Multi-view BLUP: a promising solution for post-omics data integrative prediction.

Area of Science:

• Agricultural Science
• Genetics
• Bioinformatics

Background:

• Phenotypic prediction accelerates plant breeding by forecasting crop traits.
• Multi-view data offers complementary information for comprehensive biological characterization.
• Integrating diverse data sources is crucial for advancing breeding strategies.

Purpose of the Study:

• To propose a novel Multi-View Best Linear Unbiased Prediction (MVBLUP) method for phenotypic prediction.
• To effectively integrate multi-view data for enhanced prediction accuracy in plant breeding.
• To assess the performance of MVBLUP across different crop datasets.

Main Methods:

• Developed the MVBLUP method by integrating multi-view information into the Best Linear Unbiased Prediction (BLUP) model.
• Utilized a differential evolution algorithm with early stopping to determine the importance of data views.
• Constructed a multi-view kinship matrix for incorporation into the BLUP model.

Main Results:

• The MVBLUP method demonstrated improved prediction accuracy compared to single-view approaches.
• Average prediction accuracy gains ranged from 0.038 to 0.201 across four multi-view crop datasets.
• Empirical experiments validated the effectiveness of MVBLUP in integrating multi-view data.

Conclusions:

• MVBLUP is an effective method for integrating multi-view data in phenotypic prediction.
• The proposed method offers significant improvements in prediction accuracy for plant breeding.
• MVBLUP provides a robust framework for leveraging diverse data sources in crop improvement.