Method for automatic construction of molecular pedigree of large-scale poultry population

Abstract

The application provides a method for automatically constructing a molecular pedigree of a large-scale poultry population, and the method comprises the following steps: 1) obtaining SNP molecular markers and supplementary information; 2) quality control of the SNP molecular markers; 3) calculating a relationship coefficient; 4) calculating a zero-allele sharing probability; 5) inferring the relationship; and 6) automatically constructing a pedigree. The method can automatically infer the relationship between individuals from genetic data, without relying on manual operation, greatly reducing errors in breeding records, and quickly and accurately generating a complete family pedigree in a large-scale population.

Description

Technical Field

[0001] This invention relates to the field of poultry breeding technology. Specifically, this application provides an automated method for constructing molecular pedigrees of large-scale poultry populations. Background Technology

[0002] A pedigree is a tool used to record the kinship relationships between individuals and their ancestors. It is widely used in many fields, especially in animal and plant breeding, genetic research, and forensic medicine. In poultry breeding, constructing accurate pedigrees is crucial for breeding decisions. Pedigrees help breeders trace an individual's lineage, identify superior genes, avoid inbreeding, and optimize the genetic diversity of poultry populations. Through pedigrees, breeders can understand the genetic background of each poultry individual, enabling selective breeding for specific traits and improving poultry production performance and health.

[0003] Currently, in the poultry farming industry, pedigrees are still primarily obtained through manual recording, even when combined with specific breeding methods such as monogamy or self-breeding. This method relies on breeders observing and recording the mating behavior and reproductive results of poultry during the breeding process. While this method is effective in small-scale breeding farms, as the size of the breeding population increases, manual recording becomes prone to errors, especially when it is difficult to accurately identify individual parents during breeding, thus affecting the accuracy and completeness of the pedigree. This makes traditional pedigree recording methods unreliable and inefficient in large-scale poultry breeding.

[0004] In recent years, the application of molecular marker technologies (such as SNP markers) has developed rapidly. Through whole-genome sequencing, breeders can directly obtain the genetic relationships between poultry individuals at the DNA level. Compared with traditional manual recording, molecular pedigree construction based on genetic information not only significantly improves the accuracy of pedigrees but also avoids interference from external environment, behavioral differences, and incomplete information in manual records. Summary of the Invention

[0005] Building upon the above, this invention further improves the efficiency of pedigree construction by employing molecular marker technology combined with automated algorithms to achieve automated pedigree construction for large-scale poultry populations. This method can automatically infer kinship relationships between individuals from genetic data without relying on manual operations, greatly reducing errors in breeding records, and enabling the rapid and accurate generation of complete family pedigrees even in large-scale populations.

[0006] The large-scale population pedigree construction method of this invention utilizes only the genetic information of the poultry population, as well as the sex and generational information of individuals. It does not involve other external information, such as external environmental information, social and behavioral data, or other phenotypic data. Therefore, regardless of whether the sample individuals come from different growth environments, have different social behaviors, or different body sizes and phenotypic expressions, this invention can construct the family pedigree of the target sample. It is worth noting that the large-scale population pedigree construction method mentioned in this invention is not limited to poultry populations. For other animal populations, whose genetic laws also conform to Mendel's first law (the law of segregation), this large-scale population pedigree construction method can also construct pedigrees for the target population.

[0007] The technical solution adopted in this invention is as follows: First, genomic DNA is extracted from the blood of the target population (including parents and offspring), and the sex and generational information of the sample individuals are recorded. Whole-genome sequencing is performed on the sample DNA, and SNP (Single Nucleotide Polymorphism) molecular markers are obtained through gene alignment analysis. Using the obtained SNP molecular markers, the kinship coefficient and the probability of zero identity by descent are calculated. Based on the genomic kinship coefficient, the kinship between sample individuals is inferred. Finally, using the sex and generational information of the sample individuals, a rapid and accurate family pedigree is constructed.

[0008] This invention provides an automated method for constructing molecular pedigrees of large-scale poultry populations, the method comprising the following steps:

[0009] 1) Acquisition of SNP molecular markers and supplementary information;

[0010] 2) Quality control of SNP molecular markers;

[0011] 3) Calculation of kinship coefficient;

[0012] 4) Calculation of the probability of zero allele sharing;

[0013] 5) Inference of kinship;

[0014] 6) Automated genealogy construction.

[0015] Further, step 1) includes:

[0016] 1-1) Collect blood samples from all male and female poultry retained for breeding in the selected sample population, as well as from all offspring, and extract genomic DNA;

[0017] 1-2) Collect the unique ID, gender information, and generational information for each sample individual;

[0018] 1-3) Perform whole-genome sequencing on each genomic DNA sample to obtain the genomic sequence data;

[0019] 1-4) Obtain the SNP molecular markers of the samples through comparative analysis.

[0020] Further, in steps 1-4), BWA software is used to align the sample SNP markers with the reference genome.

[0021] Further, step 2) includes:

[0022] 2-1) Genotype missing rate filtering: Set the genotype missing rate threshold to 0.05, that is, for each SNP molecular marker, if the missing rate (i.e. the proportion of missing genotypes) of the marker exceeds 5%, the SNP marker will be removed from the dataset.

[0023] 2-2) Minor allele frequency filtering: Set the minor allele frequency (MAF) threshold to 0.01, that is, for each SNP molecular marker, if the minor allele frequency is less than 1%, the SNP marker will be removed from the dataset.

[0024] 2-3) Linkage Disequilibrium Screening: A linkage disequilibrium (LD) screening strategy is used. Analysis is performed within a window containing 50 consecutive SNPs, with a step size of 5 SNPs. The window is moved forward by 5 SNPs each time it is moved, and the LD value (r) between each pair of SNPs within the window is calculated. 2 If the LD value (r) between the tag pairs 2 If the SNP value is greater than 0.2, then these SNPs with high linkage disequilibrium will be excluded.

[0025] Further, step 3) includes:

[0026] 3-1) Using the quality-controlled SNP data, construct a standardized genotype matrix M;

[0027] Given n individuals and m markers (such as SNP loci), the normalized genotype matrix M is an n×m matrix, where each element M ij This represents the genotype value of the i-th individual at the _i SNP locus. Genotype values ​​are typically encoded as 0, 1, and 2, where 0 represents homozygote (e.g., aa), 1 represents heterozygote (e.g., Aa), and 2 represents another homozygote (e.g., AA). Each element M in the genotype matrix M... ij Standardize in the following ways:

[0028]

[0029] In formula (1), g ij p is the genotype value of the i-th individual at the n-th SNP. j It is the minor allele frequency (MAF) of the j-th SNP;

[0030] This standardization process ensures that the genotype matrix has the same weights among different SNPs, thus preventing the construction of the genome relationship matrix from being affected by differences in allele frequencies.

[0031] 3-2) Calculate the genomic relationship matrix (G matrix) between individuals based on the standardized genotype matrix M;

[0032]

[0033] MM in formula (2) T This is a matrix product, which yields the genomic similarity between each pair of individuals, represented as the inner product of two normalized genotype vectors; p i The secondary allele frequency was used to calculate the normalization factor 2∑p. i (1-p i This factor represents the expected heterogeneity of all SNP sites. It ensures that the G matrix values ​​are within a reasonable range and reflect the relative genomic similarity among individuals.

[0034] In the genomic relationship matrix G, each element represents the genomic relationship coefficient, or kinship coefficient, between two individuals. The diagonal elements of the matrix (such as G...) ii This represents the relationship between each individual's genome and itself, generally close to 1. It reflects the self-similarity of an individual. Other elements in the matrix (such as G...) ij , where i≠j) represents the genomic relationship coefficient between different individuals. A larger value indicates a closer kinship between individuals; a smaller value indicates a more distant kinship.

[0035] Further, step 4) includes:

[0036] In kinship inference, IBD (Identity by Descent) indicates whether two individuals share one or two alleles at a certain locus, which are inherited from a common ancestor. IBD status can be:

[0037] Z=0: Zero allele sharing;

[0038] Z=1: Sharing one allele;

[0039] Z=2: Sharing two alleles;

[0040] IBS (Identity by State): Indicates whether two individuals share an allele at a certain locus, ignoring whether they share a common ancestor. The IBS state can be 0, 1, or 2, representing no sharing, sharing one allele, or sharing two alleles, respectively.

[0041] The probability of zero allele sharing, P(Z=0), can be calculated based on the frequency of IBS states and the theoretical distribution of alleles in the observed data, using Bayesian estimation or the method of moments. According to conditional probability, P(Z=0) can be expressed as:

[0042]

[0043] Where N(I=0) represents the number of times IBS=0 (i.e., alleles are not shared) observed at all SNP loci, which can be obtained by iterating through each SNP locus m and checking the genotypes of two individuals i and j at that locus. N(I=0|Z=0) is the expected number of times IBS=0 under the assumption that IBD=0. For allele A with frequency p and allele a with frequency 1-p, the probability that two individuals are different homozygotes (i.e., AA and aa) is:

[0044] Pr(IBS=0|Z=0)=2p 2 (1-p) 2 (4)

[0045] By summing the results at all points, we can obtain:

[0046]

[0047] in It is the estimated allele frequency of the m-th SNP locus.

[0048] Substituting Formula 5 into Formula 3 yields the value of P(Z=0). A higher P(Z=0) indicates that the two individuals are not closely related, while a lower P(Z=0) indicates a close kinship.

[0049] Further, step 5) includes:

[0050] In kinship inference, diagonal elements are not considered; only off-diagonal elements (i.e., genomic relationship coefficients between different individuals) are considered.

[0051] when Since P(Z=0)<0.1, it can be determined that the two individuals i and i are parent and child.

[0052] when Meanwhile, P(Z=0)∈(0.1,0.365), which indicates that individuals i and j are full siblings;

[0053] when at the same time It can be determined that these two individuals i and i are second-degree relatives. In the same generation, they are half-siblings (only one of their father and mother is the same). In the cross-generational relationship, if it is two generations, it is a grandparent-grandchild relationship. If it is one generation, it is a collateral relative relationship between two generations.

[0054] When the remaining cases and G ij When ≈0, it can be determined that the two individuals i and j are approximately unrelated.

[0055] Further, step 6) includes:

[0056] 6-1) Create an undirected graph to represent individuals and kinship relationships:

[0057] Create an undirected graph G using a graph structure, where nodes represent individuals, each with a unique ID, and node attributes including gender and generation information. Edges represent kinship relationships between individuals, with edge weights of G. ij The value represents the genetic similarity between individuals. Only when G... ij Value at Only when P(Z=0)<0.1 are the relationships between individuals added as edges to the graph.

[0058] 6-2) Parental inference to obtain family genealogy:

[0059] For all individuals in graph G, potential parents are selected. The generational information of these potential parents must be less than that of their offspring.

[0060] And G ij Value at Between these conditions, P(Z=0)<0.1. Further, based on gender information, the father must be male and the mother must be female.

[0061] Furthermore, the poultry is a chicken, duck, or goose, such as the Peking duck in the embodiment. Attached Figure Description

[0062] Figure 1 This is a technical roadmap for the present invention.

[0063] Figure 2 This is a graph showing the overall distribution of coefficients for different kinship types. The data represents Beijing duck populations from different genetic backgrounds involved in Examples 1 and 2.

[0064] Figure 3This is a probability distribution map of zero-shared alleles under different kinship types. The data are from Beijing duck populations with different genetic backgrounds involved in Example 1 and Example 2.

[0065] Figure 4 This is a scatter plot showing the distribution of kinship coefficients and the probability of zero-shared alleles under different kinship types. The horizontal axis Z0 represents the probability of zero-shared alleles, and the vertical axis GRM represents the kinship coefficient. The data are from Beijing duck populations under different genetic backgrounds involved in Examples 1 and 2. Detailed Implementation

[0066] The flowchart of the method for automated construction of large-scale poultry population pedigrees in this application is as follows: Figure 1 As shown.

[0067] Example 1: Construction of a 7-line pedigree of Peking ducks comprising three generations

[0068] This example collected 3648 Beijing ducks of strain 7, comprising three generations: generation 13 (528 birds, 133 males, 395 females), generation 14 (1549 birds, 346 males, 1203 females), and generation 15 (1571 birds, 322 males, 1221 females, and 28 mixed-sex birds). Generation 15 is a descendant of generation 14, and generation 14 is a descendant of generation 13. The purpose of this example is to verify the invention's ability to rapidly and accurately automate pedigree construction in large-scale poultry populations of two or more generations (see details). Figure 2 ).

[0069] The steps are as follows:

[0070] (1) Acquisition of SNP molecular markers and supplementary information: Genomic DNA was extracted from blood samples of 3,648 individuals and whole-genome sequencing was performed. The SNP molecular markers of the samples were obtained through comparative analysis, and the unique ID, sex information and generation information of each individual sample were recorded.

[0071] (2) SNP molecular marker quality control: SNP loci with a genotype deletion rate exceeding 5% and a minor allele frequency below 1% were filtered out; linkage disequilibrium screening was performed with a window size of 50 and a step size of 5, and LD values ​​(r) were removed. 2 SNP markers with a value >0.2 were selected, and 365,860 SNP sites were ultimately retained for subsequent analysis;

[0072] (3) Calculation of kinship coefficient: Based on the quality-controlled SNP molecular markers, the genomic relationship coefficient G between individuals was calculated. ij ;

[0073] (4) Calculation of zero allele sharing probability: Based on the quality-controlled SNP molecular markers, the zero allele sharing probability P (Z = 0) between individuals is calculated;

[0074] (5) Kinship inference: The kinship between individuals is determined based on the kinship coefficient and the probability of zero allele sharing. At the same time, when P(Z=0)<0.1, the two individuals i and j are determined to be first-degree parent-child relationships.

[0075] (6) Automated genealogy construction:

[0076] Using a graph structure, accurate pedigrees are constructed by inferring kinship based on the results and combining generational and gender information.

[0077] Table 1, Table 2 and Figure 3 The results showed that rapid and accurate pedigree construction was achieved in a poultry population of 3648 individuals. The accuracy rate for paternal inference was 100% for the 1571 individuals in the 15th generation, and 94% for the maternal inference; the accuracy rate for paternal inference was 99.59% for the 14th generation, and 96.87% for the maternal inference. Further paternity testing was conducted on pairs of erroneous parentages, and the results showed that the parentages were consistent. Therefore, the pedigree records were actually incorrect. This was because the Beijing ducks used in this example followed the traditional crossbreeding model, where one male mates with multiple females, resulting in high accuracy for paternal records but a higher likelihood of error in maternal records.

[0078] Example 2: Pedigree construction of populations with different genetic backgrounds

[0079] In this example, a large-scale automated molecular pedigree construction experiment was conducted on three other Beijing duck populations with different genetic backgrounds.

[0080] The sample consisted of 1215 birds of the 6th generation, including the 5th and 6th generations, with the 6th generation being the offspring of the 5th generation. The 5th generation had 109 males and 299 females; the 6th generation had 191 males and 616 females.

[0081] A total of 2,534 birds were included in the 4-lineage sample, comprising generations 40, 41, and 42. Generation 42 was the offspring of generation 41, and generation 41 was the offspring of generation 40. Generation 40 consisted of 140 males and 351 females; generation 41 consisted of 278 males and 812 females; and generation 42 consisted of 209 males and 744 females.

[0082] A total of 2422 MA strain birds were sampled, including generations 6, 7, and 8. Generation 7 is the offspring of generation 6, and generation 8 is the offspring of generation 7. Generation 6 consisted of 123 males and 306 females; generation 7 consisted of 236 males and 803 females; and generation 8 consisted of 212 males and 742 females. (See details) Figure 2 )

[0083] The steps for automated pedigree construction of large-scale populations are as follows:

[0084] (1) Acquisition of SNP molecular markers and supplementary information: Blood samples were collected from different strains of the population, genomic DNA was extracted and whole genome sequencing was performed, and SNP molecular markers of the samples were obtained through comparative analysis. The unique ID, sex and generation information of each individual were recorded.

[0085] (2) SNP molecular marker quality control: SNP loci with a genotype deletion rate exceeding 5% and a minor allele frequency below 1% were filtered out; the window size was set to 50 and the step size to 5, and linkage disequilibrium values ​​(r) were removed. 2 SNP markers with a value greater than 0.2);

[0086] (3) Calculation of kinship coefficient: Based on the quality-controlled SNP molecular markers, the genomic relationship coefficient G between individuals was calculated. ij ;

[0087] (4) Calculation of zero allele sharing probability: Based on the quality-controlled SNP molecular markers, the zero allele sharing probability P (Z = 0) between individuals is calculated;

[0088] (5) Kinship inference: The kinship between individuals is determined based on the kinship coefficient and the probability of zero allele sharing. At the same time, when P(Z=0)<0.1, the two individuals i and i are determined to be first-degree parent-child relationships.

[0089] (6) Automated genealogy construction:

[0090] Using a graph structure, accurate pedigrees are constructed by inferring kinship based on the results and combining generational and gender information.

[0091] Table 1, Table 2, Figure 3 , Figure 4 The results showed that the accuracy rate of pedigree construction for the 6-line population was 100%, for the 4-line population it was 99.55%, and for the MA lineage population it was 98.13%. After conducting paternity testing on 0.45% of individuals with inference errors in the 4-line population and 1.87% in the MA lineage population, it was found that these individuals actually had pedigree record errors, indicating that the inference method used in this invention was completely correct.

[0092] Table 1 Threshold Classification for Kinship Inference

[0093]

[0094] Table 2. Statistical distribution of kinship coefficients among the 7-line, 6-line, 4-line, and MA-lineages.

[0095]

Claims

1. A method for automatically constructing molecular pedigrees of large-scale poultry populations, characterized in that, The method includes the following steps: 1) Acquisition of SNP molecular markers and supplementary information; 2) Quality control of SNP molecular markers; 3) Calculation of kinship coefficient; 4) Calculation of the probability of zero allele sharing; 5) Inference of kinship; 6) Automated pedigree construction; Step 1) includes: 1-1) Collect blood samples from all male and female poultry used for breeding in the poultry population and all offspring, and extract genomic DNA; 1-2) Collect the unique ID, gender information, and generational information for each individual in the sample; 1-3) Perform whole-genome sequencing on the genomic DNA of each sample to obtain genomic sequence data; 1-4) Obtain the SNP molecular markers of the samples through comparative analysis; Step 2) includes: 2-1) Genotype missing rate filtering: Set the genotype missing rate threshold to 0.

05. For each SNP molecular marker, if the missing rate exceeds 5%, the SNP marker will be removed from the dataset. 2-2) Minor allele frequency filtering: Set the minor allele frequency threshold to 0.

01. For each SNP molecular marker, if the minor allele frequency is less than 1%, the SNP marker will be removed from the dataset. 2-3) Linkage imbalance screening: The linkage imbalance screening strategy is used to analyze within a window containing 50 consecutive SNPs; the step size is set to 5 SNPs, and 5 SNPs are moved forward each time the window is moved, and the LD value r² between each pair of SNPs in the window is calculated; if the LD value r² between the pairs of labels is greater than 0.2, these SNPs with high linkage imbalance are excluded. Step 3) includes: 3-1) Using the quality-controlled SNP data obtained in step 2), construct a standardized genotype matrix M; It has Individuals and One marker, standardized genotype matrix for A matrix, where each element Indicates the first The individual in the first Genotype values ​​at each SNP locus; genotype values ​​are typically encoded as 0, 1, or 2, where 0 represents homozygote, 1 represents heterozygote, and 2 represents another homozygote. Genotype matrix. Each element in Standardize in the following ways: ， In formula (1), It is the first The individual in the first Genotype values ​​of each SNP, It is the secondary allele frequency of the j-th SNP; 3-2) Based on the standardized genotype matrix Calculate the genomic relationship matrix G between individuals; ， MM in formula (2) T It is a matrix product, which yields the genomic similarity between each pair of individuals, represented as the inner product of two normalized genotype vectors; The secondary allele frequency was used to calculate the normalization factor. , which represents the expected heterogeneity of all SNP sites; Step 4) includes: Based on the observed IBS state frequencies and theoretical allele distributions, the probability of zero allele sharing P (Z=0) is calculated using Bayesian estimation or the method of moments: ， in, This indicates that the observed alleles are not shared across all SNP loci. The number of times it is determined by traversing each SNP site. Check two individuals and The genotype at that locus is obtained; It is assumed In this case, the expected The number of times, for frequency alleles and frequency alleles The two individuals are different homozygotes. and The probability is: ， Accumulate at all sites to get: ， Among them is the first Allele frequency estimates for each SNP locus; Substituting formula (5) into formula (3) will yield the result. The value; Step 5) includes: In inferring kinship, elements on the diagonal are not considered; only elements off the diagonal are considered. At the same time It can be determined that these two individuals and It is a parent-child relationship; At the same time It can be determined that these two individuals and It is a relationship between all compatriots; At the same time It can be determined that these two individuals and It is a second-degree kinship. Within the same generation, it indicates half-siblings. Across generations, if it spans two generations, it indicates a grandparent-grandchild relationship. If it spans one generation, it indicates a collateral relative relationship between two generations. When the remaining situations and At that time, it can be determined that these two individuals and They are practically unrelated. Step 6) includes: 6-1) Create an undirected graph to represent individuals and kinship relationships: Create an undirected graph G using a graph structure, where nodes represent individuals, each with a unique ID, and node attributes include gender and generation information; edges represent kinship relationships between individuals, with edge weights of . The value represents the genetic similarity between individuals; only when... Value at Between, at the same time Only when the time is right, are the relationships between individuals added as edges to the graph; 6-2) Parental deduction to obtain family genealogy: For all individuals in graph G, potential parents are selected. The generational information of these potential parents must be less than that of their offspring. Value at Between, at the same time Further, based on gender information, the father and mother must be distinguished; the father must be a male individual and the mother must be a female individual.

2. The method according to claim 1, wherein in steps 1-4), BWA software is used to align the sample SNP markers with the reference genome.

3. The method according to claim 1, wherein the poultry is a chicken, duck or goose.

4. The method according to claim 2, wherein the poultry is a Beijing duck.

Images