FMixFN: A Fast Big Data-Oriented Genomic Selection Model Based on an Iterative Conditional Expectation algorithm.

Area of Science:

• Animal breeding and genetics
• Quantitative genetics
• Bioinformatics

Background:

• Genomic selection (GS) uses dense genetic markers to identify elite breeding stock.
• Current GS software often suffers from low prediction accuracy, computational inefficiency, and limitations with large datasets.

Purpose of the Study:

• To develop an optimized genomic prediction model (FMixFN) that improves predictive ability and computational efficiency.
• To address the limitations of existing genomic selection software for large-scale applications.

Main Methods:

• Developed FMixFN, a Bayes genomic selection model utilizing four zero-mean normal distributions as prior distributions.
• Determined prior distribution variances using an F2 population.
• Employed an iterative conditional expectation algorithm for accurate and rapid genomic estimated breeding value (GEBV) calculation.

Main Results:

• FMixFN demonstrated superior computational efficiency and predictive ability compared to GBLUP, SSgblup, MIX, BayesR, BayesA, and BayesB.
• The model effectively handles large-scale sample data, suitable for large breeding programs.
• FMixFN achieved stable predictive ability and high computational efficiency.

Conclusions:

• FMixFN is a big data-oriented genomic selection model offering significant advantages in predictive ability and computational speed.
• The model's scalability makes it a valuable tool for modern breeding companies and complex breeding schedules.
• FMixFN represents a promising advancement in genomic selection methodology.