaining the optimal design from numerous possible structures to achieve the highest editing efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0029] Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The accompanying drawings are included to provide a description of preferred embodiments, and are not meant to limit the present application. Throughout the drawings, the same reference numerals are used for the same components. In the drawings:

[0030] Fig. 1 A flowchart of the artificial intelligence-based guided editing optimal design method of the present application;

[0031] Fig. 2 A schematic diagram of the optimal design method for guided editing based on artificial intelligence of the present application. DETAILED DESCRIPTION

[0032] Exemplary embodiments of the present application will be described in greater detail below with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it is understood that the present application can be embodied in various forms without being limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present application will be thoroughly and completely comprehended, and so that the scope of the present application will be completely conveyed to those skilled in the art.

[0033] The prior art has low model accuracy and poor model generalization. The model is biased and has low accuracy. The model is a black box and has no explainability. The present application obtains all candidate pegRNAs, predicts the editing efficiency of all candidate pegRNAs using a trained editing efficiency artificial intelligence prediction system, sorts the pegRNAs according to the editing efficiency from high to low, selects the top N pegRNAs with the highest editing efficiency as the optimal pegRNAs, finds all sgRNAs paired with the optimal pegRNAs according to a preset condition, determines the optimal sgRNA according to the sgRNA priority order, develops a pegRNA editing efficiency prediction algorithm based on deep learning, and trains an artificial intelligence model that can predict the efficiency of any editing without bias and accurately, and has a certain explainability for the results, thereby effectively predicting the editing efficiency of different pegRNAs. Based on the artificial intelligence prediction system, an optimal pegRNA system that can automatically design the optimal pegRNA is further developed, so that the optimal design is quickly and accurately obtained from the numerous possible structures to achieve the highest editing efficiency.

[0034] Example 1:

[0035] The present application discloses an optimal design method for guided editing based on artificial intelligence, which is described with reference to Figs. 1-2 The method comprises the following steps: step 100, obtaining all candidate pegRNAs;

[0036] Step 200, using a trained editing efficiency artificial intelligence prediction system to predict the editing efficiency of all candidate pegRNAs, and sorting the pegRNAs according to the editing efficiency from high to low, selecting the top N pegRNAs with the highest editing efficiency as the optimal pegRNAs;

[0037] Step 300, finding all sgRNAs paired with the optimal pegRNAs according to a preset condition, and determining the optimal sgRNA according to the sgRNA priority order.

[0038] The application obtains all candidate pegRNAs, predicts the editing efficiency of all candidate pegRNAs by using the trained editing efficiency artificial intelligence prediction system, and selects the top N pegRNAs with the highest editing efficiency as the optimal pegRNAs according to the descending order of the editing efficiency; all sgRNAs paired with the optimal pegRNAs are found according to the preset conditions, and the optimal sgRNA is determined according to the sgRNA priority order, a pegRNA editing efficiency prediction algorithm based on deep learning is established and developed, the trained artificial intelligence model can predict the efficiency of any editing without bias and accurately, and the result has certain interpretability, so as to effectively predict the editing efficiency of different pegRNAs, and based on the artificial intelligence prediction system, an optimal pegRNA system capable of automatically designing is further developed, so that the optimal design is quickly and accurately obtained from numerous possible structures to make the editing efficiency the highest.

[0039] Preferably, in step 300, all sgRNAs paired with the optimal pegRNA are found according to the preset conditions, and the optimal sgRNA is determined according to the sgRNA priority order, specifically comprising:

[0040] According to the distance from the sgRNA cutting site to the pegRNA cutting site, which is 0-100 bases, all sgRNAs paired with the optimal pegRNA are found, and the optimal sgRNA is determined according to the sgRNA priority order.

[0041] Preferably, in step 100, all candidate pegRNAs are obtained; specifically comprising:

[0042] According to condition one and condition two, all possible candidate PAM sequences are scanned, and then the sequence with a length of 20 bases upstream of the PAM sequence is taken as the Spacer;

[0043] According to condition three and condition four, the 3' extension sequence PBS sequence and RTT sequence of all candidate pegRNAs meeting the conditions are designed, so as to obtain all candidate pegRNAs capable of realizing the expected editing.

[0044] Preferably, the condition one specifically comprises:

[0045] The distance from the pegRNA cutting site to the target editing site is set to be not more than 50 bases.

[0046] Preferably, the condition two specifically comprises:

[0047] The type of PAM sequence is NGG or NG.

[0048] Preferably, the condition three specifically comprises:

[0049] the 3' end of the target edit sequence.

[0050] Preferably, the fourth condition; specifically includes:

[0051] Since the lengths of PBS and RTT sequences are variable, the minimum length of PBS and RTT is set to 8 bases, and the maximum length of PBS and RTT is set to 18 and 68 bases respectively according to experimental settings.

[0052] Specifically, for any intended edit, the optimal design system searches for all candidate pegRNAs and sgRNAs that can achieve the intended edit by the following conditions (the user can modify the default hyperparameters). Condition one: the distance between the pegRNA cleavage site and the target edit site is set to no more than 50 bases; Condition two: the type of PAM (Protospacer Adjacent Motif) sequence is NGG or NG; Condition three: the 3' end of the RTT sequence is at least 5 bases downstream of the 3' end of the target edit sequence, otherwise the distance is too close to have biological function; Condition four: since the lengths of PBS and RTT sequences are variable, the minimum length of PBS and RTT is set to 8 bases, and the maximum length of PBS and RTT is set to 18 and 68 bases respectively according to experimental settings; Condition five: the distance between the sgRNA cleavage site and the pegRNA cleavage site is at least 0 bases and at most 100 bases.

[0053] It should be noted that in each strand of a DNA molecule, the side with the phosphate group belongs to the 5' end, and in a deoxyribonucleotide, the phosphate is connected to the 5th carbon atom of the deoxyribose; while the -OH end is the 3' end, because the hydroxyl group is connected to the 3rd carbon atom of the deoxyribose.

[0054] The optimal design system first scans all possible candidate PAM sequences according to condition one and condition two, and then takes the sequence of 20 bases in length adjacent upstream of the PAM sequence as the Spacer.

[0055] Secondly, according to condition three and condition four, the 3' extension sequence (PBS and RTT sequence) of all candidate pegRNAs that meet the conditions is designed, thereby obtaining all candidate pegRNAs that can achieve the intended edit.

[0056] Next, the trained artificial intelligence prediction system for editing efficiency is used to predict the editing efficiency of all candidate pegRNAs, and the top 10 pegRNAs with the highest editing efficiency are selected as the optimal pegRNAs according to the editing efficiency from high to low.

[0057] Then, according to condition five, all sgRNAs paired with the pegRNA were found and the optimal sgRNA was determined according to the sgRNA priority order: 1) PE3b seed annotation 2) PE3b non-seed annotation 3) PE3 annotation 75 bases away from the Spacer of the pegRNA.

[0058] Finally, the optimal 10 pegRNAs and sgRNAs were the final optimal design of the system and the comprehensive information of the design was given.

[0059] To illustrate the wide use of the optimal design system, we extracted and screened 77,738 human pathogenic gene mutations from the ClinVar database and used the system to design optimal pegRNAs and sgRNAs to correct or install these pathogenic mutations. Among these pathogenic variants, the system designed 778,791,772 candidate pegRNAs to correct pathogenic mutations. On average, 10,018 candidate pegRNAs were designed for each pathogenic variant, reflecting the complexity of pegRNA selection, and then the system began to determine the optimal pegRNAs and sgRNAs design for each editing event from a large number of candidate designs. In total, the system designed 2 billion candidate pegRNAs and sgRNAs and stored the optimal 10 pegRNAs and sgRNAs for each editing event. To make the optimal pegRNAs and sgRNAs design of the system more accessible, we constructed OPEDVar (http: / / oped.bioinfotech.org / OPEDVar / ), a comprehensive and searchable database containing more than 77,000 pathogenic human genetic variants.

[0060] In short, the optimal design system is easy to access and user-friendly, and can be used to automatically design pegRNAs and sgRNAs for any intended editing and provide the system design results and various related information for a given targeting sequence or a given targeting site location. When the user inputs the targeting sequence or the location of the targeting site and the intended editing, the system determines all possible candidate designs for the intended editing according to the parameters set by the user, and then provides the corresponding number of optimal designs by ranking the predicted editing efficiency. On the other hand, all human pathogenic genetic variants are extracted and screened from the human genetic variant database, and the editing efficiency of all candidate designs is predicted using the system, from which the optimal design is selected. The system constructs a database of optimal guide editing designs for all human pathogenic genetic variants, which is a comprehensive searchable database for providing optimal guide editing designs for correcting these human pathogenic mutations, so as to make the mutant genes become normal genes, and help the development of multiple treatment fields including human genetic diseases, tumors, AIDS, etc.

[0061] Embodiment 2:

[0062] The embodiment provides a guided editing efficiency prediction system based on deep learning and transfer learning, which comprises an acquisition module that acquires all candidate pegRNAs;

[0063] A prediction module is configured to predict the editing efficiency of all candidate pegRNAs by using a trained editing efficiency artificial intelligence prediction system, and sort the pegRNAs according to the editing efficiency from high to low, and select the top N pegRNAs with the highest editing efficiency as optimal pegRNAs.

[0064] A sorting module is configured to find all sgRNAs paired with the optimal pegRNAs according to a preset condition, and determine the optimal sgRNA according to the sgRNA priority order.

[0065] The embodiment acquires all candidate pegRNAs, predicts the editing efficiency of all candidate pegRNAs by using a trained editing efficiency artificial intelligence prediction system, sorts the pegRNAs according to the editing efficiency from high to low, selects the top N pegRNAs with the highest editing efficiency as optimal pegRNAs, finds all sgRNAs paired with the optimal pegRNAs according to a preset condition, and determines the optimal sgRNA according to the sgRNA priority order, establishes and develops a pegRNA editing efficiency prediction algorithm based on deep learning, and the trained artificial intelligence model can predict the efficiency of any editing without bias and accurately, and has certain interpretability for the result, so as to effectively predict the editing efficiency of different pegRNAs, and based on the artificial intelligence prediction system, further develop a system capable of automatically designing optimal pegRNAs, so as to quickly and accurately obtain the optimal design from numerous possible structures to make the editing efficiency the highest.

[0066] The specific implementation of each module in the embodiment can be referred to the above-described embodiment 1, which will not be repeated here. It should be noted that the device provided in the above-described embodiment is only used as an example for the division of the above-described functional modules, and in actual application, the above-described functions can be completed by different functional modules according to needs, that is, the internal structure is divided into different functional modules to complete all or part of the functions described above.

[0067] It can be understood that the terms "first", "second" and the like used in the device of the above-described embodiment can be used to describe various units, but these units are not limited by these terms. These terms are only used to distinguish a first module from another module. For example, without departing from the scope of the present application, the first sending module can be referred to as a second sending module, and similarly, the second sending module can be referred to as a first sending module, the first sending module and the second sending module are both sending modules, but they are not the same sending module.

[0068] Embodiment 3:

[0069] The embodiment provides a computer device, including a processor and a memory for storing a program executable by the processor, when the processor executes the program stored by the memory, the above-mentioned optimal design method of guided editing based on artificial intelligence is realized.

[0070] Embodiment 4:

[0071] The embodiment provides a storage medium, which is a computer readable storage medium, and stores a computer program, when the program is executed by a processor, the processor executes the computer program stored by the memory, and the optimal design method of guided editing based on artificial intelligence in the embodiment 1 is realized.

[0072] It should be noted that the computer readable storage medium of the embodiment can be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples of computer readable storage medium can include, but are not limited to, electrical connections with one or more conductive wires, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0073] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An AI-based guided editing optimal design method, characterized in that, The method includes: Obtain all candidate pegRNAs; The editing efficiency of all candidate pegRNAs is predicted using a pre-trained AI prediction system, and they are sorted from high to low according to their editing efficiency. The top N pegRNAs with the highest editing efficiency are selected as the optimal pegRNAs. Find all sgRNAs that pair with the optimal pegRNA according to preset conditions, and determine the optimal sgRNA according to the priority order of sgRNAs; The process of finding all sgRNAs that pair with the optimal pegRNA according to preset conditions and determining the optimal sgRNA according to the priority order of sgRNAs specifically includes: finding all sgRNAs that pair with the optimal pegRNA according to the distance from the sgRNA cleavage site to the pegRNA cleavage site, which is at least 0 bases and at most 100 bases, and determining the optimal sgRNA according to the priority order of sgRNAs. The process of obtaining all candidate pegRNAs specifically includes: scanning to find all possible candidate PAM sequences according to conditions one and two, and then taking the sequence 20 bases upstream of the PAM sequence as a spacer; designing the 3' extension sequences PBS and RTT sequences of all candidate pegRNAs that meet the conditions according to conditions three and four, thereby obtaining all candidate pegRNAs that can achieve the expected editing. Condition one specifically includes: the distance from the pegRNA cleavage site to the target editing site is set to no more than 50 bases; The second condition specifically includes: the type of the PAM sequence is NGG or NG; The third condition specifically includes: the 3' end of the RTT sequence is located at least 5 bases downstream of the 3' end of the target edit sequence; Condition four specifically includes: since the lengths of PBS and RTT sequences are variable, the minimum length of both PBS and RTT is set to 8 bases according to the experimental settings, while the maximum lengths of both PBS and RTT are set to 18 and 68 bases respectively.

2. An AI-based guided editing optimal design system, used to implement the AI-based guided editing optimal design method as described in claim 1, characterized in that, The system includes: The module retrieves all candidate pegRNAs. Prediction module: Utilizes a pre-trained AI prediction system for editing efficiency to predict the editing efficiency of all candidate pegRNAs, sorts them from high to low based on editing efficiency, and selects the top N pegRNAs with the highest editing efficiency as the optimal pegRNAs; The sorting module is used to find all sgRNAs that pair with the optimal pegRNA according to preset conditions, and to determine the optimal sgRNA according to the priority order of sgRNAs.

3. A computer device, characterized in that, The computer device includes a processor and a memory for storing a processor-executable program, wherein when the processor executes the program stored in the memory, the computer device performs the method as described in claim 1.

4. A storage medium, characterized in that, A stored program, which, when executed by a processor, performs the method of claim 1.

Images