Large-scale primer pool screening system and method based on parallel computing and multi-objective optimization and storage medium

Abstract

The invention discloses a large-scale primer pool screening system and method based on parallel computing and multi-objective optimization, and belongs to the technical field of bioinformatics and molecular biology. The system comprises a candidate primer generation module, a high-speed biological characteristic analysis engine and a multi-target collaborative optimization module. The method comprises the following steps: generating an initial candidate primer pool; synchronous and efficient evaluation of specificity and dimer risk is carried out on the candidate primers by using rapid screening based on k-mer index and GPU parallel calculation; and inputting an evaluation result into an optimized multi-objective optimization algorithm for global collaborative optimization, and generating a Pareto optimal solution set which is balanced among a plurality of objectives such as maximization of target coverage, minimization of primer quantity, minimization of non-specific risk and minimization of dimer risk. According to the method, the traditional several-day-order analysis optimization process is shortened to several hours and even several minutes, normal form transformation from single-point evaluation to global intelligent optimization is achieved, and the efficiency, quality and flexibility of large-scale primer pool design are remarkably improved.

Description

Technical Field

[0001] This invention relates to the interdisciplinary field of bioinformatics and molecular biology, specifically to a large-scale primer pool screening system, method, and storage medium based on parallel computing and multi-objective optimization. Background Technology

[0002] With the rapid development of high-throughput sequencing technology, technologies such as multiplex PCR and targeted sequencing have been widely used in precision medicine fields such as pathogen detection, genetic disease screening, and tumor genotyping. These applications typically require the simultaneous and efficient amplification of tens of thousands of dispersed genomic targets, thus creating an urgent need for the efficient design and optimization of large-scale primer pools (usually containing hundreds to thousands of primer pairs).

[0003] Currently, several mature tools exist in the field of primer design, such as Primer3 and Primer-BLAST, which provide a reliable foundation for the design of single or small primer pairs. These tools primarily rely on the fundamental physicochemical properties of the primers themselves (such as melting temperature Tm, GC content, length, and terminal stability) and their specificity against specific targets for screening. However, when these traditional tools are applied to the design of large-scale primer pools, their inherent technical architecture and algorithmic strategies reveal the following serious bottlenecks and shortcomings: Homology / specificity screening is inefficient: Ensuring primer specificity (i.e., no significant similarity to non-target genomic regions) is crucial to avoiding off-target amplification. Traditional methods rely on sequence alignment tools such as BLAST to align each candidate primer one-to-one with the entire reference genome (e.g., human hg38, approximately 3 billion bases). For candidate pools containing thousands of primers, this genome-wide alignment is computationally intensive, typically taking hours to days, severely hindering rapid iteration and optimization processes.

[0004] The complexity of primer interaction assessment explodes: In multiplex reactions, primers may form stable dimers (including self-dimers and cross-dimers), severely impacting amplification efficiency. Assessing the dimer risk among all possible primer pairs in a primer pool has a computational time complexity of O(n²). For a candidate pool of dry weight primers (i.e., thousands of primers), millions or even tens of millions of combinations need to be evaluated. Using traditional sequence alignment and free energy calculation methods, this is almost impossible to complete within a feasible timeframe on conventional computing hardware, becoming a major computational bottleneck for large-scale design.

[0005] Lack of global, multi-objective collaborative optimization capabilities: Existing tools are essentially "filters," whose main function is to determine whether a single primer or primer pair passes or fails based on a series of fixed thresholds. They lack the ability to collaboratively optimize conflicting design goals from a global perspective. For example, designers often aim to achieve the best balance among several objectives: maximizing target coverage to ensure detection integrity, minimizing the total number of primers to reduce reagent costs and experimental complexity, minimizing overall nonspecific risk to improve detection accuracy, and minimizing the risk of primer dimerization to ensure amplification efficiency. These objectives are interdependent, and existing tools cannot automatically provide a Pareto optimal solution set that intelligently balances these objectives. Users can only rely on experience to manually adjust parameters and conduct multiple trial and error processes, which is cumbersome and makes it difficult to find a globally optimal solution.

[0006] The optimization strategies are rigid and lack scenario adaptability: different application scenarios have different performance requirements for primer pools (for example, cost-sensitive large-scale screening projects may focus more on the number of primers, while clinical diagnosis has extreme requirements for coverage and specificity). The filtering thresholds and rules of existing tools are usually fixed or semi-fixed, which cannot easily adjust the weight or priority of each objective in the optimization strategy according to the core needs of different application scenarios.

[0007] In summary, existing technologies have significant shortcomings in terms of computational efficiency, global optimization capability, and design process flexibility when dealing with large-scale primer pool design problems. Therefore, there is an urgent need in this field for an innovative technical solution that combines efficient bioinformatics analysis algorithms, powerful parallel computing capabilities, and intelligent multi-objective decision optimization models to provide a fast, automated, globally optimized, and customizable large-scale primer pool comprehensive screening system and method. Summary of the Invention

[0008] The purpose of this invention is to provide a large-scale primer pool screening system, method, and storage medium based on parallel computing and multi-objective optimization, so as to solve the problem that the prior art mentioned in the background art has significant deficiencies in terms of computational efficiency, global optimization capability, and design process flexibility when dealing with large-scale primer pool design problems.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A large-scale primer pool screening system based on parallel computing and multi-objective optimization includes: The candidate primer generation module is used to generate multiple candidate primers for each target based on the input target sequence set, forming an initial candidate primer pool; A high-speed biological characteristic analysis engine, connected to the candidate primer generation module, is used for rapid evaluation of primers and primer pairs in the initial candidate primer pool. The high-speed biological characteristic analysis engine includes: The specific rapid screening submodule, based on a pre-built reference genome k-mer index database, performs rapid calculation and screening of non-target region similarity for candidate primers; The dimer rapid prediction submodule employs GPU-accelerated or large-scale multi-threaded parallel computing methods to rapidly assess the dimer formation risk of candidate primer pairs. A multi-objective collaborative optimization module, connected to the high-speed biological characteristic analysis engine, is used to receive scoring data from the high-speed biological characteristic analysis engine, model the primer selection problem as a multi-objective optimization problem, and solve it using an optimization algorithm, outputting a Pareto optimal solution set that achieves a balance among multiple optimization objectives, wherein the optimization objectives include at least: maximizing target coverage, minimizing the total number of selected primers, minimizing the overall nonspecific risk, and minimizing the overall dimer formation risk.

[0010] Furthermore, in the specific rapid screening submodule, when constructing the k-mer index database, a sliding window is used to traverse the reference genome, and a unique k-mer filtering mechanism is introduced to retain only k-mer sequences that appear once in the reference genome.

[0011] Furthermore, the specific rapid screening submodule supports multi-scale k-mer analysis, which can dynamically select the k value according to the application scenario. Specifically, k=15 is used for rapid screening, k=18 is used for standard analysis, and k=21 is used for high-strictness scenarios.

[0012] Furthermore, the GPU parallel computing architecture of the dimer fast prediction submodule adopts a two-dimensional grid thread allocation strategy, with each thread responsible for computing a specific primer pair combination. It also introduces a "seed expansion" heuristic strategy, first using short sequence complete complementary matching as the initial screening condition, and then performing full sequence free energy calculation on the primer pairs that pass the initial screening.

[0013] Furthermore, the multi-objective collaborative optimization module uses the parameter-optimized NSGA-II algorithm, with a population size of 200 and an early stopping mechanism that automatically terminates the iteration when the Pareto front shows no significant improvement for several consecutive generations.

[0014] Furthermore, the multi-objective collaborative optimization module allows users to dynamically adjust the weights of each optimization objective through an interactive interface to adapt to the priority requirements of different application scenarios.

[0015] A large-scale primer pool screening method based on parallel computing and multi-objective optimization includes the following steps: Step S1: Input the target region sequence set and user configuration parameters; Step S2: Generate an initial candidate primer pool based on the target region sequence set; Step S3: Perform high-speed biological characteristic analysis in parallel, the analysis including: Step S31: Based on the pre-constructed reference genome k-mer index database, perform rapid and specific screening of all primers in the initial candidate primer pool; Step S32: Using a parallel computing method, perform a rapid dimer risk assessment on all primer pairs in the initial candidate primer pool; Step S4: Based on the results of the high-speed biological characteristic analysis described in Step S3, score each primer and primer pair; Step S5: Input the scoring data into a multi-objective optimization algorithm for iterative calculation to simultaneously optimize multiple objectives and find the Pareto optimal frontier. The multiple objectives include at least: maximizing target coverage, minimizing primer number, minimizing nonspecific risk, and minimizing dimer risk. Step S6: Output and visualize the Pareto optimal solution set; Step S7: Based on the user's selection, export the final list of selected primer sequences and a performance report.

[0016] Furthermore, in the parallel execution of high-speed biological characterization steps, when assessing the dimer risk of primer pairs, the system first searches for perfectly complementary seed pairs that meet the set length in parallel on the GPU, and then performs precise free energy calculations only on these seed pairs.

[0017] Furthermore, in the step of outputting and visualizing the Pareto optimal solution set, the Pareto front is displayed in the form of a two-dimensional scatter plot, where the horizontal axis represents the objective to be minimized and the vertical axis represents the objective to be maximized. Each scatter point represents a Pareto optimal solution scheme, and typical schemes are marked for users to select.

[0018] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a large-scale primer pool screening method based on parallel computing and multi-objective optimization.

[0019] Compared with the prior art, the beneficial effects of the present invention are: (1) Significantly improve computational efficiency: By using fast screening based on k-mer index and GPU parallel computing architecture, the time for homology analysis and dimer evaluation of primer pools is shortened from several days in traditional methods to within several hours; (2) Achieve multi-objective collaborative optimization: The primer screening problem is constructed as a multi-objective optimization model including coverage, primer quantity, specificity risk and dimer risk. An improved evolutionary algorithm is used to automatically output the Pareto optimal solution set, supporting systematic trade-offs among multiple mutually constraining objectives. (3) Enhance design flexibility and adaptability: The system allows users to dynamically adjust the weights and analysis parameters of each optimization objective according to specific application needs, and can adapt to the design requirements of different scenarios, from high-throughput screening to clinical diagnosis; (4) Provide clear decision-making basis: By visualizing the Pareto front, the trade-off between different optimization schemes is shown, providing intuitive data support for users to select the final primer pool configuration. Attached Figure Description

[0020] Figure 1 This is a diagram illustrating the overall architecture of the large-scale primer pool screening system based on parallel computing and multi-objective optimization of this invention. Figure 2 This is a flowchart of the large-scale primer pool screening method based on parallel computing and multi-objective optimization of the present invention; Figure 3 This is a schematic diagram of the Pareto front output by the multi-objective optimization module in this invention; Figure 4 The Pareto front plot after selecting the optimal solution in the example; Figure 5 A comparison of system performance metrics before and after optimization; Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figure 1 This invention provides a large-scale primer pool screening system based on parallel computing and multi-objective optimization. For example... Figure 1 As shown, the overall architecture of this system comprises three core modules: a candidate primer generation module, a high-speed biological characteristic analysis engine, and a multi-objective collaborative optimization module. These three modules are connected sequentially to achieve efficient and collaborative optimization of a large-scale primer pool.

[0023] The candidate primer generation module automatically generates multiple candidate primers that conform to basic physicochemical properties for each target sequence based on a user-input target sequence set (e.g., a FASTA format file containing thousands of targets). In practice, this module calls or encapsulates the core algorithm of an existing basic primer design program (such as Primer3) and generates a preset number (e.g., 3-10) of candidate primers for each target based on user-preset parameters (e.g., melting temperature Tm range, GC content range, primer length range, 3-end stability, etc.). The candidate primers for all targets are then aggregated to form an initial candidate primer pool, typically several times the size of the number of targets.

[0024] The high-speed biological characterization engine is connected to the candidate primer generation module and is responsible for rapidly evaluating primers in the initial candidate primer pool and all possible pairwise combinations. It includes two parallel sub-modules: The specific rapid screening submodule's core is rapid homology analysis based on the k-mer algorithm. During system initialization or first use, a k-mer index database of a reference genome (e.g., human hg38) needs to be pre-built. The construction process uses a sliding window (window size equal to the k value) to traverse the reference genome sequence. To improve the specificity and efficiency of screening, a unique k-mer filtering mechanism is introduced. During index construction, only k-mer sequences and their location information that appear exactly once in the entire reference genome are retained, while duplicate k-mers are discarded. This design transforms subsequent homology screening from full sequence alignment to an efficient database query operation. This module supports multi-scale k-mer analysis, allowing dynamic selection of the k value based on different application scenarios: for example, k=15 is chosen in rapid screening mode to prioritize speed; k=18 is chosen in standard design mode to balance speed and accuracy; and k=21 is chosen in clinical diagnostic mode where high specificity is required. During actual screening, the system decomposes each candidate primer sequence into consecutive overlapping k-mers and queries the pre-built index database. A specificity risk score is calculated based on the query results (such as the number of matched k-mers and their location distribution).

[0025] The fast dimer prediction submodule is used to efficiently assess the risk of all primer pairs in the primer pool forming stable dimers. To address the computational complexity of O(n²), this embodiment employs a GPU-accelerated parallel computing architecture. Specifically, the system maps the task of evaluating N primers and all possible N×(N-1) / 2 primer pair combinations to thousands of GPU cores. Its thread allocation strategy uses a two-dimensional grid layout, with each block and thread responsible for calculating the free energy change (ΔG) of one or a specific primer pair combination. To reduce unnecessary precise calculations, this module introduces a "seed expansion" heuristic: First, a rapid initial screening is performed in parallel on the GPU to find all "seed" primer pairs that have a perfectly complementary match of a set length (e.g., 8 bases) in a certain region of the primer sequence. Then, only these "seed pairs" that pass the initial screening and are likely to form stable dimers are subjected to subsequent time-consuming full-sequence free energy (ΔG) precise calculations. This two-stage screening mechanism significantly reduces the computational load. Finally, a dimer risk score is output for each primer pair.

[0026] The multi-objective collaborative optimization module, acting as the system's intelligent decision-making center, receives specificity scores for each primer and dimer scores for each primer pair from the high-speed biological characteristic analysis engine, while also incorporating the primers' own baseline scores (such as Tm value deviation and secondary structure). This module models the problem of "selecting the optimal subset of primers from a large-scale candidate pool" as a multi-objective optimization problem. In this embodiment, the optimization algorithm used is the parameter-tuned NSGA-II (Non-dominated sorting genetic algorithm II).

[0027] Optimization objectives: The algorithm simultaneously optimizes at least four objectives: maximizing the coverage of the target region (i.e., ensuring that as many target sites as possible are effectively covered by at least one pair of primers); minimizing the total number of primers selected in the end (to reduce cost and experimental complexity); minimizing the overall nonspecific risk of the primer pool (based on k-mer screening score); and minimizing the overall dimer formation risk of the primer pool (based on ΔG score).

[0028] Algorithm Implementation: In the actual algorithm execution, the population size is set to 200 individuals, with each individual representing a possible primer pool scheme (i.e., a subset of primers). The algorithm runs iteratively, with a standard setting of 500 generations in this embodiment. To improve efficiency, an early stopping mechanism is introduced: when the algorithm detects that no significant improvement has been observed in the Pareto optimal front for 50 consecutive generations (e.g., the improvement value of the objective function is less than a preset threshold), the iteration automatically terminates. Furthermore, this module provides a user interface, allowing users to dynamically adjust the weights of various optimization objectives. For example, in cost-sensitive projects, the weight of "minimizing the number of primers" can be increased, or in clinical testing, the weights of "maximizing coverage" and "minimizing nonspecific risk" can be increased to adapt to the priorities of different application scenarios.

[0029] Output: The algorithm does not output a single solution, but a Pareto (Pareto) result. Figure 3 The optimal solution set is a set of optimal solutions that cannot be further improved simultaneously among multiple objectives and are not mutually dominant.

[0030] Please see Figure 2 This invention also provides a large-scale primer pool screening method based on parallel computing and multi-objective optimization. This method can be executed by the above-mentioned system and specifically includes the following steps: S1: Input the target region sequence set and user configuration parameters (including primer design basic parameters, optimization target weights, analysis rigor, etc.).

[0031] S2: Based on the input target sequence set, call the candidate primer generation module to generate an initial candidate primer pool.

[0032] S3: Perform high-speed biological characterization in parallel. This step consists of two parallel sub-steps: S31: Rapid Specificity Screening. Utilizing a pre-constructed reference genome index database containing unique k-mers, all primers in the initial candidate pool are rapidly aligned and screened, and a specificity score is provided.

[0033] S32: Fast Dimer Prediction. Utilizing a GPU parallel computing architecture and a "seed expansion" strategy, this method assesses the dimer risk of all primer pairs in the initial candidate pool and provides a dimer score.

[0034] S4: Based on the analysis results of step S3, a comprehensive score is given to each primer (considering its own properties and specificity) and each primer pair (dimer risk).

[0035] S5: Input the scoring data generated in step S4 into the multi-objective collaborative optimization module. This module runs an optimized NSGA-II and other multi-objective optimization algorithms to iteratively search in multiple defined objective spaces to find the Pareto optimal front.

[0036] S6: Output and visualize the Pareto optimal solution set. For example... Figure 3 and Figure 4 As shown, the system visualizes the results in the form of two-dimensional scatter plots. Typically, the horizontal axis (X-axis) represents the target to be minimized (e.g., primer count), and the vertical axis (Y-axis) represents the target to be maximized (e.g., target coverage). Each scatter point in the plot represents a Pareto optimal solution (i.e., a feasible optimized primer pool scheme), and all optimal solutions are connected to form the Pareto front curve. The system can automatically label the front or allow users to interactively select typical schemes, such as "cost-priority scheme," "balanced scheme," and "performance-priority scheme."

[0037] S7: Based on the final solution selected by the user from the Pareto optimal solution set, the system exports the corresponding primer sequence list and a detailed performance report (including indicators such as coverage, number of primers, average specificity score, and maximum dimer risk).

[0038] Taking a human genetic disease detection panel design containing 2000 targets as an example, the specific implementation process of this invention is illustrated as follows: Input: Input 2000 target sequences in FASTA format. User configuration parameters are set as follows: Tm value 60±2℃, primer length 18-25bp, and the optimization strategy is "standard equilibrium".

[0039] Candidate pool generation: The candidate primer generation module calls the Primer3 core algorithm to generate 5 candidate primers for each target, resulting in an initial candidate pool containing 10,000 primers.

[0040] High-speed analysis: The specific screening submodule uses an index database with k=18 to quickly perform human genome (hg38) homology screening on 10,000 primers, identifying and marking primers that are highly similar to non-target regions of the genome.

[0041] The dimer prediction submodule is launched on the GPU. It first searches in parallel for all primer pairs that have 8bp perfectly complementary "seeds", and then performs precise ΔG calculation only on these candidate pairs (about 20% of the total combinations).

[0042] Multi-objective optimization: The scoring data generated in the above steps is fed into the multi-objective collaborative optimization module. The module runs the NSGA-II algorithm with optimized parameters (population 200, maximum iterations 500 generations, early stopping threshold 50 generations), and converges after about 150 generations, obtaining a Pareto optimal solution set containing about 20 non-dominated schemes.

[0043] Results output and selection: such as Figure 4 and Figure 5 As shown, the system displays the Pareto front. Users can observe a clear trade-off on the front: a reduction in the number of primers is often accompanied by a decrease in coverage. Figure 5 The comparison chart shows that the system-recommended "balanced approach" achieves 94.5% target coverage using only 650 primer pairs. Compared to the initial pool (10,000 primers), the number of primers is reduced by 93.5%, while the overall specificity score improves from 0.75 to 0.92, and the dimer risk score improves from 0.65 to 0.88 (higher scores indicate lower risk). After the user selects this approach, the system exports a final list containing 650 primer pairs.

[0044] As can be seen from the above embodiments, the present invention shortens the primer pool optimization process, which traditionally requires several days of manual trial and error and calculation, to a few hours, and automatically provides multiple global optimization schemes that achieve the best balance between cost, performance, and risk.

[0045] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A large-scale primer pool screening system based on parallel computing and multi-objective optimization, characterized in that, include: The candidate primer generation module is used to generate multiple candidate primers for each target based on the input target sequence set, forming an initial candidate primer pool; A high-speed biological characteristic analysis engine, connected to the candidate primer generation module, is used for rapid evaluation of primers and primer pairs in the initial candidate primer pool. The high-speed biological characteristic analysis engine includes: The specific rapid screening submodule, based on a pre-built reference genome k-mer index database, performs rapid calculation and screening of non-target region similarity for candidate primers; The dimer rapid prediction submodule employs GPU-accelerated or large-scale multi-threaded parallel computing methods to rapidly assess the dimer formation risk of candidate primer pairs. A multi-objective collaborative optimization module, connected to the high-speed biological characteristic analysis engine, is used to receive scoring data from the high-speed biological characteristic analysis engine, model the primer selection problem as a multi-objective optimization problem, and solve it using an optimization algorithm, outputting a Pareto optimal solution set that achieves a balance among multiple optimization objectives, wherein the optimization objectives include at least: maximizing target coverage, minimizing the total number of selected primers, minimizing the overall nonspecific risk, and minimizing the overall dimer formation risk.

2. The system according to claim 1, characterized in that, In the specific rapid screening submodule, when constructing the k-mer index database, a sliding window is used to traverse the reference genome, and a unique k-mer filtering mechanism is introduced to retain only k-mer sequences that appear once in the reference genome.

3. The system according to claim 1 or 2, characterized in that, The specific rapid screening submodule supports multi-scale k-mer analysis and can dynamically select the k value according to the application scenario, where k=15 is used for rapid screening, k=18 is used for standard analysis, and k=21 is used for high-strictness scenarios.

4. The system according to claim 1, characterized in that, The GPU parallel computing architecture of the dimer fast prediction submodule adopts a two-dimensional grid thread allocation strategy. Each thread is responsible for calculating a specific primer pair combination and introduces a "seed expansion" heuristic strategy. First, short sequence complete complementary matching is used as the initial screening condition, and then the full sequence free energy is calculated for the primer pairs that pass the initial screening.

5. The system according to claim 1, characterized in that, The multi-objective collaborative optimization module uses the parameter-optimized NSGA-II algorithm, with a population size of 200 and an early stopping mechanism. The iteration is automatically terminated when the Pareto front shows no significant improvement for several consecutive generations.

6. The system according to claim 1 or 5, characterized in that, The multi-objective collaborative optimization module also allows users to dynamically adjust the weights of each optimization objective through an interactive interface to adapt to the priority requirements of different application scenarios.

7. A large-scale primer pool screening method based on parallel computing and multi-objective optimization, characterized in that, Includes the following steps: Step S1: Input the target region sequence set and user configuration parameters; Step S2: Generate an initial candidate primer pool based on the target region sequence set; Step S3: Perform high-speed biological characteristic analysis in parallel, the analysis including: Step S31: Based on the pre-constructed reference genome k-mer index database, perform rapid and specific screening of all primers in the initial candidate primer pool; Step S32: Using a parallel computing method, perform a rapid dimer risk assessment on all primer pairs in the initial candidate primer pool; Step S4: Based on the results of the high-speed biological characteristic analysis described in Step S3, score each primer and primer pair; Step S5: Input the scoring data into a multi-objective optimization algorithm for iterative calculation to simultaneously optimize multiple objectives and find the Pareto optimal frontier. The multiple objectives include at least: maximizing target coverage, minimizing primer number, minimizing nonspecific risk, and minimizing dimer risk. Step S6: Output and visualize the Pareto optimal solution set; Step S7: Based on the user's selection, export the final list of selected primer sequences and a performance report.

8. The method according to claim 7, characterized in that, In the parallel execution of high-speed biological characteristic analysis steps, when assessing the dimer risk of primer pairs, the system first searches for perfectly complementary seed pairs that meet a set length in parallel on the GPU, and then performs precise free energy calculations only on these seed pairs.

9. The method according to claim 7, characterized in that, In the step of outputting and visualizing the Pareto optimal solution set, the Pareto front is displayed in the form of a two-dimensional scatter plot, where the horizontal axis represents the objective to be minimized and the vertical axis represents the objective to be maximized. Each scatter point represents a Pareto optimal solution scheme, and typical schemes are marked for users to select.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps of the method as described in any one of claims 7 to 9.

Images