Beyond Genomic Prediction: Combining Different Types of omics Data Can Improve Prediction of Hybrid Performance in Maize

Abstract

The ability to predict the agronomic performance of single-crosses with high precision is essential for selecting superior candidates for hybrid breeding. With recent technological advances, thousands of new parent lines, and, consequently, millions of new hybrid combinations are possible in each breeding cycle, yet only a few hundred can be produced and phenotyped in multi-environment yield trials. Well established prediction approaches such as best linear unbiased prediction (BLUP) using pedigree data and whole-genome prediction using genomic data are limited in capturing epistasis and interactions occurring within and among downstream biological strata such as transcriptome and metabolome. Because mRNA and small RNA (sRNA) sequences are involved in transcriptional, translational and post-translational processes, we expect them to provide information influencing several biological strata. However, using sRNA data of parent lines to predict hybrid performance has not yet been addressed. Here, we gathered genomic, transcriptomic (mRNA and sRNA) and metabolomic data of parent lines to evaluate the ability of the data to predict the performance of untested hybrids for important agronomic traits in grain maize. We found a considerable interaction for predictive ability between predictor and trait, with mRNA data being a superior predictor for grain yield and genomic data for grain dry matter content, while sRNA performed relatively poorly for both traits. Combining mRNA and genomic data as predictors resulted in high predictive abilities across both traits and combining other predictors improved prediction over that of the individual predictors alone. We conclude that downstream "omics" can complement genomics for hybrid prediction, and, thereby, contribute to more efficient selection of hybrid candidates.

Phenomic selection: a low-cost and high-throughput method based on indirect predictions. Proof of concept on wheat and poplar

“…It is also difficult and too expensive to genotype all individuals to apply GS, despite important economies of scales. Alternative approaches based on endophenotypes such as transcriptomes or metabolomes have been proposed to predict phenotypes (Fu et al 2012;Riedelsheimer et al 2012;Feher et al 2014;Ward et al 2015;Fernandez et al 2016;Guo et al 2016;Xu et al 2016;Zenke-Philippi et al 2016;Westhues et al 2017;Seifert et al 2018;Schrag et al 2018), but their relatively low throughput and high costs are still likely to hamper their deployment at a large scale. To increase genetic progress in this context, we propose a new approach in which we use NIRS as high-throughput phenotypes to make predictions at low costs.…”

“…We could show for wheat that for fixed material, NIRS collected on seeds, so before sowing, was efficient to run PS, which offers very interesting perspectives for this species. The studies on endophenotypic variations in maize (Riedelsheimer et al 2012;Fu et al 2012;Guo et al 2016;Schrag et al 2018), rice ) and wheat (Ward et al 2015) also demonstrated that the characterization of germinated seeds or seedlings was efficient to estimate kinships resulting in accurate predictions. These results are promising, but this needs to be tested for other species and on other datasets.…”

“…Considering this fact, we should ask the question: are there more efficient alternatives than genotyping to estimate the kinship matrix? In the last years, it was proposed to use endophenotypes (MacKay et al 2009) such as transcripts (Fu et al 2012;Guo et al 2016;Zenke-Philippi et al 2016;Westhues et al 2017), small RNAs or metabolites (Riedelsheimer et al 2012;Feher et al 2014;Ward et al 2015;Fernandez et al 2016;Xu et al 2016;Guo et al 2016;Schrag et al 2018) as regressors or to estimate kinship. These endophenotypes correspond to different molecular layers between the genome and the phenotype, which permits the integration of interactions and regulatory networks when getting closer to the phenotypes.…”

Transcriptome-based prediction of complex traits in maize

“…The performance of models based on transcript levels can be better or worse compared to those based on genetic markers depending on the trait. For example, transcriptome data performed better for predicting grain yield in hybrid maize populations, but genetic marker data performed better for predicting grain dry matter content in the same population 15 . Similarly, in a maize diversity panel, GP models that combined transcript and marker data only outperformed models using markers alone for certain traits 16 .…”

Wheat and barley biology: Towards new frontiers

“…As this special issue cannot be all‐encompassing, many other promising and important future research areas, for example for mitigating the agricultural footprint in the ecosystem, such as improved nutrient use efficiency (Distelfeld et al ; Hawkesford ; Avin‐Wittenberg et al ), root architecture (Salvi ), or root‐microbe interactions specifically for nitrogen‐fixing cereals (Rogers and Oldroyd ; Mus et al ), are not itemized. Similarly, the manipulation of recombination at free‐will in our cereal crops (Lambing and Heckmann ), or a better molecular understanding of heterosis (Schrag et al ; Seifert et al ) and the introduction of a cost‐effective and reliable hybrid seed production system in cereals (Muehleisen et al ; Whitford et al ; Tucker et al ) have not even been touched upon. Thus, many global and societal challenges, but also truly exciting research opportunities, still lie ahead of us.…”