MSXFGP: combining improved sparrow search algorithm with XGBoost for enhanced genomic prediction.

Area of Science:

• Plant breeding
• Genomics
• Bioinformatics

Background:

• High-throughput sequencing cost reduction drives genomic selection advancements.
• Existing genomic selection methods struggle with prediction accuracy in practical applications.

Purpose of the Study:

• To develop an improved genomic selection method for enhanced prediction accuracy in plant breeding.
• To optimize XGBoost parameters and feature selection for genomic prediction.

Main Methods:

• Proposed MSXFGP method utilizing a multi-strategy improved sparrow search algorithm (SSA).
• Incorporated logistic chaos mapping, elite learning, adaptive parameter adjustment, Levy flight, and early stop strategy into SSA.
• Optimized XGBoost parameters and performed feature selection concurrently using the improved SSA.

Main Results:

• MSXFGP demonstrated comparable or superior prediction accuracy against six existing genomic selection models across six datasets.
• The method showed enhanced accuracy, particularly when evaluated using the coefficient of determination R².
• A user-friendly Python utility for MSXFGP application was developed and is publicly available.

Conclusions:

• MSXFGP provides a novel and effective approach for plant genome selection.
• The method offers improved prediction accuracy, addressing limitations of current genomic selection techniques.