Estimation genetic correlation method and system based on high-precision likelihood function system

Abstract

This invention discloses a method and system for estimating genetic correlations based on a high-precision likelihood function system, relating to the field of data analysis. By constructing a reference panel feature mapping model and an accuracy mapping model, this invention overcomes the subjectivity and blindness of traditional manual selection and achieves precise learning of the correlation between quality control indicators and estimation accuracy, providing a quantitative basis for optimizing quality control strategies. Secondly, an error verification mechanism is introduced to rigorously judge the quality of the results. When the accuracy is insufficient, the reference panel is dynamically updated first, and then various influencing indicators are iteratively adjusted and continuously verified until the accuracy requirements are met. Finally, the parameters are replaced and the final result is obtained by recalculation. This approach balances the adaptability of the reference panel with the rationality of the quality control parameters, effectively reducing interference factors such as population background bias and genotyping quality fluctuations, and significantly improving the accuracy, robustness, and repeatability of genetic correlation estimation results.

Description

Technical Field

[0001] This invention belongs to the field of data analysis, and specifically relates to a method and system for estimating genetic correlations based on a high-precision likelihood function system. Background Technology

[0002] Currently, the selection of reference panels in the process of genetic correlation estimation relies on human experience, which leads to poor fit and difficulty in accurately matching the genetic background of the population. At the same time, the quality control thresholds are mostly fixed empirical values, which cannot reasonably balance data quality and the preservation of effective genetic information, and are prone to introducing bias. In addition, the accuracy of genetic correlation estimation lacks quantitative evaluation methods, making it difficult to guarantee the accuracy and robustness of the results. Moreover, when the accuracy is insufficient, the parameters cannot be adjusted in a targeted manner, resulting in poor reproducibility of the results and making it difficult to meet the reliable application requirements of complex genetic analysis. Summary of the Invention

[0003] To address the problems in related technologies, this invention proposes a method and system for estimating genetic correlations based on a high-precision likelihood function system, in order to overcome the aforementioned technical problems existing in existing related technologies.

[0004] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: This invention provides a method for estimating genetic correlations based on a high-precision likelihood function system, comprising the following steps: S1. Select the GWAS statistical test Z values ​​of several samples for which genetic correlation estimation needs to be performed and arrange them to obtain the vector set of statistical test Z values ​​to be estimated. S2. Collect historical Z-value vectors and reference panel feature data that meet the accuracy requirements, and construct the final reference panel feature mapping model. S3. Collect data on various influencing indicators and corresponding accuracy rates in historical genetic correlation estimation; then, based on this, construct the final genetic correlation estimation accuracy mapping model. S4. Download the current initial reference panel and generate a reference LD matrix. Select two sets of target vectors from the vector set in S1. Then, calculate the genetic correlation between the two sets of target vectors based on the reference LD matrix and using a high-precision likelihood function. S5. Input the current initial correlation estimation impact index data into the mapping model in S3 for mapping. If the error between the mapping result and the preset accuracy threshold is greater than or equal to the preset error threshold, execute S6; otherwise, directly output the final estimation result. S6. Input the two sets of target vectors in S4 into the mapping model in S2 for mapping; calculate the error between the mapping result and the initial panel features; if the error is greater than or equal to the preset threshold, update the reference panel; after the update is completed, repeatedly adjust the various current initial correlation estimation influence index data in S5 and verify them through the mapping model in S3 until the error requirements are met; after replacing the parameters, repeat S4 to obtain the final estimation result.

[0005] Preferably, step S1 includes the following steps: S11. Select several samples for genetic association estimation, denoted as the sample set to be estimated; obtain the whole genome data of each sample in the sample set to be estimated, to obtain the whole genome dataset to be estimated; obtain and calculate the GWAS statistical test Z value of each SNP from each whole genome data in the whole genome dataset to be estimated, and then arrange the obtained statistical test Z values ​​into a long vector according to chromosome position to obtain the vector set of statistical test Z values ​​to be estimated; accurately calculate the Z value of each SNP through GWAS analysis, and arrange it in order according to chromosome position to form a vector set, which completely preserves the whole genome genetic association signal and provides a high-dimensional, high-information-density core input for genetic association estimation.

[0006] Preferably, step S2 includes the following steps: S21. Based on the current requirements for estimating genetic correlation, obtain the corresponding estimation accuracy threshold to get the current estimation accuracy threshold; set several feature types of the whole genome linkage disequilibrium reference panel to be used in the process of estimating genetic correlation to get the reference panel feature type set; the reference panel feature type set includes ancestry matching degree, sample size, SNP density, and SNP coverage, etc. S22. Based on the reference panel feature type set, obtain two sets of GWAS statistical test Z value vectors and corresponding reference panel feature data in the genetic correlation estimation process corresponding to several historical estimation accuracy rates greater than or equal to the current estimation accuracy threshold, and obtain the historical statistical test Z value vector set and the historical reference panel feature dataset. S23. Based on the historical statistical test Z-value vector set and the historical reference panel feature dataset, construct a mapping model with two sets of statistical test Z-value vectors as input and various types of reference panel feature data as output, and obtain the final reference panel feature mapping model. The final mapping model can accurately output various reference panel feature data with two sets of GWAS statistical test Z-value vectors as input, effectively avoiding the subjectivity and blindness of traditional reference panel selection, realizing accurate matching between reference panels and data to be analyzed, and greatly improving the efficiency and scientific nature of reference panel selection.

[0007] Preferably, step S3 includes the following steps: S31. Several types of influencing indicators that affect the accuracy of genetic correlation estimation are defined to obtain a set of correlation estimation influencing indicator types; the set of correlation estimation influencing indicator types includes minor allele frequency threshold, Hardy-Weinberg equilibrium P-value threshold, and detection rate threshold, etc. Based on the set of correlation estimation impact indicators, obtain data on various correlation estimation impact indicators and corresponding genetic correlation estimation accuracy data from several historical genetic correlation estimation processes, and obtain a historical correlation estimation impact indicator dataset and a historical genetic correlation estimation accuracy dataset. S32. Based on the historical correlation estimation impact index dataset and the historical genetic correlation estimation accuracy dataset, construct a mapping model with various correlation estimation impact index data as input and genetic correlation estimation accuracy data as output, and obtain the final genetic correlation estimation accuracy mapping model. The final mapping model takes various correlation estimation impact index data as input and estimation accuracy as output. It breaks the limitation of traditional quality control thresholds relying on fixed empirical values ​​and can match the best quality control standards for different data scenarios. It not only ensures the statistical stability and genotyping reliability of the selected loci, but also retains biologically significant genetic variation information to the maximum extent, effectively avoiding problems such as estimation bias and insufficient robustness caused by improper quality control.

[0008] Preferably, step S4 includes the following steps: S41. Download the whole genome linkage disequilibrium reference panel to obtain the current initial reference panel; then extract the matching SNP list from the current initial reference panel and generate the reference LD matrix; select two sets of statistical test Z value vectors that need to be calculated for genetic correlation from the set of statistical test Z value vectors to be estimated, and denot them as the first current statistical test Z value vector and the second current statistical test Z value vector, respectively. S42. Based on the reference LD matrix and using a high-precision likelihood function, calculate the genetic correlation estimation result between the first current statistical test Z value vector and the second current statistical test Z value vector to obtain the current genetic correlation estimation result. By downloading the initial genome-wide linkage disequilibrium reference panel and generating the LD matrix, we can accurately capture potential genetic association patterns among traits; while ensuring analysis efficiency, we can achieve robust inference of genetic correlations.

[0009] Preferably, step S42 includes the following steps: S421. Based on the set of correlation estimation impact indicators, initially set the current data of various types of correlation estimation impact indicators to obtain the current initial correlation estimation impact indicator dataset. Based on the current initial correlation estimation impact index dataset, preprocessing operations are performed on the first current statistical test Z-value vector and the second current statistical test Z-value vector to obtain the first preprocessed statistical test Z-value vector and the second preprocessed statistical test Z-value vector. The preprocessing operations include matching and filtering SNP sites, Z-score standardization of each statistical test Z-value vector, and adding small perturbations to the reference LD matrix. S422. Concatenate the Z-value vectors of the two statistical test quantities that need to be estimated for genetic correlation in the preprocessed statistical test quantity Z-value vector set; construct the joint covariance matrix after concatenation to obtain the current joint covariance matrix; S423. Construct a joint log-likelihood function based on the current joint covariance matrix and maximize the likelihood function to obtain the current optimal parameters; Then, the standard error of the current optimal parameter is calculated based on the Fisher information matrix and a significance test is performed to obtain the standard error of the current optimal parameter and the current significance test result; S424. Output the current genetic correlation estimation results; By maximizing the joint log-likelihood function, the optimal estimation of core parameters such as heritability and genetic correlation was achieved. Combined with the standard error and significance test of the Fisher information matrix, the reliability of the parameter estimation was quantified, and the robustness of the results was verified by the 95% confidence interval, effectively reducing the risk of false positives.

[0010] Preferably, step S5 includes the following steps: S51. Set the current genetic correlation estimation error threshold; input the data of various influencing indicators in the current initial correlation estimation influencing indicator dataset into the final genetic correlation estimation accuracy mapping model for mapping, and obtain the current genetic correlation estimation accuracy data. S52. If the absolute value of the difference between the current genetic correlation estimation accuracy data and the current estimation accuracy threshold is greater than or equal to the current genetic correlation estimation error threshold, proceed to S6; otherwise, use the current genetic correlation estimation result as the current final genetic correlation estimation result. By comparing the absolute value of the difference between the current genetic correlation estimation accuracy data and the preset estimation accuracy threshold with the error threshold, when the absolute value of the difference exceeds the error threshold, the subsequent optimization process is initiated in a timely manner, effectively avoiding low-precision estimation results output due to improper adaptation of quality control parameters or reference panels.

[0011] Preferably, step S6 includes the following steps: S61. Set the current reference panel feature error threshold; input the first current statistical test Z value vector and the second current statistical test Z value vector into the final reference panel feature mapping model for mapping to obtain the current reference panel feature mapping dataset; then obtain the various feature data of the current initial reference panel as described in S41 according to the reference panel feature type set to obtain the current actual reference panel feature dataset. S62. Calculate the Euclidean distance between the current actual reference panel feature dataset and the current reference panel feature mapping dataset to obtain the current reference panel feature Euclidean distance data; if the current reference panel feature Euclidean distance data is greater than or equal to the current reference panel feature error threshold, download the corresponding whole-genome linkage disequilibrium reference panel according to the current reference panel feature mapping dataset, obtain the current updated reference panel and execute S63; otherwise, directly execute S63. S63. Repeatedly adjust the current initial correlation estimation influence index dataset to obtain the current adjusted correlation estimation influence index dataset; input the various influence index data in the current adjusted correlation estimation influence index dataset into the final genetic correlation estimation accuracy mapping model for mapping to obtain the current adjusted genetic correlation estimation accuracy data. The final correlation estimation impact index dataset is obtained when the absolute value of the difference between the current adjusted genetic correlation estimation accuracy data and the current estimation accuracy threshold is less than the current genetic correlation estimation error threshold. S64. Replace the current initial reference panel and the current initial correlation estimation influence index dataset in S41 with the current updated reference panel and the current final correlation estimation influence index dataset, respectively, and repeat S41 and S42 to obtain the current final genetic correlation estimation result. Based on the reference panel matching update, the initial correlation estimation impact index dataset is repeatedly adjusted and iteratively verified. The adjustment effect is continuously evaluated through the final genetic correlation estimation accuracy mapping model until the preset accuracy requirements are met. This effectively balances data quality and genetic information integrity, avoids the limitations of single-stage optimization, and ensures the scientific and efficient nature of the optimization process through quantitative judgment standards and iterative mechanisms. It significantly reduces various error interferences and has higher credibility and application value.

[0012] The genetic correlation estimation system based on the high-precision likelihood function system includes a module for obtaining the Z-value vector of the statistical test quantity to be estimated, a module for constructing a reference panel feature mapping model, a module for constructing a genetic correlation estimation accuracy mapping model, a module for calculating and obtaining the current genetic correlation estimation result, a module for determining the adoption of the accuracy of the current genetic correlation estimation result, and a module for adjusting the accuracy of the current genetic correlation estimation. The module for obtaining the Z-value vector of the statistical test statistic to be estimated selects and arranges the Z-values ​​of several samples for which genetic correlation estimation is required, thereby obtaining the Z-value vector set of the statistical test statistic to be estimated. The reference panel feature mapping model construction module sets several reference panel feature types, collects historical Z-value vectors and reference panel feature data that meet the accuracy requirements, and constructs the final reference panel feature mapping model based on this. The genetic correlation estimation accuracy mapping model construction module sets several types of correlation estimation influencing indicators and collects data on various influencing indicators and corresponding accuracy data from historical genetic correlation estimation; then, based on this, it constructs the final genetic correlation estimation accuracy mapping model. The current genetic correlation estimation result calculation and acquisition module downloads the current initial reference panel and generates a reference LD matrix. It selects two sets of target vectors from the vector set in the module for obtaining the Z-value vector of the statistical test quantity to be estimated. Then, based on the reference LD matrix and using a high-precision likelihood function, it calculates the genetic correlation between the two sets of target vectors. The current genetic correlation estimation accuracy adoption determination module inputs the current initial correlation estimation influence index data into the mapping model of the genetic correlation estimation accuracy mapping model construction module. If the error between the mapping result and the preset accuracy threshold is greater than or equal to the preset error threshold, the current genetic correlation estimation accuracy adjustment module is executed; otherwise, the final estimation result is directly output. The current genetic correlation estimation accuracy adjustment module inputs two sets of target vectors from the current genetic correlation estimation result calculation and acquisition module into the mapping model in the reference panel feature mapping model construction module for mapping; calculates the error between the mapping result and the initial panel features; if the error is greater than or equal to a preset threshold, updates the reference panel; after the update is completed, repeatedly adjusts the various current initial correlation estimation influence index data in the current genetic correlation estimation result accuracy adoption judgment module and verifies them through the mapping model in the genetic correlation estimation accuracy mapping model construction module until the error requirements are met; after replacing the parameters, the current genetic correlation estimation result calculation and acquisition module is executed again to obtain the final estimation result.

[0013] The present invention has the following beneficial effects: 1. This invention constructs a reference panel feature mapping model, which can automatically match the optimal reference panel based on the features of the vector to be analyzed, eliminating the subjectivity and blindness of traditional manual selection and ensuring the adaptability of the linkage disequilibrium matrix from the perspective of population genetic background. At the same time, it constructs an accuracy mapping model to achieve precise learning of the correlation between quality control indicators and estimation accuracy, providing a quantitative basis for quality control strategy optimization. After initially completing the genetic correlation calculation, an error verification mechanism is introduced to strictly judge the quality of the results. When the accuracy does not meet the standard, the reference panel is dynamically updated by comparing the error between the model prediction results and the actual panel features. Then, various influencing indicators are iteratively adjusted and continuously verified until the accuracy requirements are met. Finally, the parameters are replaced and recalculated to obtain the final result. Thus, it takes into account both the adaptability of the reference panel and the rationality of the quality control parameters, effectively weakens interference factors such as population background bias and genotyping quality fluctuations, and significantly improves the accuracy, robustness, and repeatability of genetic correlation estimation results, providing efficient and reliable technical support for complex genetic analysis.

[0014] 2. In this invention, by comparing the absolute value of the difference between the current genetic correlation estimation accuracy data and the preset estimation accuracy threshold with the error threshold, when the absolute value of the difference exceeds the error threshold, the subsequent optimization process is initiated in a timely manner, effectively avoiding low-precision estimation results output due to improper adaptation of quality control parameters or reference panels; and ensuring efficient output without sacrificing accuracy, strictly controlling the quality baseline of the estimation results, and avoiding resource waste caused by over-optimization.

[0015] 3. In this invention, the fit of the reference panel is scientifically determined, and the reference panel is updated in a timely manner when the fit is insufficient, thereby eliminating the systematic bias caused by the mismatch of the reference panel at the model base level. On this basis, the initial correlation estimation influence index dataset is repeatedly adjusted and iteratively verified, effectively balancing data quality and the integrity of genetic information. This avoids the limitations of single-stage optimization and ensures the scientificity and efficiency of the optimization process through quantitative judgment standards and iterative mechanisms. The final output genetic correlation estimation results can fully adapt to data characteristics and estimation needs, and significantly reduce various error interferences.

[0016] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the genetic correlation estimation method based on a high-precision likelihood function system according to the present invention. Figure 2 This is a schematic diagram illustrating the process of constructing the final reference panel feature mapping model for this invention. Figure 3 A flowchart illustrating the process of constructing the final genetic correlation estimation accuracy mapping model for this invention; Figure 4 This is a schematic diagram illustrating the calculation process of the current genetic correlation estimation results according to the present invention; Figure 5 This is a schematic diagram illustrating the process by which the present invention determines and adopts the accuracy of current genetic correlation estimation results; Figure 6 This is a schematic diagram illustrating the process of adjusting the accuracy of current genetic correlation estimation according to the present invention; Figure 7 This is a schematic diagram of the module of the genetic correlation estimation system based on the high-precision likelihood function system of the present invention; Figure 8 This is a line graph showing the iterative optimization of the genetic correlation estimation accuracy of this invention. Figure 9 This is a line graph showing the variation curve of the Euclidean distance for the reference panel feature matching in this invention. Detailed Implementation

[0019] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0020] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0021] Example 1 Please see Figure 1 This embodiment describes a method for estimating genetic correlation based on a high-precision likelihood function system, including the following steps: S1. Select the GWAS statistical test Z values ​​of several samples for which genetic correlation estimation needs to be performed and arrange them to obtain the vector set of statistical test Z values ​​to be estimated. Please see Figure 2 S1 includes the following steps: S11. Select several samples for genetic correlation estimation, denoted as the sample set to be estimated; obtain the whole genome data of each sample in the sample set to be estimated to obtain the whole genome dataset to be estimated; obtain and calculate the GWAS (Genome-Wide Association Study, a classic genetic method used to find the association between genes, traits and diseases) statistical test Z value for each SNP (single nucleotide polymorphism) in the whole genome dataset to be estimated, and then arrange the obtained statistical test Z values ​​into a long vector according to chromosome position to obtain the vector set of statistical test Z values ​​to be estimated; For example, the contents of the sample set to be estimated can be referred to as Table 1 below, the contents of the whole genome dataset to be estimated can be referred to as Table 2 and Table 3 below respectively, and the contents of the statistical test Z-value vector set to be estimated can be referred to as Table 4 below; Table 1. Schematic diagram of partial data for the sample information to be estimated. Unique sample number gender Age (years) nationality Phenotype A (essential hypertension) Phenotype B (fasting blood glucose, mmol / L) Data integrity Inclusion / Exclusion Decision HYB001 male 56 Han Chinese 1 (Case) 6.8 99.2% Included HYB002 female 61 Han Chinese 1 (Case) 7.2 98.9% Included HYB003 male 48 Han Chinese 0 (Control) 5.4 99.1% Included HYB004 female 53 Han Chinese 0 (Control) 5.1 98.8% Included HYB005 male 63 Han Chinese 1 (Case) 7.5 99.0% Included Table 2. First schematic table of partial data of the whole genome to be estimated (basic information on SNP sites) SNP ID Chromosome numbering Physical location (hg38 reference genome) Allele 1 (minor allele) Allele 2 (major allele) MAF (minor allele frequency) HWE P value Detection rate rs10757278 9 22125502 A G 0.42 0.87 99.5% rs5998391 1 58484264 C T 0.38 0.91 99.3% rs2981582 10 123256789 A C 0.29 0.79 99.4% rs1801132 1 234567890 G A 0.18 0.85 99.2% rs7903146 10 112345678 T C 0.35 0.88 99.4% Table 3. Second schematic diagram of partial whole-genome data to be estimated (genotype data) Sample number rs10757278 (genotype / code) rs599839 (genotype / code) rs2981582 (genotype / code) rs1801132 (genotype / code) rs7903146 ​​(genotype / code) HYB001 GG / 2 TT / 2 CC / 2 AA / 2 CC / 2 HYB002 AG / 1 CT / 1 AC / 1 GA / 1 CT / 1 HYB003 AA / 0 CC / 0 AA / 0 GG / 0 TT / 0 HYB004 AG / 1 TT / 2 AC / 1 AA / 2 CT / 1 HYB005 GG / 2 CT / 1 CC / 2 GA / 1 CC / 2 In this context, 0 = homozygous minor allele (such as AA in rs10757278), 1 = heterozygous (AG), and 2 = homozygous major allele (GG). Table 4. Schematic diagram of partial data for the Z-value vector of the statistical test quantity to be estimated. SNP ID chromosome Physical location (hg38) Z-score (hypertension) Z-score (fasting blood glucose) Sort number rs599839 1 158484264 1.475 1.769 1 rs1801132 1 234567890 0.689 1.164 2 rs10757278 9 22125502 3.468 4.620 3 rs7903146 10 112345678 -2.133 -2.906 4 rs2981582 10 123256789 -2.690 -3.381 5 At the sample level, the selection of suitable samples by clearly defining inclusion and exclusion criteria ensured the reliability of the data and the representativeness of the population, laying a high-quality data foundation for subsequent analysis. The systematic acquisition and standardized organization of whole-genome data not only covered the complete collection of basic SNP loci information, but also achieved data unification and operability through standardized genotype coding. In terms of the construction of statistical test measures, the Z value of each SNP was accurately calculated through GWAS analysis and arranged in an orderly manner according to chromosome position to form a vector set, which completely preserved the whole-genome genetic association signal and provided a high-dimensional, high-information-density core input for the estimation of genetic correlation. S2. Collect historical Z-value vectors and reference panel feature data that meet the accuracy requirements, and construct the final reference panel feature mapping model. S2 includes the following steps: S21. Based on the current requirements for genetic correlation estimation, obtain the corresponding estimation accuracy threshold to obtain the current estimation accuracy threshold; set several feature types of the whole-genome linkage disequilibrium reference panel (the whole-genome linkage disequilibrium reference panel is a standard reference dataset constructed based on a specific population, containing a large number of single nucleotide polymorphism sites and their linkage disequilibrium structure information throughout the whole genome, used to provide a stable and reliable reference for population genetic background and locus association patterns for genetic association analysis, genetic correlation estimation and other studies) to obtain the reference panel feature type set; the reference panel feature type set includes ancestry matching degree (referring to the degree of consistency between the population source corresponding to the reference panel and the genetic ancestral background of the sample to be analyzed, which is the core condition to ensure the accuracy of linkage disequilibrium structure), sample size (referring to the number of independent individuals used to construct the reference panel, which directly determines the stability and accuracy of linkage disequilibrium matrix estimation), SNP density (referring to the number of variant sites contained in a unit genome length, reflecting the density of marker distribution on chromosomes) and SNP coverage (referring to the completeness of the coverage of the whole genome region by the variant sites in the reference panel, reflecting the completeness of the capture of genetic regions), etc. S22. Based on the reference panel feature type set, obtain two sets of GWAS statistical test Z value vectors and corresponding reference panel feature data in the genetic correlation estimation process corresponding to several historical estimation accuracy rates greater than or equal to the current estimation accuracy threshold, and obtain the historical statistical test Z value vector set and the historical reference panel feature dataset. S23. Based on the historical statistical test Z-value vector set and the historical reference panel feature dataset, construct a mapping model with two sets of statistical test Z-value vectors as input and various types of reference panel feature data as output, and obtain the final reference panel feature mapping model. S23 includes the following steps: S231. Construct an initial reference panel feature mapping model and set a first training data ratio (e.g., 8:2 or 7:3, which can be adjusted adaptively according to the actual training situation); divide the historical statistical test Z-value vector set and the historical reference panel feature dataset according to the first training data ratio to obtain the first training dataset and the first test dataset. S232. Set a first training error threshold (10%~15%, which can be adjusted adaptively according to the actual training situation); input the first training dataset into the initial reference panel feature mapping model for training; during the training process, if the training error is less than the first training error threshold, stop training and obtain the trained reference panel feature mapping model; otherwise, continue training until the training error is less than the first training error threshold. S233. Set a first test accuracy threshold (90%~95%, which can be adjusted adaptively according to the actual test situation); input the first test dataset into the trained reference panel feature mapping model for testing; after the test is completed, obtain the first test accuracy data; if the first test accuracy data is greater than or equal to the first test accuracy threshold, use the trained reference panel feature mapping model as the final reference panel feature mapping model; otherwise, return to S232 to continue training the trained reference panel feature mapping model and repeat S233 until the first test accuracy data is greater than or equal to the first test accuracy threshold. The initial reference panel feature mapping model can adopt a fusion model of multi-scale convolution and deep fully connected neural networks; including an input layer, a one-dimensional convolutional layer (3 layers in total), four fully connected layers and an output layer; The input layer uses two sets of GWAS statistical test Z-value vectors as the original input, with the input dimension consistent with the length of the Z-value vector sites, to carry genetic association statistical features. The first one-dimensional convolutional layer has 64 convolutional kernels with a kernel size of 3; the second one-dimensional convolutional layer has 128 convolutional kernels with a kernel size of 5; and the third one-dimensional convolutional layer has 256 convolutional kernels with a kernel size of 7. These layers are used to extract local association patterns, global distribution features, and genetic structure features of the Z-value vectors layer by layer. After each convolutional layer, a pooling layer is connected, with a uniform pooling kernel size of 2, to compress the feature dimension, filter redundant noise, and retain key genetic information. The multi-dimensional feature matrix after convolution and pooling is then converted into a one-dimensional feature vector to achieve feature connection with the fully connected layer. The first, second, third, and fourth fully connected layers have 512, 256, 128, and 64 neurons, respectively, used to fuse high-dimensional abstract features layer by layer and learn the nonlinear mapping relationship between the Z-value vector and the reference panel features. In addition, both the convolutional and fully connected layers use the Gaussian Error Linear Unit (GELU) activation function to improve the stability and convergence speed of the model in fitting genetic data and avoid gradient vanishing. Two Dropout layers are connected between the fully connected layers with a dropout rate of 0.2 to reduce the risk of overfitting and enhance the generalization ability. The number of neurons in the output layer is consistent with the number of features in the four types of reference panels: ancestor matching degree, sample size, SNP density, and SNP coverage. A linear activation function is used to directly output the standardized optimal feature data of the reference panel. By constructing a reference panel feature mapping model that accurately adapts to the needs of genetic correlation estimation, intelligent and personalized reference panel selection is achieved. Specifically, firstly, based on the core requirements of genetic correlation estimation, the estimation accuracy threshold is clearly defined, setting quantitative standards for reference panel selection to ensure that the selected reference panels can meet specific accuracy requirements. Simultaneously, key reference panel feature types such as ancestry matching degree, sample size, SNP density, and SNP coverage are systematically reviewed, comprehensively covering the core factors affecting the accuracy of genetic correlation estimation, providing comprehensive and crucial feature dimensions for subsequent model construction. Furthermore, by collecting Z-value vector sets from historical genetic correlation estimation cases that met the accuracy threshold, and... By using corresponding reference panel feature data, the effectiveness and reliability of the training data are ensured, and the model is able to learn the inherent correlation between different Z-value vector features and the optimal reference panel features. The final mapping model can accurately output various suitable reference panel feature data from two sets of GWAS statistical test Z-value vectors as input, effectively avoiding the subjectivity and blindness of traditional reference panel selection and achieving accurate matching between the reference panel and the data to be analyzed. This not only greatly improves the efficiency and scientific nature of reference panel selection, but also ensures the accuracy, robustness and reproducibility of genetic correlation estimation results from the source, laying a solid foundation for subsequent high-precision genetic correlation calculations. S3. Collect data on various influencing indicators and corresponding accuracy rates in historical genetic correlation estimation; then, based on this, construct the final genetic correlation estimation accuracy mapping model. Please see Figure 3 S3 includes the following steps: S31. Several types of influencing indicators that affect the accuracy of genetic correlation estimation are defined to obtain a set of correlation estimation influencing indicator types. The set of correlation estimation influencing indicator types includes a minor allele frequency threshold (locuses with excessively low minor allele frequencies are prone to drastic fluctuations in statistics and unreliable genetic association signals due to insufficient sample representativeness; excessively high allele frequencies will eliminate low-frequency functional variations with potential functional significance), a Hardy-Weinberg equilibrium P-value threshold (Hardy-Weinberg equilibrium is the basic principle of ideal population genetic distribution; loci that deviate significantly from equilibrium are usually accompanied by systematic biases such as genotyping errors, sample contamination, population stratification, or inbreeding, which will directly interfere with linkage disequilibrium structure and Z-value distribution; however, an overly strict threshold will mistakenly screen out functional loci that are truly affected by natural selection), and a detection rate threshold (locus detection rate reflects the proportion of effective data from genotyping; an excessively low detection rate means a large number of samples are missing and the data reliability is insufficient, which will introduce systematic bias and reduce estimation robustness; while an excessively strict detection rate threshold will excessively reduce the number of usable loci and weaken the genome coverage), etc. Based on the set of correlation estimation impact indicators, obtain data on various correlation estimation impact indicators and corresponding genetic correlation estimation accuracy data from several historical genetic correlation estimation processes, and obtain a historical correlation estimation impact indicator dataset and a historical genetic correlation estimation accuracy dataset. S32. Based on the historical correlation estimation impact index dataset and the historical genetic correlation estimation accuracy dataset, construct a mapping model with various correlation estimation impact index data as input and genetic correlation estimation accuracy data as output, and obtain the final genetic correlation estimation accuracy mapping model. S32 includes the following steps: S321. Construct an initial genetic correlation estimation accuracy mapping model and set a second training data ratio (e.g., 8:2 or 7:3, which can be adaptively adjusted according to the actual training situation); divide the existing hydrogeological parameter dataset, existing aquifer environmental factor dataset, existing aquifer injection rate dataset, existing aquifer production rate dataset, and existing aquifer pressure dataset according to the second training data ratio to obtain the second training dataset and the second test dataset. S322. Set a second training error threshold (10%~15%, which can be adjusted adaptively according to the actual training situation); input the second training dataset into the initial genetic correlation estimation accuracy mapping model for training; during the training process, if the training error is less than the second training error threshold, stop training and obtain the trained genetic correlation estimation accuracy mapping model; otherwise, continue training until the training error is less than the second training error threshold. S323. Set a second test accuracy threshold (90%~95%, which can be adjusted adaptively according to the actual test situation); input the second test dataset into the trained genetic correlation estimation accuracy mapping model for testing; after the test is completed, obtain the second test accuracy data; if the second test accuracy data is greater than or equal to the second test accuracy threshold, use the trained genetic correlation estimation accuracy mapping model as the final genetic correlation estimation accuracy mapping model; otherwise, return to S322 to continue training the trained genetic correlation estimation accuracy mapping model and repeat S323 until the second test accuracy data is greater than or equal to the second test accuracy threshold. The initial genetic correlation estimation accuracy mapping model can be a lightweight deep fully connected neural network regression model, which includes an input layer, four fully connected layers, and an output layer. The input layer has three neurons, corresponding to three categories of indicators: minor allele frequency threshold, Hardy-Weinberg balance P-value threshold, and detection rate threshold. The fully connected layers in the first, second, third, and fourth layers have 64, 128, 64, and 32 neurons, respectively, used to nonlinearly fuse and abstract the quality control threshold features layer by layer, learning the mapping relationship between threshold combinations and accuracy. All fully connected layers use a leaky ReLU activation function with a negative half-axis slope of 0.01 to avoid neuron inactivation and improve the stability of small sample fitting. In addition, a dropout layer is set in each of the second and third fully connected layers with a dropout rate of 0.2 to reduce overfitting and enhance the model's generalization ability. The output layer has one neuron and uses a linear activation function to directly output the continuous genetic correlation estimation accuracy value. By constructing a precise genetic correlation estimation accuracy mapping model, the scientific optimization and dynamic adaptation of quality control thresholds were achieved. Specifically, by sorting out the core influencing indicators such as minor allele frequency thresholds, Hardy-Weinberg equilibrium P-value thresholds, and detection rate thresholds, the model comprehensively covers the key dimensions that determine data quality and the integrity of genetic information in the quality control process. It fully considers both the technical errors and signal distortion caused by overly lenient thresholds and the risks of effective site loss and genetic structure damage caused by overly strict thresholds, laying a comprehensive and practical foundation for model construction. Secondly, by collecting various quality control indicator data and corresponding accuracy data from historical genetic correlation estimation processes, the model's training accuracy was ensured. The authenticity and correlation of the training data enable the model to deeply explore the inherent patterns between different combinations of quality control thresholds and estimation accuracy. The final mapping model, with various correlation estimation influencing index data as input and estimation accuracy as output, breaks through the limitations of traditional quality control thresholds relying on fixed empirical values. It can match the best quality control standards for different data scenarios, ensuring the statistical stability and genotyping reliability of the selected loci, while preserving biologically significant genetic variation information to the maximum extent. It effectively avoids problems such as estimation bias and insufficient robustness caused by improper quality control, significantly improves the accuracy and repeatability of genetic correlation estimation, and provides high-quality data support for subsequent high-precision genetic association analysis. S4. Download the current initial reference panel and generate a reference LD matrix. Select two sets of target vectors from the vector set in S1. Then, calculate the genetic correlation between the two sets of target vectors based on the reference LD matrix and using a high-precision likelihood function. S4 includes the following steps: S41. Download the whole genome linkage disequilibrium reference panel to obtain the current initial reference panel; then extract the matching SNP list from the current initial reference panel and generate the reference LD matrix; select two sets of statistical test Z value vectors that need to be calculated for genetic correlation from the set of statistical test Z value vectors to be estimated, and denot them as the first current statistical test Z value vector and the second current statistical test Z value vector, respectively. S42. Based on the reference LD matrix and using a high-precision likelihood function, calculate the genetic correlation estimation result between the first current statistical test Z value vector and the second current statistical test Z value vector to obtain the current genetic correlation estimation result. S42 includes the following steps: S421. Based on the set of correlation estimation impact indicators, initially set the current data of various types of correlation estimation impact indicators to obtain the current initial correlation estimation impact indicator dataset. Based on the current initial correlation estimation impact index dataset, preprocessing operations are performed on the first current statistical test Z-value vector and the second current statistical test Z-value vector to obtain the first preprocessed statistical test Z-value vector and the second preprocessed statistical test Z-value vector. The preprocessing operations include matching and filtering SNP sites (e.g., filtering low-quality SNPs: minor allele frequency <1% (minor allele frequency threshold), detection rate <95% (detection rate threshold), HWE P-value <1e-6 (Hardy-Weinberg equilibrium P-value threshold)), Z-score standardization of each statistical test Z-value vector, and adding small perturbations to the reference LD matrix, etc. S422. Concatenate the Z-value vectors of the two statistical test quantities that need to be estimated for genetic correlation in the preprocessed statistical test quantity Z-value vector set; construct the joint covariance matrix after concatenation to obtain the current joint covariance matrix; S423. Construct a joint log-likelihood function based on the current joint covariance matrix and maximize the likelihood function to obtain the current optimal parameters; Then, the standard error of the current optimal parameter is calculated based on the Fisher information matrix and a significance test is performed to obtain the standard error of the current optimal parameter and the current significance test result; S424. Output the current genetic correlation estimation results; for example, the current genetic correlation estimation results can be referred to as Table 5 below: Table 5. Example of Genetic Correlation Estimation Results

[0022] Results analysis: =0.60 indicates a significant positive genetic association between hypertension and fasting blood glucose (60% shared genetic variation); a 95% confidence interval excluding 0 indicates robust results; In the data preprocessing stage, strict SNP locus quality control standards were used to systematically filter out low-quality loci with low minor allele frequencies, low detection rates, and Hardy-Weinberg imbalances, effectively eliminating interference from technical errors and data contamination. Simultaneously, Z-score standardization was employed to eliminate magnitude differences and sample size biases among different Z-score vectors, providing high-quality, comparable core input data for subsequent analysis. Adding minor perturbations to the reference LD matrix further avoided matrix singularity issues and ensured the stability of the computation process. The construction of the joint covariance matrix fully integrated the genetic background information of the two traits with the L... The D-structure features enable the model to accurately capture potential genetic association patterns among traits. By jointly maximizing the log-likelihood function, optimal estimates of core parameters such as heritability and genetic correlation are achieved. Combined with the standard error and significance test calculated by the Fisher information matrix, the reliability of parameter estimates is quantified, and the robustness of the results is verified by the 95% confidence interval, effectively reducing the risk of false positives. Thus, it can not only accurately quantify the degree of genetic sharing among traits, providing reliable data support for analyzing the genetic co-occurrence mechanism of complex traits, but also ensure the reproducibility and comparability of the results through standardized processing and calculation steps. By downloading the initial genome-wide linkage disequilibrium reference panel and generating the LD matrix, the accuracy of genetic background modeling was ensured. The application of high-precision likelihood functions fully utilized the genome-wide LD structure information, significantly improving the accuracy and statistical power of genetic association estimation compared to traditional methods. It can accurately capture potential genetic association patterns between traits. The entire technical process does not rely on raw data at the individual level, ensuring analytical efficiency while achieving robust inference of genetic associations. This provides strong technical support for elucidating the genetic co-occurrence mechanisms among complex traits and uncovering potential pathogenic pathways. S5. Input the current initial correlation estimation impact index data into the mapping model in S3 for mapping. If the error between the mapping result and the preset accuracy threshold is greater than or equal to the preset error threshold, execute S6; otherwise, directly output the final estimation result. Please see Figure 4 S5 includes the following steps: S51. Set the current genetic correlation estimation error threshold (which can be adaptively set according to the actual genetic correlation estimation requirements); input the data of various influence indicators in the current initial correlation estimation influence indicator dataset into the final genetic correlation estimation accuracy mapping model for mapping to obtain the current genetic correlation estimation accuracy data; S52. If the absolute value of the difference between the current genetic correlation estimation accuracy data and the current estimation accuracy threshold is greater than or equal to the current genetic correlation estimation error threshold, proceed to S6; otherwise, use the current genetic correlation estimation result as the current final genetic correlation estimation result. By using error threshold determination and accuracy mapping verification, precise screening and quality control of genetic correlation estimation results are achieved. Specifically, firstly, a scientifically reasonable genetic correlation estimation error threshold is set to provide a clear quantitative standard for result quality judgment and avoid bias caused by subjective judgment. Secondly, using the constructed genetic correlation estimation accuracy mapping model, various correlation estimation influencing index data actually used in the current estimation process are substituted into the model for mapping, quickly obtaining the current genetic correlation estimation accuracy data that can objectively reflect the accuracy of this estimation, thus achieving efficient quantitative evaluation of estimation quality. By comparing the absolute value of the difference between the current genetic correlation estimation accuracy data and the preset estimation accuracy threshold with the error threshold, when the absolute value of the difference exceeds the error threshold, the subsequent optimization process is initiated in a timely manner, effectively avoiding low-precision estimation results output due to improper adaptation of quality control parameters or reference panels. When the absolute value of the difference is within the error threshold range, the current estimation result is directly determined as the final result, ensuring efficient output without sacrificing accuracy. Thus, the quality baseline of the estimation results is strictly controlled, and the waste of resources caused by over-optimization is avoided, ensuring that the final output genetic correlation estimation results meet the preset accuracy requirements and have good robustness and reliability. S6. Input the two sets of target vectors in S4 into the mapping model in S2 for mapping; calculate the error between the mapping result and the initial panel features; if the error is greater than or equal to the preset threshold, update the reference panel; after the update is completed, repeatedly adjust the various current initial correlation estimation impact index data in S5 and verify them through the mapping model in S3 until the error requirements are met; after replacing the parameters, repeat S4 to obtain the final estimation result; Please see Figure 5 , Figure 6 , Figure 8 , Figure 9 S6 includes the following steps: S61. Set the current reference panel feature error threshold (which can be adaptively set according to the actual genetic correlation estimation requirements); input the first current statistical test Z value vector and the second current statistical test Z value vector into the final reference panel feature mapping model for mapping to obtain the current reference panel feature mapping dataset; then obtain the various feature data of the current initial reference panel as described in S41 according to the reference panel feature type set to obtain the current actual reference panel feature dataset. S62. Calculate the Euclidean distance between the current actual reference panel feature dataset and the current reference panel feature mapping dataset to obtain the current reference panel feature Euclidean distance data; if the current reference panel feature Euclidean distance data is greater than or equal to the current reference panel feature error threshold, download the corresponding whole-genome linkage disequilibrium reference panel according to the current reference panel feature mapping dataset, obtain the current updated reference panel and execute S63; otherwise, directly execute S63. Excessive sample size and high SNP density in the reference panel can introduce systematic bias, noise inflation, and overfitting. By prioritizing the replacement and adaptation of the reference panel, systematic biases in genetic correlation estimation can be corrected at the underlying level. This addresses the covariance structure distortion caused by key factors such as ancestral differences in the reference sample, linkage disequilibrium generational drift, and local block distortion, significantly improving the accuracy of genetic background modeling. Compared to fine-tuning quality control thresholds such as minor allele frequencies, Hardy-Weinberg equilibrium p-values, and detection rates, replacing the reference panel contributes more to reducing genetic correlation estimation bias. It can correct global basis errors in one go, avoiding irreproducible results and directional errors. Further optimization of quality control standards can further filter low-quality tail variation noise and enhance estimation stability. This priority strategy balances global bias correction and local noise reduction, ensuring both the ability to approximate the true value of genetic correlation and improving parameter robustness and statistical reliability. S63. Repeatedly adjust the current initial correlation estimation influence index dataset to obtain the current adjusted correlation estimation influence index dataset; input the various influence index data in the current adjusted correlation estimation influence index dataset into the final genetic correlation estimation accuracy mapping model for mapping to obtain the current adjusted genetic correlation estimation accuracy data. The final correlation estimation impact index dataset is obtained when the absolute value of the difference between the current adjusted genetic correlation estimation accuracy data and the current estimation accuracy threshold is less than the current genetic correlation estimation error threshold. S64. Replace the current initial reference panel and the current initial correlation estimation influence index dataset in S41 with the current updated reference panel and the current final correlation estimation influence index dataset, respectively, and repeat S41 and S42 to obtain the current final genetic correlation estimation result. By employing a dual iterative optimization mechanism involving reference panel adaptation optimization and dynamic adjustment of quality control indicators, a precise closed-loop improvement of the genetic correlation estimation process was achieved, significantly enhancing the accuracy and robustness of the final results. Specifically, an adaptively adjustable reference panel feature error threshold was first set to accommodate the personalized needs of different estimation scenarios. The optimal reference panel feature dataset, adapted to the current Z-value vector, was obtained using the final reference panel feature mapping model. The Euclidean distance between the reference panel and the actual reference panel feature dataset was calculated to scientifically determine the degree of reference panel adaptation. When the adaptation was insufficient, the reference panel was updated promptly, eliminating systematic bias caused by reference panel mismatch at the model foundation level. Based on this, further improvements were made to... The initial correlation estimation impact index dataset is repeatedly adjusted and iteratively verified. The adjustment effect is continuously evaluated through the final genetic correlation estimation accuracy mapping model until the preset accuracy requirements are met, forming the best combination of control indicators, which effectively balances data quality and the integrity of genetic information. By replacing the initial parameters with the updated reference panel and the best control indicator dataset and repeating the genetic correlation estimation process, the limitations of single-stage optimization are avoided. The scientific and efficient nature of the optimization process is ensured by quantitative judgment criteria and iterative mechanisms. The final output genetic correlation estimation results can fully adapt to data characteristics and estimation needs, greatly reduce various error interferences, and have higher reliability and application value.

[0023] Example 2 Please see Figure 7 This embodiment discloses a genetic correlation estimation system based on a high-precision likelihood function system. The system can implement the method of the above embodiment, including a module for obtaining the Z-value vector of the statistical test quantity to be estimated, a module for constructing a reference panel feature mapping model, a module for constructing a genetic correlation estimation accuracy mapping model, a module for calculating and obtaining the current genetic correlation estimation result, a module for determining the adoption of the accuracy of the current genetic correlation estimation result, and a module for adjusting the accuracy of the current genetic correlation estimation. The module for obtaining the Z-value vector of the statistical test statistic to be estimated selects and arranges the Z-values ​​of several samples for which genetic correlation estimation is required, thereby obtaining the Z-value vector set of the statistical test statistic to be estimated. The reference panel feature mapping model construction module sets several reference panel feature types, collects historical Z-value vectors and reference panel feature data that meet the accuracy requirements, and constructs the final reference panel feature mapping model based on this. The genetic correlation estimation accuracy mapping model construction module sets several types of correlation estimation influencing indicators and collects data on various influencing indicators and corresponding accuracy data from historical genetic correlation estimation; then, based on this, it constructs the final genetic correlation estimation accuracy mapping model. The current genetic correlation estimation result calculation and acquisition module downloads the current initial reference panel and generates a reference LD matrix. It selects two sets of target vectors from the vector set in the module for obtaining the Z-value vector of the statistical test quantity to be estimated. Then, based on the reference LD matrix and using a high-precision likelihood function, it calculates the genetic correlation between the two sets of target vectors. The current genetic correlation estimation accuracy adoption determination module inputs the current initial correlation estimation influence index data into the mapping model of the genetic correlation estimation accuracy mapping model construction module. If the error between the mapping result and the preset accuracy threshold is greater than or equal to the preset error threshold, the current genetic correlation estimation accuracy adjustment module is executed; otherwise, the final estimation result is directly output. The current genetic correlation estimation accuracy adjustment module inputs two sets of target vectors from the current genetic correlation estimation result calculation and acquisition module into the mapping model in the reference panel feature mapping model construction module for mapping; calculates the error between the mapping result and the initial panel features; if the error is greater than or equal to a preset threshold, updates the reference panel; after the update is completed, repeatedly adjusts the various current initial correlation estimation influence index data in the current genetic correlation estimation result accuracy adoption judgment module and verifies them through the mapping model in the genetic correlation estimation accuracy mapping model construction module until the error requirements are met; after replacing the parameters, the current genetic correlation estimation result calculation and acquisition module is executed again to obtain the final estimation result.

[0024] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0025] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present invention.

Claims

1. A method for estimating genetic correlation based on a high-precision likelihood function system, characterized in that, Includes the following steps: S1. Select the GWAS statistical test Z values ​​of several samples for which genetic correlation estimation needs to be performed and arrange them to obtain the vector set of statistical test Z values ​​to be estimated. S2. Collect historical Z-value vectors and reference panel feature data that meet the accuracy requirements, and construct the final reference panel feature mapping model. S3. Collect data on various influencing indicators and corresponding accuracy data in historical genetic correlation estimation; Based on this, a final genetic correlation estimation accuracy mapping model is then constructed; S4. Download the current initial reference panel and generate a reference LD matrix. Select two sets of target vectors from the vector set in S1. Then, calculate the genetic correlation between the two sets of target vectors based on the reference LD matrix and using a high-precision likelihood function. S5. Input the current initial correlation estimation impact index data into the mapping model in S3 for mapping. If the error between the mapping result and the preset accuracy threshold is greater than or equal to the preset error threshold, execute S6; otherwise, directly output the final estimation result. S6. Input the two sets of target vectors in S4 into the mapping model in S2 for mapping; calculate the error between the mapping result and the initial panel features; if the error is greater than or equal to the preset threshold, update the reference panel; after the update is completed, repeatedly adjust the various current initial correlation estimation influence index data in S5 and verify them through the mapping model in S3 until the error requirements are met; after replacing the parameters, repeat S4 to obtain the final estimation result.

2. The method for estimating genetic correlation based on a high-precision likelihood function system according to claim 1, characterized in that, S1 includes the following steps: S11. Select several samples for genetic correlation estimation, denoted as the sample set to be estimated; obtain the whole genome data of each sample in the sample set to be estimated to obtain the whole genome dataset to be estimated; obtain and calculate the GWAS statistical test Z value of each SNP from each whole genome data in the whole genome dataset to be estimated, and then arrange the obtained statistical test Z values ​​into a long vector according to chromosome position to obtain the vector set of statistical test Z values ​​to be estimated.

3. The method for estimating genetic correlation based on a high-precision likelihood function system according to claim 2, characterized in that, S2 includes the following steps: S21. Based on the current requirements for estimating genetic correlation, obtain the corresponding estimation accuracy threshold to get the current estimation accuracy threshold; set several feature types of the whole genome linkage disequilibrium reference panel to be used in the process of estimating genetic correlation to get the reference panel feature type set; the reference panel feature type set includes ancestry matching degree, sample size, SNP density and SNP coverage; S22. Based on the reference panel feature type set, obtain two sets of GWAS statistical test Z value vectors and corresponding reference panel feature data in the genetic correlation estimation process corresponding to several historical estimation accuracy rates greater than or equal to the current estimation accuracy threshold, and obtain the historical statistical test Z value vector set and the historical reference panel feature dataset. S23. Based on the historical statistical test Z-value vector set and the historical reference panel feature dataset, construct a mapping model with two sets of statistical test Z-value vectors as input and various types of reference panel feature data as output, and obtain the final reference panel feature mapping model.

4. The method for estimating genetic correlation based on a high-precision likelihood function system according to claim 3, characterized in that, S3 includes the following steps: S31. Several types of influencing indicators that affect the accuracy of genetic correlation estimation are defined to obtain a set of correlation estimation influencing indicator types; the set of correlation estimation influencing indicator types includes the minor allele frequency threshold, the Hardy-Weinberg equilibrium P-value threshold, and the detection rate threshold. Based on the set of correlation estimation impact indicators, obtain data on various correlation estimation impact indicators and corresponding genetic correlation estimation accuracy data from several historical genetic correlation estimation processes, and obtain a historical correlation estimation impact indicator dataset and a historical genetic correlation estimation accuracy dataset. S32. Based on the historical correlation estimation impact index dataset and the historical genetic correlation estimation accuracy dataset, construct a mapping model with various correlation estimation impact index data as input and genetic correlation estimation accuracy data as output, and obtain the final genetic correlation estimation accuracy mapping model.

5. The method for estimating genetic correlation based on a high-precision likelihood function system according to claim 4, characterized in that, S4 includes the following steps: S41. Download the whole genome linkage disequilibrium reference panel to obtain the current initial reference panel; then extract the matching SNP list from the current initial reference panel and generate the reference LD matrix; select two sets of statistical test Z value vectors that need to be calculated for genetic correlation from the set of statistical test Z value vectors to be estimated, and denot them as the first current statistical test Z value vector and the second current statistical test Z value vector, respectively. S42. Based on the reference LD matrix and using a high-precision likelihood function, calculate the genetic correlation estimation result between the first current statistical test Z value vector and the second current statistical test Z value vector to obtain the current genetic correlation estimation result.

6. The method for estimating genetic correlation based on a high-precision likelihood function system according to claim 5, characterized in that, S42 includes the following steps: S421. Based on the set of correlation estimation impact indicators, initially set the current data of various types of correlation estimation impact indicators to obtain the current initial correlation estimation impact indicator dataset. Based on the current initial correlation estimation impact index dataset, preprocessing operations are performed on the first current statistical test Z-value vector and the second current statistical test Z-value vector to obtain the first preprocessed statistical test Z-value vector and the second preprocessed statistical test Z-value vector; the preprocessing operations include matching and filtering SNP sites, Z-score standardization of each statistical test Z-value vector, and adding small perturbations to the reference LD matrix; S422. Concatenate the Z-value vectors of the two statistical test quantities that need to be estimated for genetic correlation in the preprocessed statistical test quantity Z-value vector set; construct the joint covariance matrix after concatenation to obtain the current joint covariance matrix; S423. Construct a joint log-likelihood function based on the current joint covariance matrix and maximize the likelihood function to obtain the current optimal parameters; Then, the standard error of the current optimal parameter is calculated based on the Fisher information matrix and a significance test is performed to obtain the standard error of the current optimal parameter and the current significance test result; S424. Output the current genetic correlation estimation results.

7. The method for estimating genetic correlation based on a high-precision likelihood function system according to claim 6, characterized in that, S5 includes the following steps: S51. Set the current genetic correlation estimation error threshold; input the data of various influence indicators in the current initial correlation estimation influence indicator dataset into the final genetic correlation estimation accuracy mapping model for mapping, and obtain the current genetic correlation estimation accuracy data.

8. The method for estimating genetic correlation based on a high-precision likelihood function system according to claim 7, characterized in that, S5 further includes the following steps: S52. If the absolute value of the difference between the current genetic correlation estimation accuracy data and the current estimation accuracy threshold is greater than or equal to the current genetic correlation estimation error threshold, proceed to S6; otherwise, take the current genetic correlation estimation result as the current final genetic correlation estimation result.

9. The method for estimating genetic correlation based on a high-precision likelihood function system according to claim 8, characterized in that, S6 includes the following steps: S61. Set the current reference panel feature error threshold; input the first current statistical test Z value vector and the second current statistical test Z value vector into the final reference panel feature mapping model for mapping to obtain the current reference panel feature mapping dataset; then obtain the various feature data of the current initial reference panel as described in S41 according to the reference panel feature type set to obtain the current actual reference panel feature dataset. S62. Calculate the Euclidean distance between the current actual reference panel feature dataset and the current reference panel feature mapping dataset to obtain the current reference panel feature Euclidean distance data; if the current reference panel feature Euclidean distance data is greater than or equal to the current reference panel feature error threshold, download the corresponding whole-genome linkage disequilibrium reference panel according to the current reference panel feature mapping dataset, obtain the current updated reference panel and execute S63; otherwise, directly execute S63. S63. Repeatedly adjust the current initial correlation estimation influence index dataset to obtain the current adjusted correlation estimation influence index dataset; input the various influence index data in the current adjusted correlation estimation influence index dataset into the final genetic correlation estimation accuracy mapping model for mapping to obtain the current adjusted genetic correlation estimation accuracy data. The final correlation estimation impact index dataset is obtained when the absolute value of the difference between the current adjusted genetic correlation estimation accuracy data and the current estimation accuracy threshold is less than the current genetic correlation estimation error threshold. S64. Replace the current initial reference panel and the current initial correlation estimation influence index dataset in S41 with the current updated reference panel and the current final correlation estimation influence index dataset, respectively, and repeat S41 and S42 to obtain the current final genetic correlation estimation result.

10. A system for implementing the method for estimating genetic correlation based on a high-precision likelihood function system as described in any one of claims 1-9.

Images