Application of machine learning to explore the genomic prediction accuracy of fall dormancy in autotetraploid alfalfa

Abstract

Fall dormancy (FD) is an essential trait to overcome winter damage and for alfalfa (Medicago sativa) cultivar selection. The plant regrowth height after autumn clipping is an indirect way to evaluate FD. Transcriptomics, proteomics, and quantitative trait locus mapping have revealed crucial genes correlated with FD; however, these genes cannot predict alfalfa FD very well. Here, we conducted genomic prediction of FD using whole-genome SNP markers based on machine learning-related methods, including support vector machine (SVM) regression, and regularization-related methods, such as Lasso and ridge regression. The results showed that using SVM regression with linear kernel and the top 3000 genome-wide association study (GWAS)-associated markers achieved the highest prediction accuracy for FD of 64.1%. For plant regrowth height, the prediction accuracy was 59.0% using the 3000 GWAS-associated markers and the SVM linear model. This was better than the results using whole-genome markers (25.0%). Therefore, the method we explored for alfalfa FD prediction outperformed the other models, such as Lasso and ElasticNet. The study suggests the feasibility of using machine learning to predict FD with GWAS-associated markers, and the GWAS-associated markers combined with machine learning would benefit FD-related traits as well. Application of the methodology may provide potential targets for FD selection, which would accelerate genetic research and molecular breeding of alfalfa with optimized FD.

Acknowledgements

We thank the Medium Term Library of National Grass Seed Resources of China and the U.S. National Plant Germplasm System (USDA GRIN) for providing the alfalfa accessions. This work was supported by the National Natural Science Foundation of China (No. 31971758), the breeding forage and grain legumes to increase China’s and EU’s protein self-sufficiency, collaborative research key project between China and EU (2017YFE0111000/EUCLEG 727312), Key Projects in Science and Technology of Inner Mongolia (2021ZD0031), and the China Scholarship Council (201903250068). The funding bodies played no role in the study’s design, the collection, analysis, and interpretation of data, or the writing of the manuscript.

Author contributions

Q.C.Y. conceived and designed the experiments. J.M.K., R.C.L., M.N.L., and Y.S. planted the alfalfa accessions. F.Z., J.M.K., R.C.L., F.H., X.Q.J., C.F.Y., and X.J.Y. collected the phenotypes. F.Z., J.K., Y.W.W., Z.W., and Z.W.Z. analyzed the data. F.Z., Z.W.Z., Z.W., and Q.C.Y. wrote the paper. All authors read and approved the final manuscript.

Data availability

All RAD raw sequence data were upload to the National Genomics Data Center (NGDC, https://bigd.big.ac.cn/) under BioProject PRJCA004024 (https://ngdc.cncb.ac.cn/search/?dbId=biosample&q=PRJCA004024&page=1) and NCBI Sequence Read Archive with Bioproject ID: PRJNA739212 (https://www.ncbi.nlm.nih.gov/sra/?term=PRJNA739212). The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Conflict of interest

The authors declare that they have no conflicts of interest.

References

[1] Teuber L, Taggard K, Gibbs L et al. Fall dormancy. Standard tests to characterize alfalfa cultivars. St. Paul, MN, USA: North American Alfalfa Improvement Conference, 1998, A-1-A-2.

[2] Barnes DK, Smith DM, Stucker RE et al. Fall dormancy in alfalfa: a valuable predictive tool. Washington, D.C., USA: The 26th Alfalfa Improvement Conference, 1979, 34.

[3] Smith D . Association of fall growth habit and winter survival in alfalfa. Can J Plant Sci. 1961; 41: 244-51.

[4] Stout DG . Fall growth and stand persistence of alfalfa in interior British Columbia. Can J Plant Sci. 1985; 65: 935-41.

[5] Stout DG, Hall JW . Fall growth and winter survival of alfalfa in interior British Columbia. Can J Plant Sci. 1989; 69: 491-9.

[6] Avci M, Hatipoglu R, Çinar S et al. Assessment of yield and quality characteristics of alfalfa (Medicago sativa L.) cultivars with different fall dormancy rating. Legume Res. 2018; 41: 369-73.

[7] Li X, Alarcón-Zúñiga B, Kang J et al. Mapping fall dormancy and winter injury in tetraploid alfalfa. Crop Sci. 2015; 55: 1995-2011.

[8] Munjal G, Hao J, Teuber LR et al. Selection mapping identifies loci underpinning autumn dormancy in alfalfa (Medicago sativa). G3 (Bethesda). 2018; 8: 461- 8.

[9] Adhikari L, Lindstrom OM, Markham J et al. Dissecting key adaptation traits in the polyploid perennial Medicago sativa using GBS-SNP mapping. Front Plant Sci. 2018; 9: 934.

[10] Cunningham S, Gana J, Volenec J et al. Winter hardiness, root physiology, and gene expression in successive fall dormancy selections from ‘Mesilla’ and ‘CUF 101’ alfalfa. Crop Sci. 2001; 41: 1091-8.

[11] Du H, Shi Y, Li D et al. Screening and identification of key genes regulating fall dormancy in alfalfa leaves. PLoS One. 2017; 12: e0188964.

[12] Liu Z-Y, Baoyin T, Li X-L et al. How fall dormancy benefits alfalfa winter-survival? Physiologic and transcriptomic analyses of dormancy process. BMC Plant Biol. 2019; 19: 1-13.

[13] Du H, Shi Y, Li D et al. Proteomics reveals key proteins participating in growth difference between fall dormant and non-dormant alfalfa in terminal buds. J Proteome. 2018; 173: 126-38.

[14] Bertrand A, Bipfubusa M, Claessens A et al. Effect of photoperiod prior to cold acclimation on freezing tolerance and carbohydrate metabolism in alfalfa (Medicago sativa L.). Plant Sci. 2017; 264: 122-8.

[15] Biazzi E, Nazzicari N, Pecetti L et al. Genome-wide association mapping and genomic selection for alfalfa (Medicago sativa) forage quality traits. PLoS One. 2017; 12: e0169234.

[16] Wang Z, Qiang H, Zhao H et al. Association mapping for fiber-related traits and digestibility in alfalfa (Medicago sativa). Front Plant Sci. 2016; 7: 331.

[17] Yu L-X, Liu X, Boge W et al. Genome-wide association study identifies loci for salt tolerance during germination in autotetraploid alfalfa (Medicago sativa L.) using genotyping-by-sequencing. Front Plant Sci. 2016; 7: 956.

[18] Zhang T, Yu L-X, Zheng P et al. Identification of loci associated with drought resistance traits in heterozygous autotetraploid alfalfa (Medicago sativa L.) using genome-wide association studies with genotyping by sequencing. PLoS One. 2015; 10: e0138931.

[19] Liu X-P, Yu L-X . Genome-wide association mapping of loci associated with plant growth and forage production under salt stress in alfalfa (Medicago sativa L.). Front Plant Sci. 2017; 8: 853.

[20] Lin S, Medina CA, Boge B et al. Identification of genetic loci associated with forage quality in response to water deficit in autotetraploid alfalfa (Medicago sativa L.). BMC Plant Biol. 2020; 20: 1-18.

[21] Jia C, Wu X, Chen M et al. Identification of genetic loci associated with crude protein and mineral concentrations in alfalfa (Medicago sativa) using association mapping. BMC Plant Biol. 2017; 17: 97.

[22] Wang Z, Wang X, Zhang H et al. A genome-wide association study approach to the identification of candidate genes underlying agronomic traits in alfalfa (Medicago sativa L.). Plant Biotechnol J. 2020; 18: 611-3.

[23] Yu LX, Zheng P, Zhang T et al. Genotyping-by-sequencing-based genome-wide association studies on verticillium wilt resistance in autotetraploid alfalfa (Medicago sativa L.). Mol Plant Pathol. 2017; 18: 187-94.

[24] Sakiroglu M, Brummer EC . Identification of loci controlling forage yield and nutritive value in diploid alfalfa using GBS-GWAS. Theor Appl Genet. 2017; 130: 261-8.

[25] Crossa J, Pérez-Rodríguez P, Cuevas J et al. Genomic selection in plant breeding: methods, models, and perspectives. Trends Plant Sci. 2017; 22: 961-75.

[26] Annicchiarico P, Nazzicari N, Li X et al. Accuracy of genomic selection for alfalfa biomass yield in different reference populations. BMC Genomics. 2015; 16: 1-13.

[27] Li X, Wei Y, Acharya A et al. Genomic prediction of biomass yield in two selection cycles of a tetraploid alfalfa breeding population. Plant Genome. 2015; 8: plantgenome2014.12.0090.

[28] Tong H, Nikoloski Z . Machine learning approaches for crop improvement: leveraging phenotypic and genotypic big data. J Plant Physiol. 2021; 257: 153354.

[29] Whitmire CD, Vance JM, Rasheed HK et al. Using machine learning and feature selection for alfalfa yield prediction. AI. 2021; 2: 71-88.

[30] Browning BL, Zhou Y, Browning SR . A one-penny imputed genome from next-generation reference panels. Am J Hum Genet. 2018; 103: 338-48.

[31] Zingaretti LM, Gezan SA, Ferrão LFV et al. Exploring deep learning for complex trait genomic prediction in polyploid outcrossing species. Front Plant Sci. 2020; 11: 25.

[32] Wang J, Zhou Z, Zhang Z et al. Expanding the BLUP alphabet for genomic prediction adaptable to the genetic architectures of complex traits. Heredity. 2018; 121: 648-62.

[33] Long R, Zhang F, Zhang Z et al. Genome assembly of alfalfa cultivar Zhongmu-4 and identification of SNPs associated with agronomic traits. Genomics Proteomics Bioinformatics. 2022; 20: 14-28.

[34] Zhang F, Kang J, Long R et al. High-density linkage map construction and mapping QTL for yield and yield components in autotetraploid alfalfa using RAD-seq. BMC Plant Biol. 2019; 19: 1-12.

[35] Muleta KT, Bulli P, Zhang Z et al. Unlocking diversity in germplasm collections via genomic selection: a case study based on quantitative adult plant resistance to stripe rust in spring wheat. Plant Genome. 2017; 10: plantgenome2016.12.0124.

[36] Medina CA, Kaur H, Ray I et al. Strategies to increase prediction accuracy in genomic selection of complex traits in alfalfa (Medicago sativa L.). Cells. 2021; 10: 3372.

[37] Li X, Brummer EC . Applied genetics and genomics in alfalfa breeding. Agronomy. 2012; 2: 40-61.

[38] Xianglin L, Liqiang W . Alfalfa fall dormancy and its relationship to winter hardiness and yield. Acta Pratacul Sin. 2004; 13: 57-61.

[39] Bolger AM, Lohse M, Usadel B . Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014; 30: 2114-20.

[40] Li H . Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv 2013;13033997.

[41] Li H, Handsaker B, Wysoker A et al. The sequence alignment/map format and SAMtools. Bioinformatics. 2009; 25: 2078-9.

[42] Van der Auwera GA, Carneiro MO, Hartl C et al. From FastQ data to high-confidence variant calls: the genome analysis toolkit best practices pipeline. Curr Protoc Bioinformatics. 2013; 43: 11.10.1-33.

[43] Koboldt DC, Zhang Q, Larson DE et al. VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res. 2012; 22: 568-76.

[44] Danecek P, Auton A, Abecasis G et al. The variant call format and VCFtools. Bioinformatics. 2011; 27: 2156-8.

[45] Huang M, Liu X, Zhou Y et al. BLINK: a package for the next level of genome-wide association studies with both individuals and markers in the millions. GigaScience. 2019; 8: giy154.

[46] Turner SD . qqman: an R package for visualizing GWAS results using QQ and Manhattan plots. bioRxiv. 2014; 005165.

[47] Brault C, Segura V, This P et al. Across-population genomic prediction in grapevine opens up promising prospects for breeding. Hortic Res. 2022; 9: 9.

[48] Cappa EP, Chen C, Klutsch JG et al. Multiple-trait analyses improved the accuracy of genomic prediction and the power of genome-wide association of productivity and climate change-adaptive traits in lodgepole pine. BMC Genomics. 2022; 23: 536.