New Computational Approach Enhances Prediction of Trait Dynamics in Crop Breeding

ezbrenntow

New member
Advances in high-throughput phenotyping (HTP) and genotyping technologies have significantly advanced the breeding of crop varieties with desirable traits through genomic prediction. However, there is still a lack of understanding regarding how multiple traits express themselves at various stages throughout the plant’s growth cycle.


A research team, including scientists from the IPK Leibniz Institute and the Max Planck Institute of Molecular Plant Physiology, has developed a computational method to address this issue. Their findings were published in Nature Plants.


The phenome of a plant represents the totality of traits expressed at any given moment, influenced by genetic factors, environmental conditions, and their interactions. Understanding how a crop’s phenome evolves over time can help predict specific traits at key developmental stages. This challenge is complicated by the complex interdependencies between traits and the way phenomes vary across different genotypes over a plant’s life cycle.


Traditional genomic prediction (GP) methods train machine learning models using trait data from a population of genotypes at a single time point based on genetic markers. However, these methods do not yet address the challenge of predicting trait dynamics across the entire growth period of the plant.


To overcome this limitation, the team introduced dynamicGP, a computational method designed to predict trait dynamics during crop development using time-series phenotypic data from HTP platforms.


“We demonstrated that dynamicGP is an efficient tool for predicting genotype-specific dynamics for multiple traits. This is achieved by combining genomic prediction with dynamic mode decomposition (DMD),” explained David Hobby, a researcher at the Max Planck Institute of Molecular Plant Physiology and one of the lead authors of the study.





By applying genetic markers and HTP data from a maize multi-parent advanced generation inter-cross population and an Arabidopsis thaliana diversity panel, the team showed that dynamicGP outperformed traditional genomic prediction methods for multiple traits.


“We found that the developmental dynamics of traits with more stable heritability over time can be predicted with higher accuracy, offering insight into a key factor affecting trait predictability over a plant’s development,” said Dr. Marc Heuermann, a researcher at the IPK Leibniz Institute and co-author of the study.


The introduction of dynamicGP offers a new way to explore and integrate the dynamic interactions between genotype and phenotype throughout crop development, improving the prediction accuracy of important agronomic traits. Future improvements of dynamicGP could include incorporating environmental factors using extensions of DMD, leading to further refinements that will significantly impact crop breeding for specific regions and precision agriculture.
 
Top