Related Experiment Video
Updated: Jan 18, 2026
15:30
A Telemetric, Gravimetric Platform for Real-Time Physiological Phenotyping of Plant–Environment Interactions
Published on: August 5, 2020
12.4K
ResDeepGS: A deep learning-based method for crop phenotype prediction
Chaokun Yan1, Jiabao Li1, Qi Feng1
1School of Computer and Information Engineering, Henan University, Kaifeng, Henan, China; Academy for Advanced Interdisciplinary Studies, Henan University, Zhengzhou, Henan, China.
Methods (San Diego, Calif.)
|September 10, 2025
Summary
Genomic selection (GS) accelerates crop improvement by predicting genetic potential. A new deep learning method, ResDeepGS, enhances prediction accuracy, offering a robust solution for future food security.
Area of Science:
- Agricultural Science
- Genetics
- Computational Biology
Background:
- Genomic selection (GS) uses genomic markers for crop and animal breeding.
- Traditional methods struggle with complex gene interactions and large datasets.
- Deep learning offers potential for capturing nonlinear relationships and gene interactions.
Purpose of the Study:
- To propose a novel crop phenotype prediction method, ResDeepGS, using deep learning.
- To improve the efficiency and reliability of feature selection in genomic data.
- To enhance the accuracy of predicting crop traits for accelerated breeding.
Main Methods:
- Developed ResDeepGS, a deep learning model with feature selection and phenotype prediction modules.
- Utilized incremental recursive feature elimination for efficient feature selection.
- Employed an enhanced multi-layer convolutional neural network with residual structures and dropout for phenotype prediction.
Main Results:
- ResDeepGS outperformed state-of-the-art methods on wheat, maize, and soybean datasets.
- Achieved 5-9% improvement in prediction accuracy on the wheat dataset.
- Demonstrated superior performance in genomic selection tasks.
Conclusions:
- ResDeepGS offers a robust and adaptable solution for genomic selection.
- The method enhances crop breeding efficiency and contributes to addressing food security.
- Deep learning advancements significantly improve phenotype prediction accuracy.
