Department of Crop and Soil Sciences, Washington State University, Pullman, WA 99164, USA.
Department of Plant and Environmental Sciences, New Mexico State University, Las Cruces, NM 88003, USA.
Genes (Basel). 2020 Jul 11;11(7):779. doi: 10.3390/genes11070779.
Achieving optimal predictive ability is key to increasing the relevance of implementing genomic selection (GS) approaches in plant breeding programs. The potential of an item-based collaborative filtering (IBCF) recommender system in the context of multi-trait, multi-environment GS has been explored. Different GS scenarios for IBCF were evaluated for a diverse population of winter wheat lines adapted to the Pacific Northwest region of the US. Predictions across years through cross-validations resulted in improved predictive ability when there is a high correlation between environments. Using multiple spectral traits collected from high-throughput phenotyping resulted in better GS accuracies for grain yield (GY) compared to using only single traits for predictions. Trait adjustments through various Bayesian regression models using genomic information from SNP markers was the most effective in achieving improved accuracies for GY, heading date, and plant height among the GS scenarios evaluated. Bayesian LASSO had the highest predictive ability compared to other models for phenotypic trait adjustments. IBCF gave competitive accuracies compared to a genomic best linear unbiased predictor (GBLUP) model for predicting different traits. Overall, an IBCF approach could be used as an alternative to traditional prediction models for important target traits in wheat breeding programs.
实现最佳预测能力是提高基因组选择（GS）方法在植物育种计划中相关性的关键。本文探讨了基于项目的协同过滤（IBCF）推荐系统在多性状、多环境 GS 背景下的潜力。针对适应美国太平洋西北地区的冬小麦群体，评估了不同的 IBCF GS 场景。通过交叉验证进行跨年度预测，当环境之间存在高度相关性时，预测能力会得到提高。与仅使用单一性状进行预测相比，使用来自高通量表型的多个光谱性状可提高产量（GY）的 GS 准确性。通过使用 SNP 标记的基因组信息进行各种贝叶斯回归模型进行性状调整，是在评估的 GS 场景中提高 GY、抽穗期和株高准确性的最有效方法。与其他模型相比，贝叶斯 LASSO 用于表型性状调整的预测能力最高。与基因组最佳线性无偏预测（GBLUP）模型相比，IBCF 在预测不同性状方面具有竞争力。总体而言，IBCF 方法可以作为小麦育种计划中重要目标性状的传统预测模型的替代方法。
