Phenotype-associated genes (PAGs) have higher predicted probabilities compared with uncategorized gene models in maize, rice, and Arabidopsis. (A) Distribution of predicted probabilities of PAGs, premature stop enriched–common gene (PSE-C) models, and uncharacterized maize gene models. (B) Distribution of predicted probabilities of a subset of validated gene models (n = 96) retrieved from the literature in rice (Huang et al. 2022) using a model trained on maize data with 99 genomic sequence–derived and evolutionary features. (C) Distribution of predicted probabilities of a subset of validated gene models (n = 2018) retrieved from the literature in Arabidopsis (Lloyd and Meinke 2012) using a model trained on maize data with 99 genomic sequence–derived and evolutionary features. (D) Predicted probability distribution of newly validated PAGs (n = 27, curated from the MaizeGDB since November 27, 2021) compared with hold-out PAGs, hold-out PSE-Cs, premature stop enriched–rare genes (PSE-Rs), and uncategorized gene models.
