Optimising parent selection in plant breeding: comparing metaheuristic algorithms for genotype building.

Area of Science:

• Plant breeding and genetics
• Computational biology
• Bioinformatics

Background:

• Genomic-assisted breeding aims to develop superior crop genotypes by stacking desirable haplotypes.
• Optimal Haplotype Selection (OHS) faces challenges in identifying optimal parent sets due to combinatorial complexity.
• Metaheuristic search algorithms (MSAs) offer potential solutions for complex optimisation problems in parent selection.

Purpose of the Study:

• To evaluate the performance of MSAs (GA, DE, PSO, SA) for parent selection in genotype building (GB).
• To compare two GB objectives: OHS and Optimal Population Value (OPV).
• To assess the long-term impact of OHS and OPV on genetic gain and diversity.

Main Methods:

• Utilised a wheat population (583 lines) genotyped for 29,972 SNPs and phenotyped for stripe rust.
• Applied GA, DE, PSO, and SA to optimise parent selection for OHS and OPV.
• Simulated 100 breeding cycles to evaluate long-term genetic gain and diversity retention.

Main Results:

• Genetic Algorithm (GA) demonstrated high fitness and rapid convergence.
• Optimal Haplotype Selection (OHS) outperformed OPV and Genomic Estimated Breeding Value (GEBV)-based selection in long-term genetic gain and diversity.
• OHS maintained heterozygosity and additive variance, crucial for sustainable crop improvement.

Conclusions:

• Metaheuristic algorithms effectively optimise parent selection for genotype building.
• OHS is a promising strategy for enhancing long-term genetic gain and maintaining diversity in crop breeding programs.
• Prioritising collective parent set performance over individual ranking improves selection outcomes in genomic-assisted breeding.