Grzybowski, Marcin

Abstract Estimates of plant traits derived from hyperspectral reflectance data have the potential to efficiently substitute for traits, which are time or labor intensive to manually score. Typical workflows for estimating plant traits from hyperspectral reflectance data employ supervised classification models that can require substantial ground truth datasets for training. We explore the potential of an unsupervised approach, autoencoders, to extract meaningful traits from plant hyperspectral reflectance data using measurements of the reflectance of 2151 individual wavelengths of light from the leaves of maize ( Zea mays ) plants harvested from 1658 field plots in a replicated field trial. A subset of autoencoder‐derived variables exhibited significant repeatability, indicating that a substantial proportion of the total variance in these variables was explained by difference between maize genotypes, while other autoencoder variables appear to capture variation resulting from changes in leaf reflectance between different batches of data collection. Several of the repeatable latent variables were significantly correlated with other traits scored from the same maize field experiment, including one autoencoder‐derived latent variable (LV8) that predicted plant chlorophyll content modestly better than a supervised model trained on the same data. In at least one case, genome‐wide association study hits for variation in autoencoder‐derived variables were proximal to genes with known or plausible links to leaf phenotypes expected to alter hyperspectral reflectance. In aggregate, these results suggest that an unsupervised, autoencoder‐based approach can identify meaningful and genetically controlled variation in high‐dimensional, high‐throughput phenotyping data and link identified variables back to known plant traits of interest.

SUMMARY Transcriptome‐wide association studies (TWAS) can provide single gene resolution for candidate genes in plants, complementing genome‐wide association studies (GWAS) but efforts in plants have been met with, at best, mixed success. We generated expression data from 693 maize genotypes, measured in a common field experiment, sampled over a 2‐h period to minimize diurnal and environmental effects, using full‐length RNA‐seq to maximize the accurate estimation of transcript abundance. TWAS could identify roughly 10 times as many genes likely to play a role in flowering time regulation as GWAS conducted data from the same experiment. TWAS using mature leaf tissue identified known true‐positive flowering time genes known to act in the shoot apical meristem, and trait data from a new environment enabled the identification of additional flowering time genes without the need for new expression data. eQTL analysis of TWAS‐tagged genes identified at least one additional known maize flowering time gene through trans ‐eQTL interactions. Collectively these results suggest the gene expression resource described here can link genes to functions across different plant phenotypes expressed in a range of tissues and scored in different experiments.