SNP (Single Nucleotide Polymorphism) molecular marker combination for identifying Tianfu broiler chicken and application of SNP molecular marker combination
By fusing multiple machine learning algorithms, 20 highly discriminative SNP molecular marker combinations were selected and a Tianfu broiler breed identification model was constructed. This model solves the problems of low efficiency and insufficient accuracy in poultry breed identification in existing technologies, and achieves efficient and accurate Tianfu broiler breed identification and breeding support.
Patent Information
- Application Number
- CN202511978622.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-25
- Publication Date
- 2026-02-10
AI Technical Summary
Existing technologies for poultry breed identification have limitations such as strong subjectivity, susceptibility to environmental influences, and long cycles. Furthermore, they lack multi-algorithm fusion strategies, making it difficult to meet the needs of modern, efficient, and precise breeding. In particular, there are no systematic reports on the application of specific SNP combinations for Tianfu broiler chickens.
A multi-machine learning algorithm fusion method was used to screen 20 high-discrimination SNP molecular marker combinations to construct a Tianfu broiler breed identification model. A high-precision and high-efficiency breed identification model was constructed using 20 specific SNP sites. Whole genome resequencing was performed using an Illumina HiseqX10 instrument. The importance of SNPs was evaluated using multiple indicators, and the model was optimized using a support vector machine-radial basis function algorithm.
It achieves high accuracy and efficiency in identifying Tianfu broiler breeds, with an average accuracy rate of 97.58% and an average ROC_AUC of 96.04%, significantly reducing testing costs. It is robust and practical, and suitable for breed identification, genetic resource protection, and breeding practices.
Smart Images
Figure CN121496071A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of molecular biology, and in particular to a combination of SNP molecular markers for identifying Tianfu broiler chickens and its applications. Background Technology
[0002] Accurate identification of poultry breeds and protection of genetic resources are important aspects of modern livestock and poultry breeding. Traditional breed identification methods based on phenotypic characteristics (such as body size, feather color, and growth performance) have limitations such as strong subjectivity, susceptibility to environmental influences, and long cycles, making it difficult to meet the needs of modern efficient and precise breeding.
[0003] Machine learning algorithms have demonstrated great potential in genome selection, genotype-phenotype association analysis, and breed identification. Through feature selection and model optimization, machine learning can screen for highly discriminative locus combinations from massive amounts of SNPs, constructing efficient and robust classification models. However, existing studies mostly rely on single algorithms or a few methods, and have not fully utilized multi-algorithm fusion strategies to improve the comprehensiveness and reliability of SNP screening. Furthermore, there are no systematic reports on specific SNP combinations for Tianfu broiler chickens and their application in breed identification.
[0004] Therefore, developing a method for constructing the molecular identity card of Tianfu broiler chicken based on the fusion of multiple machine learning algorithms, screening out the minimum SNP set with high discriminative power, and constructing a high-precision and high-efficiency breed identification model is of great significance for realizing the protection of Tianfu broiler chicken germplasm resources, purebred identification and molecular breeding, and also has broad application prospects. Summary of the Invention
[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a combination of SNP molecular markers for identifying Tianfu broiler chickens and its application. The SNP molecular marker combination for identifying Tianfu broiler chickens consists of 20 SNP molecular markers with high discriminative power, constructing a Tianfu broiler chicken breed identification model. The SNP combination includes 33_3540036, 1_146933112, 1_66815463, 16_1759334, 10_10329321, 2_24734158, 3_18405453, 2_126275214, 4_24261336, 1_78050128, 2_65063243, 1_18 The invention identifies 20 specific loci: 1306873, 9_5903512, 6_34822674, 7_27874044, 6_10268252, 1_106149218, 2_138543556, 4_82085883, and 2_126956497. The SNP molecular marker combination of this invention exhibits significant species specificity for Tianfu broiler chickens, enabling rapid identification of genuine Tianfu broiler chickens with limited genotypic information. The breed identification model constructed based on this SNP combination achieves an average accuracy of 97.58% and an average ROC_AUC of 96.04%, providing efficient and accurate technical support for Tianfu broiler chicken breed identification and breeding program implementation.
[0006] To achieve the above technical effects, the following technical solution is adopted: A combination of SNP molecular markers for identifying Tianfu broiler chickens, with the following information on the 20 SNP sites in the combination:
[0007] The physical locations of the 20 SNP loci were determined based on the alignment of the chicken whole genome standard sequence, the version number of which is GCA_000002315.5 GRCg6a; the URL is: https: / / www.ncbi.nlm.nih.gov / datasets / genome / GCF_000002315.6 / .
[0008] An application of SNP molecular marker combinations for identifying Tianfu broiler chickens, the specific application method is as follows: Step (1): Obtain whole genome resequencing data of Tianfu broiler chicken and control breed; Step (2): Perform quality control on the SNP data, including a deletion rate ≤ 2% and a minimum allele frequency ≥ 1%; Genotype data were converted into numerical data of 0, 1, and 2 using a digital encoding method (REF / REF = 0, REF / ALT or ALT / REF = 1, ALT / ALT = 2). Missing values were filled with the median. Strict quality control was performed on SNP sites, with screening criteria of no more than 2% SNP missing rate and no less than 1% minimum allele frequency to ensure the accuracy and reliability of subsequent analysis.
[0009] Step (3): 80% of each variety is used as the training set and 20% as the test set; Step (4): Calculate the overall importance score for each SNP; Step (5): Select the top 20 SNP sites based on the comprehensive importance score; Step (6): A Tianfu broiler breed identification model was constructed using nine machine learning algorithms; Step (7): Score the model and obtain the optimal model based on the average accuracy and average ROC_AUC.
[0010] Furthermore, in step (1), whole-genome resequencing data of Tianfu broiler chickens and control breeds are obtained. The breeds include Changshun green-shelled chickens, Chishui black-boned chickens, Puding tall-legged chickens, Qiandongnan small fragrant chickens, Wumeng black-boned chickens, Weining chickens, Xingyi dwarf chickens, Yaoshan chickens, Xinyang black-feathered chicken breeding lines, Tianfu broiler chickens, and domestic chickens. The data of Changshun green-shelled chickens, Chishui black-boned chickens, Puding tall-legged chickens, Qiandongnan small fragrant chickens, Wumeng black-boned chickens, Weining chickens, Xingyi dwarf chickens, Yaoshan chickens, Xinyang black-feathered chicken breeding lines, and Tianfu broiler chickens are obtained by whole-genome resequencing through whole blood extraction. The pre-genome resequencing data of Red Junglefowl is obtained through open databases.
[0011] Specifically: Genomic DNA was extracted from the blood of 30 chickens of each breed, including Tianfu broiler chickens, Changshun green-shelled chickens, Chishui black-boned chickens, Puding tall-legged chickens, Qiandongnan small fragrant chickens, Wumeng black-boned chickens, Weining chickens, Xingyi dwarf chickens, Yaoshan chickens, and Xinyang black-feathered chicken breeding lines. The genomic DNA was resequencing the entire genome using an Illumina Hiseq X10 instrument. Whole-genome resequencing data from 30 Red Junglefowl chickens were obtained from the NCBI open database.
[0012] Furthermore, 80% of each variety was used as the training set and 20% as the test set.
[0013] Furthermore, in step (3), the comprehensive importance score is calculated by combining the following indicators: Mutual Information score, Allele Frequency Difference score, Random Forest Importance score, and LASSO Coefficients score. The average value of each indicator after standardization is taken as the comprehensive importance score; thus realizing the multi-dimensional assessment of the importance of SNP sites.
[0014] Furthermore, the nine machine learning algorithms in step (5) include: L2 penalized logistic regression, L1 penalized logistic regression, Elastic Net Logistic Regression; Random Forest, ExtraTrees, Gradient Boosting; Support Vector Machine Radial Basis Function (SVM_RBF), K-Nearest Neighbors (KNN), and NaiveBayes.
[0015] Furthermore, in step (6), the optimal model is constructed based on the support vector machine-radial basis function algorithm, and the model performance indicators are: average accuracy: 97.58%; average ROC-AUC: 96.04%; average F1 score: 86.11%; average balanced accuracy: 87.06%.
[0016] Furthermore, the SNP molecular marker combination is applied in the identification, genetic resource protection, or breeding screening of Tianfu broiler chickens.
[0017] The beneficial effects of this invention are as follows: 1. High accuracy: The identification model based on 20 SNP sites has an average accuracy of 97.58% and an average ROC_AUC of 96.04%.
[0018] 2. High efficiency: Accurate identification can be achieved with only 20 SNP sites, significantly reducing detection costs.
[0019] 3. Strong robustness: It adopts a multi-algorithm fusion strategy to avoid the limitations of a single method.
[0020] 4. Highly practical: It can be directly applied to variety identification, genetic resource protection, and breeding practices. Attached Figure Description
[0021] Figure 1 This is a schematic diagram of the workflow of an embodiment of the present invention; Figure 2This invention ranks the top 30 important SNPs based on machine learning predictions. Figure 3 This invention relates to information on 11 chicken breeds. Detailed Implementation
[0022] The present invention will be further described below with reference to the accompanying drawings. The scope of protection of the present invention is not limited to the following description: Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0023] Example 1: Method for constructing SNP combinations in Tianfu broiler chickens like Figure 1 As shown: 1. Genomic DNA was extracted from the blood of Tianfu broiler chickens, Changshun green-shelled chickens, Chishui black-boned chickens, Puding tall-legged chickens, Qiandongnan small fragrant chickens, Wumeng black-boned chickens, Weining chickens, Xingyi dwarf chickens, Yaoshan chickens, Xinyang black-feathered chicken breeding lines, Tianfu broiler chickens, and Red Junglefowl. Genomic DNA was extracted from the blood of 30 chickens of each breed from the Changshun green-shelled chicken, Chishui black-boned chicken, Puding tall-legged chicken, Qiandongnan small fragrant chicken, Wumeng black-boned chicken, Weining chicken, Xingyi dwarf chicken, Yaoshan chicken, and Xinyang black-feathered chicken breeding lines. The genomic DNA was resequencing the entire genome using an Illumina Hiseq X10 instrument. Whole genome resequencing data from 30 Red Junglefowl were obtained from the open database NCBI. Figure 3 As shown.
[0024] For the Tianfu broiler breed, samples of this breed are identified based on their names and binary labels are generated (target breed is labeled 1, other breeds are labeled 0). The genotype of each SNP locus is encoded: homozygous reference genotype (REF / REF) is encoded as 0, heterozygous genotype (REF / ALT or ALT / REF) is encoded as 1, and homozygous substitute genotype (ALT / ALT) is encoded as 2. Missing data are filled with the median value, generating a matrix of sample × SNP locus genotype.
[0025] Furthermore, 80% of each variety was used as the training set and 20% as the test set.
[0026] The SNP sites in the matrix are subjected to double quality filtering. First, SNP sites with a deletion rate higher than 2% are removed. Then, based on the calculated allele frequency of each SNP site, SNP sites with an allele frequency lower than 1% are removed, resulting in a set of candidate SNP sites.
[0027] Multi-indicator SNP importance assessment: For each candidate SNP locus of a variety, four importance indices are calculated: mutual information score: measures the statistical correlation between the SNP and the variety label; allele frequency difference: the absolute difference in allele frequencies between the target variety and other varieties; random forest importance: the feature importance score based on 100 trees; absolute value of Lasso coefficient: the absolute value of the standardized coefficient obtained through LassoCV cross-validation; after standardizing each index, the average value is taken to obtain the comprehensive importance score.
[0028] The standardized formula for the mutual information score is:
[0029] Among them, A n mi is the standardized value of the mutual information score of the nth candidate SNP site. n The mutual information score for the nth candidate SNP site; u mi The average mutual information score of all candidate SNP sites for each variety; σ mi The standard deviation of the mutual information scores of all candidate SNP sites for each variety, where ε is a constant 1×10⁻⁶. -9 .
[0030] The standardized formula for the allele frequency difference score is:
[0031] Among them, B n The standardized score for allele frequency difference at the nth candidate SNP locus; af n The allele frequency difference score for the nth candidate SNP locus; u af The average allele frequency difference score for all candidate SNP loci for each variety; σ af The standard deviation of allele frequency difference scores for all candidate SNP loci for each variety, where ε is a constant 1×10⁻⁶. -9 .
[0032] The standardized formula for the importance score of the random forest is as follows:
[0033] Among them, C n rf is the standardized value of the random forest importance score for the nth candidate SNP site; n The random forest importance score for the nth candidate SNP site; u rf The average random forest importance score of all candidate SNP sites for each variety; σ rf The standard deviation of the random forest importance scores for all candidate SNP sites for each species, where ε is a constant 1 × 10⁻⁶.-9 .
[0034] The standardized formula for the absolute value of the Lasso coefficient is:
[0035] Among them, D n The Lasso coefficient is the standardized score of the absolute value of the Lasso coefficient for the nth candidate SNP site; lasso n The absolute score of the Lasso coefficient for the nth candidate SNP site; u lasso The average of the absolute scores of the Lasso coefficients of all candidate SNP sites for each variety; σ lasso The standard deviation of the absolute values of the Lasso coefficients of all candidate SNP sites for each variety, where ε is a constant 1 × 10⁻⁶. -9 .
[0036] The formula for the comprehensive importance score of the nth SNP locus is: E n = (A n +B n +C n +D n ) / 4 Among them, E n A represents the overall importance score for the nth SNP locus; n B is the standardized value of the mutual information score of the nth candidate SNP site; n C is the standardized score of allele frequency difference at the nth candidate SNP locus; n D is the standardized value of the random forest importance score for the nth candidate SNP site; n The Lasso coefficient of the nth candidate SNP site is a standardized score.
[0037] Example 2: Construction and Optimization of Multi-Algorithm Variety Identification Model like Figure 2 As shown in the table below: Based on the overall importance score of SNPs, the top 20 SNP loci were selected. Locus information is as follows:
[0038] For the 20 SNP loci listed in the table above, nine machine learning algorithms were trained and compared, including: L2 penalized logistic regression, L1 penalized logistic regression, ElasticNet logistic regression; Random Forest, ExtraTrees, GradientBoosting; Support Vector Machine Radial Basis Function (SVM_RBF), K-Nearest Neighbors (KNN), and NaiveBayes.
[0039] Five-fold stratified cross-validation was used to evaluate the performance of each model, recording four core metrics: accuracy, ROC AUC, F1 score, and balanced accuracy. By comparing the performance of all SNP combination sizes and algorithm combinations, the optimal SNP panel size and optimal algorithm for each variety were determined.
[0040] The optimal algorithm for obtaining the information from the 20 SNP sites was ultimately determined to be: Support Vector Machine Radial Basis Function (SVM_RBF). The final model is retrained on the full training set using the best SNP combination and the best algorithm, saving the model parameters, SNP site list, median filler, and normalizer.
[0041] Example 3: Application of Tianfu Broiler Chicken SNP Combination Prediction Model The model based on Example 2 provides a unified prediction function. The input is a dictionary of SNP loci genotypes (keys are in the format "chromosome_location" and values are genotype strings or numbers). It automatically processes missing values and outputs the predicted probability and classification label of the target variety.
[0042] A dedicated SNP combination and optimization model was established for Tianfu broiler chickens, supporting large-scale parallel computing and output.
[0043] In addition to the prediction results, the system also provides SNP importance ranking and detailed model performance analysis, facilitating the understanding of the identification criteria and model reliability in practical operations. The breed identification model built based on this SNP combination (Support Vector Machine-Radial Basis Function) has an average accuracy of 97.58% and an average ROC_AUC of 96.04%, providing efficient and accurate technical support for the identification and implementation of breeding programs for Tianfu broiler chickens.
[0044] In summary, this invention discloses a combination of SNP molecular markers for identifying Tianfu broiler chickens and its application. The SNP molecular marker combination for identifying Tianfu broiler chickens consists of 20 SNP molecular markers with high discriminative power, constructing a Tianfu broiler chicken breed identification model. The SNP combination includes 33_3540036, 1_146933112, 1_66815463, 16_1759334, 10_10329321, 2_24734158, 3_18405453, 2_126275214, 4_24261336, 1_78050128, 2_65063243, 1_181306873, and 9_5903. The invention identifies 20 specific loci: 512, 6_34822674, 7_27874044, 6_10268252, 1_106149218, 2_138543556, 4_82085883, and 2_126956497. The SNP molecular marker combination of this invention exhibits significant species specificity for Tianfu broiler chickens, representing unique variation and important trait-determining loci. It enables rapid identification of genuine Tianfu broiler chickens using relatively little genotypic information. The breed identification model constructed based on this SNP combination has an average accuracy rate of 97.58% and an average ROC_AUC of 96.04%, providing efficient and accurate technical support for Tianfu broiler breed identification and breeding program implementation.
[0045] Therefore, those skilled in the art will recognize that although embodiments of the present invention have been shown and described in detail herein, many other variations or modifications conforming to the principles of the present invention can be directly determined or derived from the disclosure of the present invention without departing from the spirit and scope of the invention. Therefore, the scope of the present invention should be understood and recognized as covering all such other variations or modifications.
Claims
1. A combination of SNP molecular markers for identifying Tianfu broiler chicken, characterized in that, The specific information of the 20 SNP sites in the SNP molecular marker combination is as follows: ； The physical locations of the 20 SNP loci were determined based on the alignment of the chicken whole genome standard sequence; the version number of the chicken whole genome standard sequence is GCA_000002315.5 GRCg6a; the URL is: https: / / www.ncbi.nlm.nih.gov / datasets / genome / GCF_000002315.6 / .
2. The application of the method for identifying SNP molecular marker combinations in Tianfu broiler chicken as described in claim 1, characterized in that, The specific application method is as follows: Step (1): Obtain whole-genome resequencing data of Tianfu broiler chicken and control breed; Step (2): Perform quality control on the SNP data, including a deletion rate ≤ 2% and a minimum allele frequency ≥ 1%; Step (3): Calculate the overall importance score for each SNP; Step (4): Based on the comprehensive importance score, the top 20 SNP sites described in claim 1 are selected; Step (5): A Tianfu broiler breed identification model was constructed using nine machine learning algorithms; Step (6): Score the model and obtain the optimal model based on the average accuracy and average ROC_AUC.
3. The application of the method for identifying SNP molecular marker combinations in Tianfu broiler chicken as described in claim 2, characterized in that, In step (1), whole-genome resequencing data of Tianfu broiler chicken and control breeds are obtained. The breeds include Changshun green-shelled chicken, Chishui black-bone chicken, Puding tall-legged chicken, Qiandongnan small fragrant chicken, Wumeng black-bone chicken, Weining chicken, Xingyi dwarf chicken, Yaoshan chicken, Xinyang black-feathered chicken breeding line, Tianfu broiler chicken, and Hongjun chicken. The data of Changshun green-shelled chicken, Chishui black-bone chicken, Puding tall-legged chicken, Qiandongnan small fragrant chicken, Wumeng black-bone chicken, Weining chicken, Xingyi dwarf chicken, Yaoshan chicken, Xinyang black-feathered chicken breeding line, and Tianfu broiler chicken are obtained by whole-genome resequencing after extracting whole blood. The whole-genome resequencing data of Hongjun chicken is obtained through an open database.
4. The application of the method for identifying SNP molecular marker combinations in Tianfu broiler chicken as described in claim 2, characterized in that, In step (3), the overall importance score is calculated by combining the following indicators: Mutual Information score, Allele Frequency Difference score, Random Forest Importance score, and LASSO Coefficients score. The average value of each indicator after standardization is taken as the overall importance score.
5. The application of the method for identifying SNP molecular marker combinations in Tianfu broiler chicken as described in claim 2, characterized in that, The nine machine learning algorithms in step (5) include: L2 penalized logistic regression, L1 penalized logistic regression, Elastic Net Logistic Regression; Random Forest, ExtraTrees, GradientBoosting; Support Vector Machine Radial Basis Function (SVM_RBF), K-Nearest Neighbors (KNN), and NaiveBayes.
6. The application of the method for identifying SNP molecular marker combinations in Tianfu broiler chicken as described in claim 2, characterized in that, In step (6), the optimal model is constructed based on the support vector machine-radial basis function algorithm. The model performance indicators are: average accuracy: 97.58%; average ROC-AUC: 96.04%; average F1 score: 86.11%; average balanced accuracy: 87.06%.
7. The SNP molecular marker combination for identifying Tianfu broiler chicken as described in claim 1, characterized in that, Application of the SNP molecular marker combination in the identification, genetic resource protection or breeding screening of Tianfu broiler breed.