Novel bayesian networks for genomic prediction of developmental traits in biomass sorghum

Authors: DOS SANTOS, JONATHAN - Cornell University; FERNANDES, SAMUEL - University Of Illinois; LOZANO, ROBERTO - Cornell University; BROWN, PATRICK - Uc Davis Medical Center; Buckler, Edward - Ed; GARCIA, ANTONIO - Universidad De Sao Paulo; GORE, MICHAEL - Cornell University

Submitted to: bioRxiv
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/26/2019
Publication Date: 6/26/2019
Citation: Dos Santos, J.P., Fernandes, S.B., Lozano, R., Brown, P.K., Buckler IV, E.S., Garcia, A.A., Gore, M.A. 2019. Novel bayesian networks for genomic prediction of developmental traits in biomass sorghum. bioRxiv. https://doi.org/10.1101/677179.
DOI: https://doi.org/10.1101/677179

Interpretive Summary: An important component of renewable energy and carbon sequestration will be biomass crops. Sorghum is a desirable biomass crop because it is photosynthetic and water-use efficient and also requires less fertilizer. Breeding and selecting for varieties that will produce the highest levels of biomass is challenging given the size of the plants and plots necessary for evaluation. This research tested whether a combination of molecular markers, multiple early-season measurements, and statistical models could identify high-yielding varieties of sorghum better and sooner than standard approaches. Results showed that the combination of these approaches improved detection of high-yielding varieties by 36 to 52%. This combined approach will allow breeders to more quickly and accurately identify biomass sorghum varieties and could potentially be applied to other species being bred for biomass, such as trees.

Technical Abstract: The ability to connect genetic information between traits over time allow Bayesian networks to offer a powerful probabilistic framework to construct genomic prediction models. In this study, we phenotyped a diversity panel of 869 biomass sorghum (Sorghum bicolor (L.) Moench] lines, which had been genotyped with 100,435 SNP markers, for plant height (PH) with biweekly measurements from 30 to 120 days after planting (DAP) and for end-of-season dry biomass yield (DBY) in four environments. We evaluated five genomic prediction models: Bayesian network (BN), Pleiotropic Bayesian network (PBN), Dynamic Bayesian network (DBN), multi-trait GBLUP (MTr-GBLUP), and multi-time GBLUP (MTi-GBLUP) models. In 5-fold cross-validation, prediction accuracies ranged from 0.48 (PBN) to 0.51 (MTr-GBLUP) for DBY and from 0.47 (DBN, DAP120) to 0.74 (MTi-GBLUP, DAP60) for PH. Forward-chaining cross-validation further improved prediction accuracies of the DBN, MTi-GBLUP and MTr-GBLUP models for PH (training slice: 30-45 DAP) by 36.4-52.4% relative to the BN and PBN models. Coincidence indices (target: biomass, secondary: PH) and a coincidence index based on lines (PH time series) showed that the ranking of lines by PH changed minimally after 45 DAP. These results suggest a two-level indirect selection method for PH at harvest (first-level target trait) and DBY (second-level target trait) could be conducted earlier in the season based on ranking of lines by PH at 45 DAP (secondary trait). With the advance of high-throughput phenotyping technologies, our proposed two-level indirect selection framework could be valuable for enhancing genetic gain per unit of time when selecting on developmental traits.