Exploiting genomic knowledge in optimising molecular breeding programmes: algorithms from evolutionary computing.

Area of Science:

• Genetics
• Computational Biology
• Animal and Plant Breeding

Background:

• Limited studies quantify benefits of molecular genetic markers in experimental breeding.
• Optimal mating strategies for experimental breeding programs remain underexplored.
• In silico modeling using evolutionary computation (EC) is proposed as a solution.

Purpose of the Study:

• To explore the application of evolutionary computation (EC) for optimizing experimental breeding programs.
• To compare the effectiveness of genetic algorithms (G-algorithms) with knowledge of the search space against those without (F-algorithms).
• To assess the implications of machine learning-based algorithms for post-genomic era breeding.

Main Methods:

• Reviewed EC approaches relevant to experimental breeding.
• Conducted in silico experiments on a biologically relevant landscape.
• Compared G-algorithms (informed) and F-algorithms (uninformed) in optimizing breeding parameters.

Main Results:

• Both F-algorithms and G-algorithms demonstrated comparable quality and effectiveness on the tested landscapes.
• G-algorithms were not pre-equipped with knowledge of epistatic pathway interactions.
• The study provides freely available, non-proprietary code for in silico modeling.

Conclusions:

• In silico modeling using EC is a viable approach for optimizing experimental breeding programs.
• Machine learning-based algorithms have significant potential for optimizing breeding in the post-genomic era.
• Further research may explore G-algorithms with prior knowledge of genetic interactions.