Subsampling and DNA pooling can increase gains through genomic selection in switchgrass

Authors: TILHOU, NEAL - University Of Wisconsin; Casler, Michael

Submitted to: The Plant Genome
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/22/2021
Publication Date: 12/5/2021
Citation: Tilhou, N., Casler, M.D. 2021. Subsampling and DNA pooling can increase gains through genomic selection in switchgrass. The Plant Genome. 14(3). Article e20149. https://doi.org/10.1002/tpg2.20149.
DOI: https://doi.org/10.1002/tpg2.20149

Interpretive Summary: Switchgrass is one of the energy grass candidates for development of perennial biomass systems to supply feedstock for bioenergy production. Significant breeding efforts are required to make switchgrass production systems more economically sustainable. Genomic selection is one of the principal strategies being used to accelerate the breeding process toward making switchgrass more productive and sustainable. Because genomic selection requires considerable expense in DNA sequencing costs, this paper investigated several strategies for pooling DNA samples prior to sequencing, based either on pooling individuals with similar trait values or at random. Pooling strategies were shown to have reduced accuracy compared to using individual plants, but the reduction in sequencing cost was greater, suggesting that DNA pooling is a viable strategy for decreasing the cost of sequencing and improving the efficiency of switchgrass breeding. These results were sufficiently generalizable that they should apply to other breeding programs, including switchgrass and other species.

Technical Abstract: Genomic selection (GS) can accelerate breeding cycles in perennial crops such as the bioenergy grass switchgrass (Panicum virgatum L.). The sequencing costs of GS can be reduced by pooing DNA samples in the training population (TP), preferentially sequencing TP phenotypic outliers, or pooling evaluation population (EP) DNA samples. These methods were simulated for two traits (spring vigor and anthesis date) in three breeding populations of switchgrass. Sequencing only the outlier 50% of the TP phenotype distribution resulted in a penalty of <5% in GS predictive ability, measured using cross-validation. Predictive ability decreased when sequencing progressively fewer DNA pools due to increasing loss of phenotypic information. However, TPs constructed from only two phenotypically contrasting DNA samples retained a mean of >80% predictive ability relative to individual sequencing. Pooling DNA within the EP requires novel group testing methods. One- and two-dimensional EP DNA pooling allows greater than one EP individual to be evaluated per DNA sample sequenced. However, there was a trade-off between predictive ability and EP sequencing savings. To determine the benefit of reduced sequencing methods, breeding progress was calculated for seven GS scenarios while assuming constant sequencing resources. Reduced TP sequencing and most EP pooling methods were superior to individual sequence-based GS when sequencing budgets were restricted (2000 DNA samples per 5-year cycle). Only one reduced sequencing strategy was superior to individual sequencing when sequencing budgets large (8000 DNA samples per 5-year cycle). This study highlights multiple routes for reduced sequencing costs in GS, some of which could be valuable for future switchgrass yield improvements.