KineticGP: a computational framework for genomic prediction of leaf photosynthesis traits

Crop traits are the integrated outcome of genetic factors, environment effects, and their complex interactions, rendering accurate prediction from genetic markers alone a challenging problem. Here we present KineticGP, a computational framework that combines genomic prediction with genotype-specific kinetic models of C4 photosynthesis to make predictions of leaf photosynthesis traits across genotypes from a multiple parent advanced generation intercross maize population. Using genetic markers and gas exchange measurements from three field seasons, we show that KineticGP outperforms a baseline genomic prediction model for photosynthesis rate at saturating light by 86% for unseen genotypes across two seen seasons. In addition, KineticGP allowed surveying the genetic variability in enzyme kinetic parameters that can be used to raise targets for improvement of photosynthesis. The approach paves the way for interrogating and integrating the dynamic interactions between genotype and environment to improve the prediction accuracy of photosynthetic traits.