Genotype Performance Prediction Using RR-BLUP Regression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting crop genotype performance under specific environmental conditions are inefficient and costly, as they rely on traditional crop models that are not optimized for local conditions or specific genotypes, and require extensive trial data, which is time-consuming and prone to errors due to geographical variations.
Innovation Solution
An electronic system combining species-level environmental data with genotypic performance data using a crop model and a machine-learning linear regression model to generate performance predictions for specific genotypes under given conditions, achieving high determination coefficient R2 values (≥0.8) for yield and other parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional crop models are used to predict genotype performance, then species-level predictions can be obtained, but the models cannot be optimized for specific local conditions or specific genotypes
Solution Approach 1:
The patent applies local quality by transitioning from species-level crop models to genotype-specific models. Each genotype receives customized model parameters and environmental coefficients based on its unique characteristics, allowing the system to provide locally optimized predictions rather than generic species-level estimates. This is achieved through the genotype-specific model generation module that creates tailored prediction models for each genotype.
Solution Approach 2:
The patent segments the prediction system into distinct components: species-level crop models are divided into multiple genotype-specific models. This segmentation allows each genotype to have its own specialized prediction model with customized parameters, improving prediction accuracy for specific genotypes while maintaining the broader species-level framework.
2Measurement precision
If extensive trial data is collected to improve prediction accuracy, then genotype performance can be assessed more precisely, but the process becomes time-consuming and expensive
Solution Approach 1:
The patent applies preliminary action by pre-processing environmental data and pre-calculating environmental coefficients before actual genotype trials. The system prepares species-specific environmental models in advance, and when a new genotype is introduced, the genotype-specific model can be quickly generated using pre-computed environmental data, significantly reducing the time required for genotype assessment.
Solution Approach 2:
The patent uses copying by creating genotype-specific models that are derived from and copy the structure of species-level crop models. Instead of building entirely new models from scratch for each genotype, the system copies the proven species-level model framework and adapts it with genotype-specific parameters, reducing development time while maintaining accuracy.
3Reliability
If multiple genotypes are tested at multiple sites to account for environmental variations, then prediction reliability improves, but the complexity and cost of the testing network increases
Solution Approach 1:
The patent applies parameter changes by transforming environmental data into standardized environmental coefficients that can be applied across different locations. Instead of conducting physical trials at multiple sites, the system changes the parameters of the environmental models to reflect different environmental conditions, allowing virtual testing across multiple environments without the physical complexity of a multi-site trial network.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
An automated system is described for predicting performance of a crop genotype under particular environmental conditions. The system 1 is operable in a training phase, during which crop performance information 5' and corresponding environmental data 2' are used to train a predictor model 10 using a RR-BLUP regression algorithm 9. Predictor variables 8 for the regression algorithm are generated from the environment data 2' using a standard crop model 3. In an operation mode of the system, the trained model is used to generate predicted value(s) 7 of the performance parameters 5 of a particular genotype on the basis of environmental limiting factors 8 derived using the same crop model 3 from specified environmental conditions 2. A training method, a prediction method and a computer program product are also described.