A PARADIGM SHIFT IN BREEDING: FROM GENOMICS TO PHENOMICS

Chun-Peng Chen

doi:10.7273/000005456

Breeding for quantitative traits is studied by genomics and phenomics. In past decades, trait improvements have been mainly driven by genomics studies. Breeders broadly apply statistical approaches of Genome-Wide Association Studies (GWAS) and Genomic Selection (GS) to identify causal alleles and to evaluate breeding values, respectively. However, the implementations are usually limited to command-line environments and are not friendly to users. More research groups can participate in the studies if there's a user-friendly tool freeing researchers from learning multiple methods from scratch. In addition to genetic variation, how individuals respond to environments is another critical factor improving quantitative traits. The problems are investigated by phenomics studies and they usually require intensive efforts on collecting phenotypes over multiple environments. The advance of unmanned aerial vehicles (UAV) technology makes it possible to evaluate traits of interest on a large scale. However, challenges arise when interpreting raw reads into meaningful information. The works involve multiple computational methods and require considerable manual effort. Strong demand for an integrated segmentation tool emerges when phenomics studies get more attention. Besides the macroscope of phenomics driven by UAV, popularized hyperspectral scanners have stimulated phenomics research in the scope of single plants or seeds. One example in wheat is the evaluation of falling numbers (FN), which was believed to be correlated with the end-use quality. Low FNs are known to be contributed by two major causes: Preharvest sprouting (PHS) and late maturity α-amylase (LMA). Yet it's recently found that only PHS will lead to poor end-use quality, and this finding shows the need to differentiate PHS and LMA instantly before samples are priced.We cover three described problems in this thesis. The literature review is presented in the first chapter, and the second chapter demonstrates how the solution, intelligent prediction, and association tools (iPat), integrate GWAS and GS models into a graphical interface. In the macroscope of phenomics, we show that the tool, greenfield imagery decoder (GRID), can segment aerial images into subplots precisely with little human effort. Lastly, we address the micro-scope phenomics by showing the potential of spectroscopy in identifying low FN kernels from PHS or LMA samples. Overall, incorporating both genomics and phenomics can drive the breeding process more effectively.

A PARADIGM SHIFT IN BREEDING: FROM GENOMICS TO PHENOMICS

Files and links (1)

Abstract

Metrics

Details