FMixFN

FMixFN applies a Bayesian genomic selection model using four zero-mean normal prior distributions to improve predictive accuracy and computational efficiency for estimating genomic breeding values in breeding programs.

Key Features:

• Iterative Conditional Expectation Algorithm: Uses an iterative conditional expectation algorithm to derive genomic estimated breeding values (GEBV) with improved accuracy and speed.
• Four Zero-Mean Normal Priors: Employs four zero-mean normal distributions as prior distributions with parameters determined from F2 population data.
• High Computational Efficiency: Demonstrates superior computational efficiency and scalability for large-scale sample data compared with methods such as GBLUP, SSgblup, MIX, BayesR, BayesA, and BayesB.
• Improved Predictive Ability: Provides enhanced predictive accuracy for selection using dense genetic markers relative to traditional methods.

Scientific Applications:

• Genomic Selection in Breeding: Application to genomic selection in animal and plant breeding programs for estimating breeding values.
• Elite Stock Selection: Selection of elite breeding stock using dense genetic marker data to improve prediction of genetic merit.
• Large-Scale Breeding Programs: Integration and analysis of large datasets and combined breeding schedules in extensive breeding programs.

Methodology:

Implements a Bayesian framework with four zero-mean normal prior distributions whose variances are determined from F2 population data and performs inference using an iterative conditional expectation algorithm to estimate GEBV.