Sequencing data heredity estimation method and system, computer device and computer program product

By optimizing the heritability estimation using the Girard-Hutchinson stochastic estimation method, the problem of high computational complexity in existing technologies is solved, and efficient heritability estimation is achieved under large-scale samples while maintaining computational accuracy.

CN121034402APending Publication Date: 2025-11-28ZHEJIANG CHINESE MEDICAL UNIVERSITY +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202511155690.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-08-18
Publication Date
2025-11-28

AI Technical Summary

Technical Problem

Existing technologies suffer from high computational complexity when estimating the genetic rate of millions of samples, leading to memory overflows and exhaustion of computing resources, making them unsuitable for large-scale data processing needs.

Method used

The Girard-Hutchinson stochastic estimation method is used to calculate the trace of the kinship matrix using the following formula: tr(K2). The calculation process is optimized by increasing the number of iterations and decomposing the samples.

Benefits of technology

With increasing sample size, especially at the million level, it achieves efficient heritability estimation, avoids sample loss, maintains computational accuracy, and reduces computational complexity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121034402A_ABST
    Figure CN121034402A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of biological information, in particular to a genetic rate estimation method and system for sequencing data, a computer device and a computer program product. And representing phenotypic data of the concerned character as y = X beta + e, calculating a matrix trace of a genetic relationship matrix by adopting a Girard-Hutchinson random estimation method, and estimating the genetic rate according to a formula when judging that the sequencing data are from the same data source. The invention provides an optimized heredity estimation method, through a controllable random algorithm, on the premise of balancing calculation precision and calculation amount, the heredity can be solved under sample amount increase, especially under a million-level sample scale, and especially under a moment estimation framework, full-sample calculation can be used, large-scale data can be efficiently processed, sample loss is avoided, and the calculation efficiency is improved. And calculation precision is maintained.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of bioinformatics technology, and in particular, to a heritability estimation method, a system, a computer device, and a computer program product for sequencing data. Background Art

[0002] Heritability is a core indicator for quantifying the genetic contribution of phenotypic variation. The heritability of single nucleotide polymorphisms (SNPs) refers to the degree of contribution of single nucleotide polymorphisms to the genetic variation of complex traits, that is, what proportion of the variation of complex traits is determined by genetic factors. Heritability estimation is a conventional analysis task in statistical genetics.

[0003] One method for estimating heritability in the prior art is the heritability estimation method based on the linear mixed model (LMM). The heritability is obtained through maximum likelihood estimation (MLE). The calculation bottleneck lies in calculating where K is the kinship matrix between n individuals based on m genetic markers, and its computational complexity is With the emergence of samples in the millions, the computational cost of the K matrix becomes extremely large. For the current heritability estimation method based on the linear mixed model, the upper limit of the samples for estimating heritability is generally around 50,000. When the sample size expands to the level of hundreds of thousands or even millions, the operations of calculating, storing, and inverting the K matrix will cause memory overflow and depletion of computing resources, severely restricting the application of this method in samples of the million level.

[0004] Another method for estimating heritability in the prior art is the method of moment (MoM). The bottleneck of the method of moment lies in calculating the trace tr(K 2 ). The conventional amount of calculation also requires calculating K first. Then, through an approximation calculation strategy, the calculation process of tr(K 2 ) can be transformed into a way of reducing the sample size (subsampling) (n1 < n) to estimate tr(K 2 ), so as to obtain an approximate calculation. Although the amount of calculation can be reduced, it brings a decrease in calculation accuracy.

[0005] The increasing computational demands associated with larger sample sizes are a common phenomenon. With the rapid development of high-throughput sequencing technology, especially with the emergence of millions of samples in the biomedical field, carrying tens of millions of SNP data, traditional methods for calculating heritability face significant challenges when sample sizes reach biobank scales (n≥50,000). Traditional heritability estimation methods are limited by the trade-off between computational complexity and sample size, making them ill-suited for processing millions of data points. Summary of the Invention

[0006] To address the above problems, the purpose of this invention is to provide a method for estimating the heritability of sequencing data;

[0007] Another objective of this invention is to provide a system for estimating the heritability of sequencing data.

[0008] A method for estimating the heritability of sequencing data, comprising:

[0009] Step S1: Express the phenotypic data of the trait of interest as y = Xβ + e;

[0010] Where y represents the standardized phenotype of the trait of interest; X is the standardized individual genotype matrix, an n×m matrix, where n is the sample size of the sequencing data and m is the number of biallelic markers; β is the genetic effect vector. I m It is an m×m identity matrix, h 2 e is the heritability of phenotype y; e is the residual. I n It is an n×n identity matrix. It is the residual variance;

[0011] Step S2, using the Girard-Hutchinson stochastic estimation method, calculate the matrix trace tr(K) of the kinship matrix based on the following formula. 2 ):

[0012] E(L 2,B )=tr(K 2 );

[0013] Where z ~ N(0,1), z b Let n be a vector of length n;

[0014] E(L 2,B ) for L 2,B The expected value is B, where B is the number of iterations. Given an n×n kinship matrix, i refers to the i-th individual, and j refers to the j-th individual;

[0015] Step S3: When the sequencing data are determined to come from the same data source, the heritability is estimated according to the following formula.

[0016]

[0017] Where, m e For the number of valid tags,

[0018] The estimated variance is

[0019]

[0020] The heritability estimation method for sequencing data described in this invention The data source is divided into two independent parts: a molecule and a denominator. The molecule contains phenotypic information, while the denominator comes from a genetically related population. Both the molecule and denominator are distributed data sources. In step S3, when determining that the sequencing data comes from a distributed data source, the heritability is estimated using the following formula:

[0021] in, 'n' represents the sample size of the data source in the denominator, and 'n' represents the sample size of the data source in the numerator. in in It comes from the genotype of the denominator. The data matrix contains m columns of biallelic genes that are consistent with the X matrix of the molecular part.

[0022] but The estimated variance is

[0023]

[0024] in,

[0025]

[0026] The heritability estimation method for sequencing data described in this invention, wherein the sample size of the sequencing data in step S1 is greater than 10 5 The number of iterations is greater than or equal to 20, and η is 0.1 or 0.05.

[0027] The formulas for estimating the heritability and residual variance of phenotype y are as follows:

[0028]

[0029] The heritability estimation method for sequencing data described in this invention determines the number of iterations using the following formula:

[0030]

[0031] in,

[0032]

[0033] η takes the value 0.05 or 0.1.

[0034] tr(K 4 The trace of the fourth-order kinship matrix is ​​estimated using the following formula:

[0035]

[0036] A system for estimating the heritability of sequencing data, comprising,

[0037] The data standardization module represents the phenotypic data of the traits of interest as y = Xβ + e;

[0038] Where y represents the standardized phenotype of the trait of interest; X is the standardized individual genotype matrix, an n×m matrix, where n is the sample size of the sequencing data and m is the number of biallelic markers; β is the genetic effect vector. I m It is an m×m identity matrix, h 2 e is the heritability of phenotype y; e is the residual. I n It is an n×n identity matrix. It is the residual variance;

[0039] The matrix trace calculation module, connected to the data standardization module, uses the Girard-Hutchinson stochastic estimation method to calculate the matrix trace tr(K) of the kinship matrix based on the following formula. 2 ):

[0040] E(L 2,B )=tr(K 2 );

[0041] Where z ~ N(0,1), z b Let n be a vector of length n;

[0042] E(L 2,B ) for L 2,B The expected value is B, where B is the number of iterations. Given an n×n kinship matrix, i refers to the i-th individual, and j refers to the j-th individual;

[0043] The heritability estimation module, connected to the data standardization module and the matrix trace calculation module, estimates the heritability based on the following formula when the sequencing data comes from the same data source.

[0044]

[0045] Where, m e For the number of valid tags,

[0046] The estimated variance is

[0047]

[0048] The heritability estimation system for sequencing data described in this invention The data source is divided into two independent parts: a numerator and a denominator. The numerator contains phenotypic information, while the denominator comes from a genetically related population. Both the numerator and denominator are from a distributed data source. When determining that the sequencing data comes from a distributed data source, the heritability estimation module estimates the heritability using the following formula:

[0049]

[0050] in, 'n' represents the sample size of the data source in the denominator, and 'n' represents the sample size of the data source in the numerator. in in It comes from the genotype of the denominator. The data matrix contains m columns of biallelic genes, and these columns are identical to the X matrix of the molecular part. The estimated variance is

[0051]

[0052] in,

[0053]

[0054] The heritability estimation system for sequencing data described in this invention has a sample size of more than 10. 5 The number of iterations is greater than or equal to 20, and η is 0.1 or 0.05.

[0055] The formulas for estimating the heritability and residual variance of phenotype y are as follows:

[0056]

[0057] The heritability estimation system for sequencing data described in this invention determines the number of iterations using the following formula:

[0058]

[0059] in,

[0060]

[0061] η takes the value 0.05 or 0.1.

[0062] tr(K 4 The trace of the fourth-order kinship matrix is ​​estimated using the following formula:

[0063]

[0064] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the heritability estimation method.

[0065] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the heritability estimation method.

[0066] Beneficial effects: This invention proposes an optimized heritability estimation method. Through a controllable stochastic algorithm, it can solve for heritability under the premise of balancing computational accuracy and computational cost, especially at the scale of millions of samples. In particular, under the moment estimation framework, it can use full sample calculation, efficiently process large-scale data, avoid sample loss, and maintain computational accuracy. Attached Figure Description

[0067] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0068] Figure 2 This is a schematic diagram of the system structure of the present invention;

[0069] Figure 3 The results are randomized estimates of the heritability of 81 traits. Detailed Implementation

[0070] 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.

[0071] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0072] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0073] Reference Figures 1 to 3 A method for estimating the heritability of sequencing data, comprising:

[0074] Step S1: Express the phenotypic data of the trait of interest as y = Xβ + e;

[0075] Where y represents the standardized phenotype of the trait of interest; X is the standardized individual genotype matrix, an n×m matrix, where n is the sample size of the sequencing data and m is the number of biallelic markers; β is the genetic effect vector. I m It is an m×m identity matrix, h 2 e is the heritability of phenotype y; e is the residual. I n It is an n×n identity matrix. It is the residual variance;

[0076] Step S2, using the Girard-Hutchinson stochastic estimation method, calculate the matrix trace tr(K) of the kinship matrix based on the following formula. 2 ):

[0077] E(L 2,B )=tr(K 2 );

[0078] Where z ~ N(0,1), z b Let n be a vector of length n;

[0079] E(L 2,B ) for L 2,B The expected value is B, where B is the number of iterations. Given an n×n kinship matrix, i refers to the i-th individual, and j refers to the j-th individual;

[0080] Step S3, when it is determined that the sequencing data comes from the same data source, the heritability is estimated according to the following formula

[0081]

[0082] where m e is the number of effective markers,

[0083] the estimated variance of

[0084]

[0085] The present invention proposes an optimized method for estimating heritability. Through a controllable random algorithm, on the premise of balancing the calculation accuracy and the amount of calculation, the heritability can be solved when the sample size increases, especially at the scale of millions of samples. Especially in the framework of moment estimation, the full sample can be used for calculation, efficiently processing large-scale data, avoiding sample loss, and maintaining the calculation accuracy.

[0086] Estimating heritability The numerator part in the formula of The computational time complexity is There is no computational bottleneck, and the amount of calculation is mainly concentrated in the calculation step of the denominator tr(K 2 ).

[0087] In the prior art, the calculation of tr(K 2 ) is based on the following formula The direct computational time complexity is For a large sample size, such as about 500,000 samples in the UK Biobank, this will bring an extremely high computational burden. The method of reducing the sampling rate is to extract n1 samples (n1 < n) from n and construct K1. From The computational complexity of However, the calculation accuracy is reduced accordingly, and there will be a situation of data mismatch. The present invention adopts the Girard-Hutchinson random estimation method to calculate the matrix trace tr(K 2 ) of the kinship matrix based on the following formula. The computational time complexity of tr(K 2 ) is reduced to

[0088] The method for estimating the heritability of the sequencing data described in the present invention The data source is divided into two independent parts: a molecule and a denominator. The molecule contains phenotypic information, while the denominator comes from a genetically related population. In practice, the two may be separated, forming a distributed data source. That is, the data sources for the molecule and the denominator are distributed data sources. In step S3, when determining that the sequencing data comes from a distributed data source, the heritability is estimated according to the following formula:

[0089] in, 'n' represents the sample size of the data source in the denominator, and 'n' represents the sample size of the data source in the numerator.

[0090] in It comes from the genotype of the denominator. The data matrix contains m columns of biallelic genes, and these columns are identical to the X matrix of the molecular part. The estimated variance is

[0091]

[0092] in,

[0093]

[0094] The heritability estimation method for sequencing data described in this invention, wherein the sample size of the sequencing data in step S1 is greater than 10 5 The number of iterations is greater than or equal to 20, and η is 0.1 or 0.05.

[0095] The formulas for estimating the heritability and residual variance of phenotype y are as follows:

[0096]

[0097] The heritability estimation method for sequencing data described in this invention determines the number of iterations using the following formula:

[0098]

[0099] in,

[0100]

[0101] η takes the value 0.05 or 0.1.

[0102] tr(K 4 The trace of the fourth-order kinship matrix is ​​estimated using the following formula:

[0103]

[0104] A larger number of iterations helps improve estimation accuracy and reduce variance. Controlling the number of iterations is important both theoretically and practically when dealing with extremely large datasets. The number of iterations can range from 10 to 200 or even higher. Preferably, the number of iterations is greater than or equal to 20.

[0105] Reference Figure 2 A system for estimating the heritability of sequencing data, comprising,

[0106] Data standardization module 1 represents the phenotypic data of the traits of interest as y = Xβ + e;

[0107] Where y represents the standardized phenotype of the trait of interest; X is the standardized individual genotype matrix, an n×m matrix, where n is the sample size of the sequencing data and m is the number of biallelic markers; β is the genetic effect vector. I m It is an m×m identity matrix, h 2 e is the heritability of phenotype y; e is the residual. I n It is an n×n identity matrix. It is the residual variance;

[0108] Matrix trace calculation module 2, connected to the data standardization module 1, calculates the matrix trace tr(K) of the kinship matrix using the Girard-Hutchinson stochastic estimation method based on the following formula. 2 ):

[0109]

[0110] Where z ~ N(0, 1), z b Let n be a vector of length n;

[0111] E(L 2,B ) for L 2,B The expected value is B, where B is the number of iterations. Given an n×n kinship matrix, i refers to the i-th individual, and j refers to the j-th individual;

[0112] Heritability estimation module 3, connected to data standardization module 1 and matrix trace calculation module 2, estimates heritability based on the following formula when the sequencing data comes from the same data source.

[0113]

[0114] Where, m e For the number of valid tags,

[0115] The estimated variance is

[0116]

[0117] The heritability estimation system for sequencing data described in this invention The data source is divided into two independent parts: a numerator and a denominator. The numerator contains phenotypic information, while the denominator comes from a genetically related population. In practice, these two parts may be separated, forming a distributed data source. That is, the data sources for both the numerator and denominator are distributed data sources. When determining whether the sequencing data comes from a distributed data source, the heritability estimation module estimates the heritability using the following formula:

[0118] in, 'n' represents the sample size of the data source in the denominator, and 'n' represents the sample size of the data source in the numerator.

[0119] in in It comes from the genotype of the denominator. The data matrix contains m columns of biallelic genes, and these columns are identical to the X matrix of the molecular part. The estimated variance is

[0120]

[0121] in,

[0122]

[0123] The heritability estimation system for sequencing data described in this invention has a sample size of more than 10. 5 The number of iterations is greater than or equal to 20, and η is 0.1 or 0.05.

[0124] The formulas for estimating the heritability and residual variance of phenotype y are as follows:

[0125]

[0126] The heritability estimation system for sequencing data described in this invention determines the number of iterations using the following formula:

[0127]

[0128] in,

[0129]

[0130] η takes the value 0.05 or 0.1.

[0131] tr(K 4 The trace of the fourth-order kinship matrix is ​​estimated using the following formula:

[0132]

[0133] tr(K 4 ) is a key parameter that determines sampling performance.

[0134] The present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the heritability estimation method.

[0135] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the heritability estimation method.

[0136] Figure 3 This paper presents the randomized estimation results of the heritability of 81 traits. Real-world data were tested using 292,223 unrelated British individuals (field 22021) from the UK Biobank. Selection criteria included: Minor Allele Frequency (MAF) > 0.01, deletion rate < 0.05, Hardy-Weinberg proportion test p-value > 1e-6, and genotype call rate > 0.95. Finally, 525,460 autosomal SNPs were retained for analysis. The heritability of the 81 quantitative traits was estimated, with B0 = 10 as the initial iteration number, increasing in step sizes of 10. Subsequently, η and m were estimated. e h 2 Until 81 traits The average value was 0.0518, which is very close to the preset threshold. The data shows that in a biobank data test of nearly 300,000 unrelated independent individuals, a single estimation took an average of about 4.5 hours to complete using 10 CPUs.

[0137] The description and accompanying drawings provide typical embodiments of specific structures for specific implementations. Other modifications are possible based on the spirit of the invention. While the above-described invention presents preferred embodiments, these are not intended to be limiting.

[0138] For those skilled in the art, various changes and modifications will undoubtedly be apparent after reading the above description. Therefore, the appended claims should be construed as covering all changes and modifications that encompass the true intent and scope of the invention. Any and all equivalent scope and content within the scope of the claims should be considered to remain within the intent and scope of the invention.

Claims

1. A method for estimating the heritability of sequencing data, characterized in that, include: Step S1: Express the phenotypic data of the trait of interest as y = Xβ + e; Where y represents the standardized phenotype of the trait of interest; X is the standardized individual genotype matrix, an n×m matrix, where n is the sample size of the sequencing data and m is the number of biallelic markers; β is the genetic effect vector. I m It is an m×m identity matrix, h 2 e is the heritability of phenotype y; e is the residual. I n It is an n×n identity matrix. It is the residual variance; Step S2, using the Girard-Hutchinson stochastic estimation method, calculate the matrix trace tr(K) of the kinship matrix based on the following formula. 2 ): E(L 2,B )=tr(K 2 ); Where z ~ N(0, 1), z b Let n be a vector of length n; E(L 2,B ) for L 2,B The expected value is B, where B is the number of iterations. Given an n×n kinship matrix, i refers to the i-th individual, and j refers to the j-th individual; Step S3: When the sequencing data are determined to come from the same data source, the heritability is estimated according to the following formula. Where, m e For the number of valid tags, The estimated variance is 2. The method for estimating the heritability of sequencing data according to claim 1, characterized in that, The data source is divided into two independent parts: a molecule and a denominator. The molecule contains phenotypic information, while the denominator comes from a genetically related population. Both the molecule and denominator are distributed data sources. In step S3, when determining that the sequencing data comes from a distributed data source, the heritability is estimated using the following formula: in, 'n' represents the sample size of the data source in the denominator, and 'n' represents the sample size of the data source in the numerator. in It comes from the genotype of the denominator. The data matrix contains m columns of biallelic genes, and these columns are identical to the X matrix of the molecular part. The estimated variance is in, 3. The method for estimating the heritability of sequencing data according to claim 1, characterized in that, The sample size of the sequencing data in step S1 is greater than 10. 5 The number of iterations is greater than or equal to 20, and η is 0.1 or 0.

05. The formulas for estimating the heritability and residual variance of phenotype y are as follows:

4. The method for estimating the heritability of sequencing data according to claim 1, characterized in that, The number of iterations is determined by the following formula: in, η takes the value 0.05 or 0.

1. tr(K 4 The trace of the fourth-order kinship matrix is ​​estimated using the following formula:

5. A system for estimating the heritability of sequencing data, characterized in that, include, The data standardization module represents the phenotypic data of the traits of interest as y = Xβ + e; Where y represents the standardized phenotype of the trait of interest; X is the standardized individual genotype matrix, an n×m matrix, where n is the sample size of the sequencing data and m is the number of biallelic markers; β is the genetic effect vector. I m It is an m×m identity matrix, h 2 e is the heritability of phenotype y; e is the residual. I n It is an n×n identity matrix. It is the residual variance; The matrix trace calculation module, connected to the data standardization module, uses the Girard-Hutchinson stochastic estimation method to calculate the matrix trace tr(K) of the kinship matrix based on the following formula. 2 ): E(L 2,B )=tr(K 2 ); Where z ~ N(0, 1), z b Let n be a vector of length n; E(L 2,B ) for L 2,B The expected value is B, where B is the number of iterations. Given an n×n kinship matrix, i refers to the i-th individual, and j refers to the j-th individual; The heritability estimation module, connected to the data standardization module and the matrix trace calculation module, estimates the heritability based on the following formula when the sequencing data comes from the same data source. Where, m e For the number of valid tags, The estimated variance is 6. The heritability estimation system for sequencing data according to claim 5, characterized in that, The data source is divided into two independent parts: a numerator and a denominator. The numerator contains phenotypic information, while the denominator comes from a genetically related population. Both the numerator and denominator are from a distributed data source. When determining that the sequencing data comes from a distributed data source, the heritability estimation module estimates the heritability using the following formula: in, 'n' represents the sample size of the data source in the denominator, and 'n' represents the sample size of the data source in the numerator. in in It comes from the genotype of the denominator. The data matrix contains m columns of biallelic genes, and these columns are identical to the X matrix of the molecular part. The estimated variance is in, 7. The heritability estimation system for sequencing data according to claim 5, characterized in that, The sample size of the sequencing data is greater than 10. 5 The number of iterations is greater than or equal to 20, and η is 0.1 or 0.

05. The formulas for estimating the heritability and residual variance of phenotype y are as follows:

8. The heritability estimation system for sequencing data according to claim 5, characterized in that, The number of iterations is determined by the following formula: in, η takes the value 0.05 or 0.

1. tr(K 4 The trace of the fourth-order kinship matrix is ​​estimated using the following formula:

9. A computer device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the heritability estimation method according to any one of claims 1-5.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the heritability estimation method according to any one of claims 1-5.