Customized digital twins for plants from a physiological base model

Overview

This offering provides a framework for building customized digital twins of plant systems from a shared physiological base model. The approach combines genomic information with biologically informative process models to predict how novel genotypes will perform under untested scenarios, including those anticipated from climate change. By connecting genetic variation directly to physiological parameters, the framework enables better forecasting of genotype-by-environment interactions and supports decision-making for crop improvement.

The solution is especially relevant for organizations working with economically valuable crops such as rice, cotton, soybean, and wheat. It addresses a key limitation in current plant modeling: existing models often lack the right level of abstraction to integrate new -omics datasets across temporal and spatial scales. The resulting digital twins serve as flexible, agile tools for research and applied breeding programs.

Technical specifications

• Minimum base model developed in Python, with modular components for carbon processes, water transport, nitrogen cycling, and growth and development
• Parsimonious architecture that represents common physiological mechanisms while exposing outward-facing parameters that reflect genetic variation
• Application-specific derivations generated from the base model using physiological datasets already collected by the research group
• Genotype-specific process-based modeling approach demonstrated to capture genotype-by-environment interactions under both controlled and field conditions
• Integration pathway for publicly available genomic resources with process-based simulation
• Species- and germplasm-tailored outputs designed as digital twins of empirical experiments

Technology readiness level

The base model architecture was validated in prior published work demonstrating the value of genotype-specific process-based models for capturing genotype-by-environment interactions. The minimum base model is anticipated for completion in January 2024, after which the team plans to derive several tailored models over a one-year period. The primary developer of the model is a post-doctoral researcher, and the team is seeking partnership funding to support this role. Application-specific digital twins for rice, cotton, soybean, and wheat are the immediate next development targets.