Construction method and application of high-density genotype interpolation panel for dairy cows in southern China
By constructing a high-density genotype interpolation panel in southern China, the problem of low interpolation accuracy of southern China's dairy cows in the existing technology is solved, efficient and accurate genotype interpolation is achieved, and sequencing costs are reduced, and economical and practical solutions are provided for genetic evaluation and breeding.
Patent Information
- Application Number
- CN202411630793.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-15
- Publication Date
- 2025-03-07
AI Technical Summary
The existing high-density genotype interpolation reference panels are mainly based on foreign dairy cow breeds, and cannot effectively solve the problem of interpolation accuracy in the southern Chinese dairy cow population, resulting in limited analysis such as genetic evaluation.
A high-density genotype interpolation panel for Chinese Southern dairy cows was constructed. By collecting 50K SNP chip data and hair samples from cows far from their kinship with Chinese Southern dairy cows, genotyping detection and interpolation reference panel were constructed. Data quality control and interpolation operations were used for software such as Beagle v5.4 and PLINK v1.9, and the interpolation process was optimized to improve accuracy.
It improves the genetic similarity between the reference population and the target population, significantly improves the accuracy and efficiency of imputation, especially in groups with close kinship, reduces the cost of high coverage sequencing, and provides a more economical solution for large-scale genetic assessment and breeding.
Smart Images
Figure CN119580844A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a construction method and application of a high-density genotype interpolation panel for dairy cows in southern China, and belongs to the field of genotype interpolation panels. Background Art
[0002] With the development of sequencing technology, whole genome sequencing (WGS) has become the best choice for population genetic evaluation. However, large-scale high-coverage resequencing of livestock is expensive. Genotype interpolation technology can use reference panels to predict missing genotypes or untyped markers in low-density chips to solve this problem.
[0003] The accuracy of genotype interpolation is affected by many factors, such as the number of animals in the reference population, interpolation software, etc. Previous studies have found that the genetic structure of the reference population and the target population will significantly affect the interpolation accuracy. The more similar the genetic structure of the two is, the higher the interpolation accuracy.
[0004] At the same time, genotype interpolation technology can be divided into one-step interpolation and two-step interpolation according to the interpolation steps. One-step interpolation refers to the direct inference of low-density genotyping results to the whole genome sequencing level, and two-step interpolation refers to the process of inferring genotypes from low density to high density (HD) and finally to the whole genome sequencing level. In terms of interpolation effect, the two-step method is more accurate than the one-step method and has become the mainstream interpolation method.
[0005] Existing high-density genotype interpolation reference panels are usually constructed based on foreign dairy cattle breeds, and the genetic background of the reference population is mostly European or North American dairy cows. Using the high-density markers in the genotype chip, low-density chip data is interpolated in a two-step manner through interpolation software (such as Beagle or Minimac, etc.), thereby achieving full genome coverage. Due to the high cost of large-scale high-coverage resequencing of livestock, genotype interpolation technology is an effective solution, but the accuracy of interpolation is affected by many factors, especially the genetic similarity between the reference population and the target population and the interpolation method. In the southern Chinese dairy cattle population, there is currently a lack of a public high-density interpolation platform for this population, which will significantly affect the accuracy of interpolation, thereby limiting comprehensive analysis such as genetic evaluation. Summary of the invention
[0006] Purpose of the invention: The technical problem to be solved by the present invention is to provide a method for constructing a high-density genotype interpolation panel for dairy cows in southern China and its application.
[0007] Technical solution: To solve the above technical problems, the present invention provides a method for constructing a high-density genotype interpolation panel for dairy cows in southern China, comprising the following steps:
[0008] (1) Chip data processing: 50K SNP chip data and hair samples of cattle were collected. The cattle were distantly related to southern Chinese dairy cows. Genotyping was performed on the hair samples, and the cattle ARS-UCD 1.2 assembled genome was used as the reference genome.
[0009] (2) Imputation reference panel: SNP capture pipeline of CattleGTEx was used and phased using Beagle v5.4;
[0010] (3) Quality control of raw data: PLINK v1.9 was used to perform preliminary screening and cleaning of SNP data, specifying the --snps-only just-acgt and --chr 1-29 parameters. --snps-only just-acgt ensures that only SNP sites with A, C, G, and T bases are retained, and other types of markers are removed; --chr 1-29 limits the analyzed chromosome range to autosomes 1 to 29 of cattle, and removes duplicate SNP sites;
[0011] (4) Data correction: The ARS-UCD 1.2 version of the cattle genome was used as the reference genome to correct the reference alleles and alternative alleles;
[0012] (5) Interpolation panel construction: The data from different pastures were merged into a single dataset using the --merge-list function of PLINK v1.9. The merged data were converted into VCF file format and phased using Beagle v5.4.
[0013] The hair sample collection method in step (1) is to collect hair from the base of the cow's tail, and each individual provides at least 50 hairs with hair follicles to ensure sufficient DNA for subsequent genotyping.
[0014] The correction method in step (4) is to make the ref and alt allele definitions of all sites accurate and consistent based on the ARS-UCD 1.2 genome sequence, ensuring that the data has consistent allele identification between different individuals and sources.
[0015] Among them, the construction method also includes the optimization of the interpolation process, including the following steps: first, the 50K SNP chip data of Holstein cattle in southern China is interpolated to a high-density level using a high-density genotype interpolation panel for dairy cows in southern China; VCFtools v0.1.17 is used to perform quality control on the HD data, and the quality control conditions are a maximum missing rate of 5% and a minor allele frequency range of 1%-99% to ensure high-quality interpolation data; after quality control, the HD data is further interpolated to the WGS level using the CattleGTEx whole genome sequencing interpolation panel.
[0016] The invention also provides application of the construction method in genetic analysis and breeding.
[0017] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: 1. For the dairy cow population in southern China, the present invention constructs a high-density genotype interpolation panel for dairy cows in southern China, improves the genetic similarity between the reference population and the target population, solves the problem of decreased interpolation accuracy due to genetic differences, and effectively improves the accuracy and efficiency of interpolation, especially for groups with closer kinship, which show higher accuracy; 2. The data used in the interpolation process include 10K and 50K SNP chip data show that the interpolation panel has a wide adaptability to different data types and can maintain consistent performance in different types of genetic data, verifying the reliability and practicality of the interpolation panel; 3. Compared with the one-step interpolation technology, through the two-step interpolation process (first interpolation to a high-density level, and then interpolation to the whole genome sequencing level), this method of gradually improving the resolution can more finely complete the data, reduce interpolation errors, and improve the interpolation effect; 4. Based on this optimized interpolation process and algorithm, the cost of high-coverage sequencing can be significantly reduced, while maintaining high interpolation accuracy and efficiency, providing a more economical solution for large-scale genetic evaluation, breeding decisions and disease association studies; 5. This method provides an effective solution for the optimization of interpolation strategies. Through the use of high-density panels and a two-step interpolation process, the integrity and accuracy of the interpolation results are improved, and the missing information in the genetic data can be better dealt with, providing a reliable scientific basis for further genetic analysis and breeding, and providing technical support for promoting genetic research and breeding practices in regional dairy cattle populations. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 is the average interpolation accuracy of each chromosome of different populations in the 10K dataset;
[0019] Figure 2 is the average imputation accuracy of each chromosome of different populations in the 50K dataset;
[0020] Figure 3 is the average imputation accuracy of all chromosomes in different populations in the same dataset;
[0021] Figure 4 The average imputation accuracy per chromosome for the one-step and two-step methods;
[0022] Figure 5 The distribution of SNP numbers within the accuracy interval of one-step and two-step imputation methods. DETAILED DESCRIPTION
[0023] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.
[0024] 1. Construction of interpolation data platform
[0025] (1) Chip data processing
[0026] The present invention collected 50K SNP chip data of 999 European Holstein cattle (distantly related) from public databases. At the same time, 12,417 individuals of Jersey and Holstein cattle were selected from 8 large-scale ranches in southern China, of which 999 Holstein cattle were used to verify the population, and the remaining 11,418 cattle were included in the construction of a high-density genotype interpolation panel for dairy cows in southern China. The sample collection method is strict, and all samples are collected from the hair at the base of the cow's tail to ensure that each individual provides at least 50 hairs with hair follicles to ensure sufficient DNA for subsequent genotyping. The collected hair samples were sent to Newgene Biotech (Shanghai) Co., Ltd. and genotyped using the GGPBovine 100K SNP Chips chip, in which the bovine ARS-UCD 1.2 assembled genome was used as a reference genome.
[0027] (2) Interpolation reference panel
[0028] The whole genome sequencing horizontal interpolation panel used in the present invention comes from CattleGTEx, which integrates WGS data from multiple sources, with a total of 8,886 individuals and 192 breeds, of which the main breeds are: 906 Holstein cattle, 694 Jersey cattle, 573 Angus cattle, 326 German black spotted cattle, and 322 Simmental cattle. Its construction process is to use the SNP capture process of CattleGTEx and use Beagle v5.4 for phasing.
[0029] 2. Analysis methods and models
[0030] (1) Interpolation panel construction
[0031] This paper selected 11,418 cattle from 8 large-scale ranches in southern China and included them in the interpolation panel construction process. Since these data come from different ranches, in order to ensure data quality and consistency, the original data needs to be quality controlled. The quality control steps include the following:
[0032] First, PLINK v1.9 was used to perform preliminary screening and cleaning of SNP data. The --snps-only just-acgt and --chr 1-29 parameters were specified, where --snps-only just-acgt ensured that only SNP sites with A, C, G, and T bases were retained and other types of markers were removed; --chr 1-29 limited the chromosome range of the analysis to autosomes 1 to 29 of cattle. In order to further improve the data quality, duplicate SNP sites were removed to avoid inaccurate results due to redundant information in subsequent analysis.
[0033] After quality control is completed, the ARS-UCD 1.2 version of the cattle genome is used as the reference genome to calibrate the reference allele (ref) and alternative allele (alt). That is, according to the ARS-UCD 1.2 genome sequence, ensure that the ref and alt allele definitions of all sites are accurate and consistent. Ensure that the data has consistent allele identification between different individuals and sources.
[0034] After data correction, the data from the eight pastures were merged into a single dataset using the --merge-list function of PLINK v1.9, and the merged data were converted into VCF (Variant Call Format) file format for subsequent imputation analysis. Subsequently, Beagle v5.4 was used to perform phasing on the merged dataset, that is, to determine the exact position and arrangement order of alleles on the chromosome to support further construction and performance evaluation of the imputation panel.
[0035] (2) Optimization of interpolation process
[0036] In order to optimize the interpolation process of SNP chip data, this study designed and compared two interpolation schemes to analyze the impact of different processes on interpolation accuracy. The two schemes used different interpolation strategies and eventually interpolated the 50K SNP chip data of Holstein cattle in southern China to the whole genome sequencing (WGS) level. The specific process is as follows:
[0037] Solution 1: One-step interpolation
[0038] In the first approach, the CattleGTEx whole genome sequencing imputation panel and Beagle v5.4 software were used to directly impute the 50K SNP chip data of Holstein cattle in southern China to the whole genome sequencing (WGS) level.
[0039] Solution 2: Two-step interpolation
[0040] Scheme 2 adopted a step-by-step interpolation strategy. First, the 50K SNP chip data of Holstein cattle in southern China were interpolated to the high-density (HD) level using the high-density genotype interpolation panel of southern Chinese dairy cows. Next, VCFtools v0.1.17 was used to perform quality control on the HD data, with the quality control conditions of a maximum missing rate of 5% and a minor allele frequency range of 1%-99% to ensure high-quality interpolation data. After quality control, the HD data were further interpolated to the WGS level using the CattleGTEx whole genome sequencing interpolation panel.
[0041] The performance of the two processes at different imputation stages can be evaluated by comparing the imputation results of Scheme 1 and Scheme 2. To ensure the consistency of the processes, the imputation processes of the two schemes were run with default options and uninformative linkage maps (1 cM / 1 Mb).
[0042] (3) Evaluating interpolation panel performance
[0043] In order to systematically evaluate the performance of the interpolation panel, the present invention adopts multi-level interpolation test and comparative analysis. First, 50K SNP chip data of 999 Holstein cattle in Europe and 100K SNP chip data of 999 Holstein cattle in southern China were processed. PLINK v1.9 software was used to downsample to 10K and 50K SNP sites based on LD (linkage disequilibrium) to construct test data sets of different densities. Then, Beagle v5.4 was used to phase and interpolate the SNP chip data of all autosomes (chr1-29).
[0044] The interpolation panel evaluation of the present invention is divided into the following aspects, aiming to fully understand the impact of different data types and kinship on interpolation accuracy and quality, and provide a scientific basis for optimizing interpolation strategies:
[0045] ① Interpolation performance analysis of chips with different densities
[0046] In this section, the accuracy of imputation to high density (HD) level was evaluated at different chip densities of 10K and 50K. The 10K SNP and 50K SNP chip data of European Holstein and southern Chinese Holstein were imputed to HD level using the high-density genotype imputation panel of southern Chinese dairy cows. By comparing the imputation results of chips with different densities, the imputation effect under different data densities was analyzed.
[0047] ②The impact of kinship on interpolation accuracy
[0048] This part mainly examines how the genetic relationship between individuals and the reference panel affects the interpolation accuracy. The interpolation accuracy of European Holstein cattle and southern Chinese Holstein cattle to HD level is compared under 10K and 50K SNP chip data, the impact of genetic relationship on the interpolation effect is analyzed, and the role of the closeness of genetic relationship in interpolation accuracy is confirmed.
[0049] ③Comparison of interpolation strategies
[0050] To further optimize the imputation process, the effects of different imputation strategies were compared. First, the 50K SNP chip data of Holstein cattle in southern China were directly imputed to the whole genome sequencing (WGS) level. Then, a two-step method was used: the 50K data were first imputed to the HD level using a high-density genotype imputation panel of dairy cows in southern China, and then imputed to the WGS level after quality control. By comparing the imputation accuracy of the two strategies, it was evaluated which imputation process could improve accuracy more effectively.
[0051] The above multi-level evaluation process ensures a comprehensive understanding of the performance of the imputation panel under different conditions. Through these evaluation results, the impact of kinship and chip density on imputation accuracy is quantified, and the effects of different imputation strategies at different levels are verified, providing theoretical support and data basis for the design of more efficient imputation strategies in the future.
[0052] The present invention uses the constructed high-density genotype interpolation panel of southern dairy cows to interpolate the 10K and 50K SNP chip data to the high-density (HD) level. When comparing the interpolation performance of different groups, it was found that the interpolation accuracy of the southern Chinese group on each autosome (chr1-29) was higher than that of the European group ( Figure 1 , Figure 2 ). This result can be attributed to the closer relationship between the southern Chinese population and the imputation panel, which improves the accuracy of allele prediction. Subsequent analysis of the average imputation accuracy of all autosomes (chr1-29) showed that the southern Chinese population was better than the European population in overall accuracy. This phenomenon verifies that kinship has a significant impact on imputation accuracy: populations with closer kinship tend to achieve higher imputation accuracy.
[0053] When using the high-density genotype imputation panel for dairy cows in southern China to impute 50K SNP chip data, the imputation accuracy of the southern Chinese population exceeded 0.95, while the accuracy of the European population was slightly lower, but the overall imputation accuracy exceeded 0.8 ( Figure 3 ). These results suggest that the high-density genotype imputation panel for dairy cattle in southern China can provide a high level of imputation accuracy, especially in closely related groups.
[0054] In addition, to evaluate the impact of different interpolation processes on the interpolation effect, the present invention uses 50K SNP chip data of Holstein cattle in southern China as input data to compare the accuracy of the two-step interpolation process and the one-step interpolation process. The two-step interpolation process, that is, first interpolating the data to a high-density level and then further interpolating to the whole genome sequencing level, has a higher interpolation accuracy than the one-step method ( Figure 4 ) and the number of SNPs with high imputation accuracy increased significantly in the two-step process, while the number of SNPs with low imputation accuracy decreased ( Figure 5 ). This shows that the step-by-step interpolation process effectively improves the accuracy of the final interpolation by gradually improving the density and quality of the data.
[0055] In summary, the high-density genotype imputation panel for dairy cows in southern China performed well in improving imputation accuracy, especially in closely related groups. The two-step imputation process is superior to the one-step method, providing an important basis for optimizing the imputation strategy.
Claims
1. A method for constructing a high-density genotype interpolation panel for dairy cows in southern China, characterized in that: The following steps are involved: (1) Chip data processing: 50K SNP chip data and hair samples of cattle were collected. The cattle were distantly related to southern Chinese dairy cows. Genotyping tests were performed on hair samples, where the bovine ARS-UCD 1.2 assembled genome was used as the reference genome; (2) Imputation reference panel: SNP capture pipeline of CattleGTEx was used and phased using Beagle v5.4; (3) Quality control of raw data: PLINK v1.9 was used to perform preliminary screening and cleaning of SNP data, specifying the --snps-only just-acgt and --chr 1-29 parameters. --snps-only just-acgt ensures that only SNP sites with A, C, G, and T bases are retained, and other types of markers are removed; --chr 1-29 limits the analyzed chromosome range to autosomes 1 to 29 of cattle, and removes duplicate SNP sites; (4) Data correction: The ARS-UCD 1.2 version of the cattle genome was used as the reference genome to correct the reference alleles and alternative alleles; (5) Interpolation panel construction: The data from different pastures were merged into a single dataset using the --merge-list function of PLINK v1.
9. The merged data were converted into VCF file format and phased using Beagle v5.
4.
2. The construction method according to claim 1, characterized in that: The hair sample collection method in step (1) is to collect hair from the base of the cow's tail, with each individual providing at least 50 hairs with hair follicles to ensure sufficient DNA for subsequent genotyping.
3. The construction method according to claim 1, characterized in that: The correction method described in step (4) is to make the ref and alt allele definitions of all sites accurate and consistent based on the ARS-UCD 1.2 genome sequence, ensuring that the data has consistent allele identification between different individuals and sources.
4. The construction method according to claim 1, characterized in that: The construction method also includes optimization of the interpolation process, including the following steps: first, the 50K SNP chip data of Holstein cattle in southern China is interpolated to a high-density level using a high-density genotype interpolation panel for dairy cows in southern China; VCFtools v0.1.17 is used to perform quality control on the HD data, and the quality control conditions are a maximum missing rate of 5% and a minor allele frequency range of 1%-99% to ensure high-quality interpolation data; after quality control, the HD data is further interpolated to the WGS level using the CattleGTEx whole genome sequencing interpolation panel.
5. Application of the construction method according to claim 1 in genetic analysis and breeding.