Soybean Yield, Protein, Oil (YPO) Predictions via Genomic and Phenomic Prediction

In this research, we first conducted a genome wide association study (GWAS) on soybean seed yield, protein, and oil. Results of this largely overlapped with previous reports in soybean. Next, we trained a genomic best linear unbiased prediction (GBLUP) model for genomic prediction of the three seed traits, using molecular marker data. We additionally trained phenomic prediction models using three different machine learning models (partial least squares regression, random forest, and extreme gradient boosting) to predict each of the three traits. This was trained on vegetation indices that were taken at two different timepoints. We then compared the predictive ability between the GBLUP and different phenomic prediction models, and their ability to rank soybean accessions for selection of specified traits.