Genomic Prediction and Heritability Estimation for Daughter Pregnancy Rate in U.S. Holstein Cows Using SNP, Epistasis and Haplotype Effects
Abstract
1. Introduction
2. Results and Discussion
2.1. Initial Prediction Model Based on Genomic Heritability Estimates
2.2. Prediction Accuracies of Two SNP Models
2.3. Prediction Accuracies of Epistasis Models
2.4. Prediction Accuracy of Haplotype Models
2.5. Accuracy of Integrated Models with Epistasis and Haplotype Effects
2.6. Contributions of Intra- and Inter-Chromosome A × A Effects to Prediction Accuracy
2.7. Genomic Heritability Estimates from 10-Fold Validations
2.8. Sample Size and Prediction Accuracy
2.9. Prediction Accuracy in Training Populations
3. Materials and Methods
3.1. Holstein Populations and Genotyping Data
3.2. Mixed Model for GBLUP and GREML
3.3. Evaluation of Prediction Accuracy Using Cross-Validation
3.4. Initial Selection of Epistasis Effects for Prediction Models
4. Conclusions
Supplementary Materials
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
References
- VanRaden, P.; Sanders, A.; Tooker, M.; Miller, R.; Norman, H.; Kuhn, M.; Wiggans, G. Development of a national genetic evaluation for cow fertility. J. Dairy Sci. 2004, 87, 2285–2292. [Google Scholar] [CrossRef] [PubMed]
- Parker Gaddis, K.L.; Null, D.J.; Cole, J.B. Explorations in genome-wide association studies and network analyses with dairy cattle fertility traits. J. Dairy Sci. 2016, 99, 6420–6435. [Google Scholar] [CrossRef]
- Liang, Z.; Prakapenka, D.; VanRaden, P.M.; Jiang, J.; Ma, L.; Da, Y. A Million-cow genome-wide association study of three fertility traits in US Holstein cows. Int. J. Mol. Sci. 2023, 24, 10496. [Google Scholar] [CrossRef]
- CDCB Genomic Evaluations. Available online: https://uscdcb.com/genomic-evaluations/ (accessed on 7 June 2025).
- Derbyshire, M.C.; Khentry, Y.; Severn-Ellis, A.; Mwape, V.; Saad, N.S.M.; Newman, T.E.; Taiwo, A.; Regmi, R.; Buchwaldt, L.; Denton-Giles, M. Modeling first order additive× additive epistasis improves accuracy of genomic prediction for sclerotinia stem rot resistance in canola. Plant Genome 2021, 14, e20088. [Google Scholar] [CrossRef] [PubMed]
- Liang, Z.; Prakapenka, D.; Parker Gaddis, K.L.; VandeHaar, M.J.; Weigel, K.A.; Tempelman, R.J.; Koltes, J.E.; Santos, J.E.P.; White, H.M.; Peñagaricano, F. Impact of epistasis effects on the accuracy of predicting phenotypic values of residual feed intake in U. S Holstein cows. Front. Genet. 2022, 13, 2930. [Google Scholar] [CrossRef] [PubMed]
- Vojgani, E.; Pook, T.; Martini, J.W.; Hölker, A.C.; Mayer, M.; Schön, C.-C.; Simianer, H. Accounting for epistasis improves genomic prediction of phenotypes with univariate and bivariate models across environments. Theor. Appl. Genet. 2021, 134, 2913–2930. [Google Scholar] [CrossRef]
- Calleja-Rodriguez, A.; Chen, Z.; Suontama, M.; Pan, J.; Wu, H.X. Genomic predictions with nonadditive effects improved estimates of additive effects and predictions of total genetic values in Pinus sylvestris. Front. Plant Sci. 2021, 12, 666820. [Google Scholar] [CrossRef]
- Raffo, M.A.; Sarup, P.; Guo, X.; Liu, H.; Andersen, J.R.; Orabi, J.; Jahoor, A.; Jensen, J. Improvement of genomic prediction in advanced wheat breeding lines by including additive-by-additive epistasis. Theor. Appl. Genet. 2022, 135, 965–978. [Google Scholar] [CrossRef]
- Johnson, M.S.; Reddy, G.; Desai, M.M. Epistasis and evolution: Recent advances and an outlook for prediction. BMC Biol. 2023, 21, 120. [Google Scholar] [CrossRef]
- Akdemir, D.; Jannink, J.-L. Locally epistatic genomic relationship matrices for genomic association and prediction. Genetics 2015, 199, 857–871. [Google Scholar] [CrossRef]
- Won, S.; Park, J.-E.; Son, J.-H.; Lee, S.-H.; Park, B.H.; Park, M.; Park, W.-C.; Chai, H.-H.; Kim, H.; Lee, J. Genomic Prediction Accuracy Using Haplotypes Defined by Size and Hierarchical Clustering Based on Linkage Disequilibrium. Front. Genet. 2020, 11, 134. [Google Scholar] [CrossRef] [PubMed]
- Gao, N.; Teng, J.; Ye, S.; Yuan, X.; Huang, S.; Zhang, H.; Zhang, X.; Li, J.; Zhang, Z. Genomic Prediction of Complex Phenotypes Using Genic Similarity Based Relatedness Matrix. Front. Genet. 2018, 9, 364. [Google Scholar] [CrossRef]
- Liang, Z.; Tan, C.; Prakapenka, D.; Ma, L.; Da, Y. Haplotype analysis of genomic prediction using structural and functional genomic information for seven human phenotypes. Front. Genet. 2020, 11, 588907. [Google Scholar] [CrossRef]
- Difabachew, Y.F.; Frisch, M.; Langstroff, A.L.; Stahl, A.; Wittkop, B.; Snowdon, R.J.; Koch, M.; Kirchhoff, M.; Cselényi, L.; Wolf, M. Genomic prediction with haplotype blocks in wheat. Front. Plant Sci. 2023, 14, 1168547. [Google Scholar] [CrossRef]
- Alemu, A.; Batista, L.; Singh, P.K.; Ceplitis, A.; Chawade, A. Haplotype-tagged SNPs improve genomic prediction accuracy for Fusarium head blight resistance and yield-related traits in wheat. Theor. Appl. Genet. 2023, 136, 92. [Google Scholar] [CrossRef]
- Da, Y.; Liang, Z.; Prakapenka, D. Multifactorial methods integrating haplotype and epistasis effects for genomic estimation and prediction of quantitative traits. Front. Genet. 2022, 13, 922369. [Google Scholar] [CrossRef]
- Liang, Z.; Prakapenka, D.; Da, Y. Comparison of the accuracy of epistasis and haplotype models for genomic prediction of seven human phenotypes. Biomolecules 2023, 13, 1478. [Google Scholar] [CrossRef] [PubMed]
- VanRaden, P.M. Efficient methods to compute genomic predictions. J. Dairy Sci. 2008, 91, 4414–4423. [Google Scholar] [CrossRef] [PubMed]
- Fernando, R.L.; Cheng, H.; Garrick, D.J. An efficient exact method to obtain GBLUP and single-step GBLUP when the genomic relationship matrix is singular. Genet. Sel. Evol. 2016, 48, 80. [Google Scholar] [CrossRef]
- Misztal, I. Inexpensive Computation of the Inverse of the Genomic Relationship Matrix in Populations with Small Effective Population Size. Genetics 2015, 115, 182089. [Google Scholar] [CrossRef]
- VanRaden, P.M.; Sun, C.; O’Connell, J.R. Fast imputation using medium or low-coverage sequence data. BMC Genet. 2015, 16, 82. [Google Scholar] [CrossRef] [PubMed]
- Henderson, C. Best linear unbiased prediction of nonadditive genetic merits in noninbred populations. J. Anim. Sci. 1985, 60, 111–117. [Google Scholar] [CrossRef]
- Su, G.; Christensen, O.F.; Ostersen, T.; Henryon, M.; Lund, M.S. Estimating additive and non-additive genetic variances and predicting genetic merits using genome-wide dense single nucleotide polymorphism markers. PloS One 2012, 7, e45293. [Google Scholar] [CrossRef]
- Muñoz, P.R.; Resende, M.F.; Gezan, S.A.; Resende, M.D.V.; de los Campos, G.; Kirst, M.; Huber, D.; Peter, G.F. Unraveling additive from nonadditive effects using genomic relationship matrices. Genetics 2014, 198, 1759–1768. [Google Scholar] [CrossRef] [PubMed]
- Vitezica, Z.G.; Legarra, A.; Toro, M.A.; Varona, L. Orthogonal estimates of variances for additive, dominance, and epistatic effects in populations. Genetics 2017, 206, 1297–1307. [Google Scholar] [CrossRef]
- Jiang, Y.; Reif, J.C. Efficient algorithms for calculating epistatic genomic relationship matrices. Genetics 2020, 216, 651–669. [Google Scholar] [CrossRef]
- Legarra, A.; Robert-Granie, C.; Manfredi, E.; Elsen, J.M. Performance of genomic selection in mice. Genetics 2008, 180, 611–618. [Google Scholar] [CrossRef]
Sample 1 (n = 50,606) | Subscript in Equations (1) and (2) | A | A + D | Full Model | A + AA | A + AAinter + AAintra |
---|---|---|---|---|---|---|
A | 1 | 0.053 | 0.053 | 0.047 | 0.047 | 0.045 |
D | 2 | - | 0.006 | 0.004 | - | - |
A × A | 3 | - | - | 0.066 | 0.067 | - |
A × Ainter | 3 | - | - | - | - | 0.048 |
A × Aintra | 4 | - | - | - | - | 0.019 |
A × D | 4 | - | - | 0.000 | - | - |
D × D | 5 | - | - | 0.000 | - | - |
A × A × A | 6 | - | - | 0.000 | - | - |
A× D × D | 7 | - | - | 0.000 | - | - |
A × D × D | 8 | - | - | 0.000 | - | - |
D × D× D | 9 | - | - | 0.000 | - | - |
Total Heritability | 0.053 | 0.059 | 0.116 | 0.114 | 0.112 | |
Initial Model | - | - | - | A + AA | A + AAinter + AAintra | |
Sample 2 (n = 133,934) | Subscript in Equations (1) and (2) | A | A + D | Full Model | A + AA | |
A | 1 | 0.044 | 0.044 | 0.037 | 0.036 | |
D | 2 | - | 0.005 | 0.003 | - | |
A × A | 3 | - | - | 0.060 | 0.061 | |
A × D | 4 | - | - | 0.000 | - | |
D × D | 5 | - | - | 0.000 | - | |
A × A × A | 6 | - | - | 0.000 | - | |
A× D × D | 7 | - | - | 0.000 | - | |
A × D × D | 8 | - | - | 0.000 | - | |
D × D× D | 9 | - | - | 0.000 | - | |
Total heritability | 0.044 ± 0.000 | 0.049 ± 0.002 | 0.099 | 0.097 | ||
Initial model | - | - | - | A + AA |
Prediction Accuracy | Accuracy Increase over A-Model (%) | |||
---|---|---|---|---|
Model | T | V | T | V |
SNP model | ||||
A (n = 25,827) | 0.383 ± 0.004 | 0.162 ± 0.013 | 0.00 | 0.00 |
A + D (n = 25,827) | 0.415 ± 0.028 | 0.161 ± 0.012 | 8.36 | −0.62 |
A (n = 50,606) | 0.327 ± 0.002 | 0.184 ± 0.014 | 0.00 | 0.00 |
A + D (n = 50,606) | 0.356 ± 0.004 | 0.185 ± 0.014 | 8.86 | 0.54 |
A (n = 70,000) | 0.305 ± 0.001 | 0.201 ± 0.008 | 0.00 | 0.00 |
A + D (n = 70,000) | 0.317 ± 0.002 | 0.201 ± 0.008 | 3.93 | 0.00 |
A (n = 90,000) | 0.304 ± 0.002 | 0.208 ± 0.012 | 0.00 | 0.00 |
A + D (n = 90,000) | 0.322 ± 0.003 | 0.208 ± 0.012 | 5.92 | 0.00 |
A (n = 133,934) | 0.289 ± 0.001 | 0.205 ± 0.007 | 0.00 | 0.00 |
A + D (n = 133,934) | 0.310 ± 0.002 | 0.207 ± 0.007 | 7.27 | 0.97 |
Global epistasis model | ||||
A + AA (n = 25,827) | 0.649 ± 0.075 | 0.165 ± 0.014 | 69.45 | 1.85 |
A + AA (n = 50,606) | 0.594 ± 0.010 | 0.193 ± 0.014 | 81.65 | 4.89 |
A + D + AA (n = 50,606) | 0.601 ± 0.011 | 0.193 ± 0.014 | 83.79 | 4.89 |
A + AA (n = 70,000) | 0.506 ± 0.007 | 0.208 ± 0.008 | 65.90 | 3.48 |
A + AA (n = 90,000) | 0.520 ± 0.001 | 0.216 ± 0.011 | 71.05 | 3.85 |
A + AA (n = 133,934) | 0.520 ± 0.004 | 0.215 ± 0.006 | 79.93 | 4.88 |
A + D + AA (n = 133,934) | 0.524 ± 0.004 | 0.215 ± 0.006 | 81.31 | 4.88 |
Haplotype model | ||||
A + H (n = 25,827) | 0.462 ± 0.016 | 0.164 ± 0.014 | 20.63 | 1.23 |
A + H (n = 50,606) | 0.406 ± 0.004 | 0.188 ± 0.014 | 14.04 | 2.17 |
A + H (n = 70,000) | 0.362 ± 0.003 | 0.204 ± 0.008 | 18.69 | 1.49 |
A + H (n = 90,000) | 0.357 ± 0.002 | 0.211 ± 0.012 | 17.43 | 1.44 |
A + H (n = 133,934) | 0.341 ± 0.002 | 0.210 ± 0.006 | 17.99 | 2.44 |
Integrated model | ||||
A + AA + H (n = 25,827) | 0.633 ± 0.076 | 0.166 ± 0.014 | 65.27 | 2.47 |
A + AA + H (n = 50,606) | 0.572 ± 0.013 | 0.192 ± 0.014 | 74.92 | 4.35 |
A + AA + H (n = 70,000) | 0.487 ± 0.007 | 0.208 ± 0.008 | 59.67 | 3.48 |
A + AA + H (n = 90,000) | 0.505 ± 0.011 | 0.216 ± 0.011 | 66.12 | 3.85 |
A + AA + H (n = 133,934) | 0.509 ± 0.004 | 0.215 ± 0.006 | 76.12 | 4.88 |
Prediction Accuracy | Accuracy Increase over A-Model (%) | |||
---|---|---|---|---|
Model | T | V | T | V |
A + AAinter + AAintra | 0.586 ± 0.010 | 0.193 ± 0.013 | 79.20 | 4.89 |
A + AAinter | 0.589 ± 0.010 | 0.192 ± 0.014 | 80.12 | 4.35 |
A + AAintra | 0.464 ± 0.009 | 0.189 ± 0.013 | 41.89 | 2.72 |
A + AAinter +AAinta + H | 0.568 ± 0.012 | 0.192 ± 0.013 | 59.55 | 4.35 |
A + AAinter + H | 0.569 ± 0.012 | 0.192 ± 0.014 | 74.01 | 4.35 |
A + AAintra + H | 0.477 ± 0.009 | 0.190 ± 0.013 | 45.87 | 3.26 |
Model | Sample 1 with 50,606 Cows | ||||||
---|---|---|---|---|---|---|---|
A | D | HH | AA | AAinter | AAintra | Total | |
A | 0.054 ± 0.001 | - | - | - | - | - | 0.054 ± 0.001 |
A + D | 0.054 ± 0.001 | 0.006 ± 0.001 | - | - | - | - | 0.060 ± 0.001 |
A + AA | 0.047 ± 0.001 | - | - | 0.067 ± 0.003 | - | - | 0.114 ± 0.003 |
A + H | 0.040 ± 0.002 | - | 0.030 ± 0.002 | - | - | - | 0.070 ± 0.001 |
A + AAinter | 0.047 ± 0.001 | - | - | - | 0.065 ± 0.003 | - | 0.112 ± 0.003 |
A + AAintra | 0.047 ± 0.002 | - | - | - | - | 0.036 ± 0.003 | 0.083 ± 0.002 |
A + AAinter + H | 0.040 ± 0.002 | - | 0.021 ± 0.003 | - | 0.047 ± 0.005 | - | 0.108 ± 0.003 |
A + AAintra + H | 0.039 ± 0.002 | - | 0.023 ± 0.002 | - | - | 0.024 ± 0.002 | 0.086 ± 0.002 |
A + AAinter + AAintra | 0.046 ± 0.001 | - | - | - | 0.047 ± 0.004 | 0.020 ± 0.003 | 0.112 ± 0.003 |
A + AAinter +AAinta + H | 0.039 ± 0.002 | - | 0.019 ± 0.003 | - | 0.035 ± 0.004 | 0.014 ± 0.003 | 0.108 ± 0.003 |
A + H + AA | 0.040 ± 0.002 | - | 0.021 ± 0.003 | 0.049 ± 0.005 | - | - | 0.109 ± 0.003 |
Sample 2 with 133,934 cows | |||||||
A | 0.044 ± 0.001 | - | - | 0.044 ± 0.001 | 0.044 ± 0.001 | ||
A + D | 0.044 ± 0.001 | 0.005 ± 0.001 | - | - | 0.049 ± 0.001 | ||
A + AA | 0.037 ± 0.001 | - | - | 0.062 ± 0.001 | 0.099 ± 0.001 | ||
A + H | 0.034 ± 0.001 | - | 0.022 ± 0.000 | - | - | - | 0.056 ± 0.001 |
A + H + AA | 0.033 ± 0.001 | - | 0.011 ± 0.000 | 0.052 ± 0.001 | 0.096 ± 0.001 |
Sample 3 (m = 75,209) | Sample 1 (m = 75,209) | Sample 2 (m = 74,855) | ||||
---|---|---|---|---|---|---|
GRM | No. Cows (n) | Rank | No. Cows (n) | Rank | No. Cows (n) | Rank |
A () | 25,827 | 25,817 | 50,606 | 50,594 | 133,934 | 74,853 |
D () | 25,827 | 25,818 | 50,606 | 50,595 | 133,934 | 74,853 |
AA () | 25,827 | 25,818 | 50,606 | 50,595 | 133,934 | 133,896 |
AD () | 25,827 | 25,818 | 50,606 | 50,595 | 133,934 | 133,896 |
DD () | 25,827 | 25,818 | 50,606 | 50,595 | 133,934 | 133,896 |
AAA () | 25,827 | 25,818 | 50,606 | 50,595 | 133,934 | 133,896 |
AAD () | 25,827 | 25,818 | 50,606 | 50,595 | 133,934 | 133,896 |
ADD () | 25,827 | 25,818 | 50,606 | 50,595 | 133,934 | 133,896 |
DDD () | 25,827 | 25,818 | 50,606 | 50,595 | 133,934 | 133,896 |
AH () | 25,827 | 25,817 | 50,606 | 50,594 | 133,934 | 133,897 |
Sample 3 (Subset of Sample 1) | Sample 1 | Sample 2 | Sample 4 (Subset of Sample 2) | Sample 5 (Subset of Sample 2) | |
---|---|---|---|---|---|
Number of cows | 25,827 | 50,606 | 133,934 | 70,000 | 90,000 |
Number of SNPs | 75,209 | 75,209 | 74,855 | 74,855 | 74,855 |
Purpose | Identifying the best haplotype model | Identifying the best epistasis model | Large sample analysis of prediction accuracy and heritability estimates | Additional evidence for the impact of sample size on prediction accuracy | Additional evidence for the impact of sample size on prediction accuracy |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Yang, R.; Prakapenka, D.; Liang, Z.; Da, Y. Genomic Prediction and Heritability Estimation for Daughter Pregnancy Rate in U.S. Holstein Cows Using SNP, Epistasis and Haplotype Effects. Int. J. Mol. Sci. 2025, 26, 5687. https://doi.org/10.3390/ijms26125687
Yang R, Prakapenka D, Liang Z, Da Y. Genomic Prediction and Heritability Estimation for Daughter Pregnancy Rate in U.S. Holstein Cows Using SNP, Epistasis and Haplotype Effects. International Journal of Molecular Sciences. 2025; 26(12):5687. https://doi.org/10.3390/ijms26125687
Chicago/Turabian StyleYang, Ruifei, Dzianis Prakapenka, Zuoxiang Liang, and Yang Da. 2025. "Genomic Prediction and Heritability Estimation for Daughter Pregnancy Rate in U.S. Holstein Cows Using SNP, Epistasis and Haplotype Effects" International Journal of Molecular Sciences 26, no. 12: 5687. https://doi.org/10.3390/ijms26125687
APA StyleYang, R., Prakapenka, D., Liang, Z., & Da, Y. (2025). Genomic Prediction and Heritability Estimation for Daughter Pregnancy Rate in U.S. Holstein Cows Using SNP, Epistasis and Haplotype Effects. International Journal of Molecular Sciences, 26(12), 5687. https://doi.org/10.3390/ijms26125687