Professor Li Lin's team, in collaboration with professor Chen Hong's team, both from Huazhong Agricultural University, published a paper titled "AutoGP: An Intelligent Breeding Platform for Enhancing Maize Genomic Selection" in Plant Communications online on Jan 9.

The study introduced the intelligent breeding platform AutoGP (http://autogp.hzau.edu.cn), which aims to simplify the application process of genomic selection (GS) technology by combining high-quality single nucleotide polymorphism (SNP) selection, a convenient field phenotype collection system, and a fully functional GS module, providing breeders with a simple and reliable breeding tool.

Through validation on the maize CUBIC population, AutoGP is able to select the optimal parental combinations, offering a new pathway to accelerate crop improvement and streamline breeding processes.

AutoGP offers five core functional modules: model training, phenotype prediction, integrated training and prediction, optimal parental selection, and integrated training and selection. Users can flexibly choose from a model library containing five machine learning models (SVM, XGBoost, GBDT, MLP, and RF) or four commonly used deep learning models (DeepGS, DLGWAS, DNNGP, and SoyDNGP) to determine the most suitable GS model.

Furthermore, AutoGP is equipped with an environmental information embedding module that supports model training, phenotype prediction, and integrated training and prediction. The introduction of this module enables users to conduct unified model phenotype predictions for different regions, significantly enhancing the adaptability and practicality of the models.

AutoGP offers breeders an integrated solution. [Photo/news.hzau.edu.cn]