High-Throughput Genotyping Analyses and Image-based Phenotyping in Sorghum bicolor

Abstract

Sorghum bicolor is a valuable plant grown commercially for grain, forage, sugar, and lignocellulosic biomass production. Increasing yields for these applications without increasing inputs is necessary to sustainably meet future food and fuel demand. The generation of superior plant cultivars that produce more without increased input is facilitated by methods that can rapidly and accurately acquire plant genotypic and phenotypic data, and this dissertation describes the development and application of genomic and phenomic methods to improve crop productivity. The sensitivity and specificity with which genetic variants are called from sorghum genomic sequence data was improved by developing a variant calling workflow; this workflow interrelates different sources of genomic sequence data to inform the modern machine learning techniques implemented within the Broad Institute's Genome Analysis Toolkit (GATK). Genetic variants called in this manner have been used to dissect the genetic basis of agriculturally important traits and improve the sorghum reference genome assembly. Additionally, to increase the rate at which the morphology of plants can be evaluated, an image-based phenotyping platform was developed to acquire measurements of sorghum shoot architecture traits using a depth camera. Depth images of plants are used to generate 3D reconstructions, and these reconstructions are used to measure phenotypes, to identify the genetic bases of shoot architecture, and as input to plant and crop modeling applications. This research facilitates the rapid and accurate acquisition of the data necessary to increase the rate of crop improvement