Advances in multi-trait genomic prediction approaches: classification, comparative analysis, and perspectives.

Area of Science:

• Agricultural Science
• Genetics
• Bioinformatics

Background:

• Organismal traits are interconnected, presenting trade-offs that complicate breeding efforts.
• Multi-trait genomic prediction (MT-GP) is crucial for managing these complex trait relationships in crop improvement.
• Existing MT-GP models lack systematic comparative analysis, hindering optimal model selection.

Purpose of the Study:

• To classify and summarize existing multi-trait genomic prediction (MT-GP) models.
• To systematically compare the advantages, limitations, and assumptions of different MT-GP approaches.
• To guide the development of next-generation MT-GP models for crop breeding.

Main Methods:

• Classification of existing MT-GP models based on their principles and assumptions.
• Extensive comparative analysis of 10 traits in an Oryza sativa diversity panel.
• Cross-validation scenarios simulating practical breeding conditions.

Main Results:

• A comprehensive classification of current MT-GP models is presented.
• Comparative analysis revealed varying performance and limitations of different MT-GP approaches.
• Insights into model selection for specific breeding objectives were gained.

Conclusions:

• Systematic comparison of MT-GP models is essential for effective crop breeding.
• Understanding model assumptions and limitations is key to successful trait improvement.
• This work provides a foundation for developing advanced MT-GP strategies to address crop breeding challenges.