Genomic selection (GS) holds great promise to accelerate breeding progress of plants, and the advancement of across-population GS is crucial for realizing its full potential. However, conventional across-population GS relies heavily on precisely aligned markers across dense genotypes, while the viability of utilizing flexible low-density markers remains underexplored. This study developed GS-Impute, a residual convolutional denoising autoencoder-based neural network framework that enables accurate genotype imputation for low-density across-population GS. A key breakthrough of GS-Impute is the automatic matching algorithm, which successfully resolves the persistent challenge of targeted training with both sporadic and systematic missing data. Additionally, GS-Impute incorporates a data augmentation approach and several advanced techniques to increase the imputation accuracy, including residual blocks, dynamic learning rate optimization, and layer normalization. Comprehensive evaluation across rice and maize breeding populations demonstrated that GS-Impute outperforms the latest versions of established benchmark tools, including Beagle5.4, Minimac4, and STICI. Importantly, results reveal that GS-Impute makes across-population GS feasible with low-density markers, establishing a resource-efficient strategy that has the potential to transform genomic breeding programs.
Wang et al. (Sun,) studied this question.