An algorithm to compute optimal genetic contributions in selection programs with large numbers of candidates

Area of Science:

• Animal breeding and genetics
• Quantitative genetics
• Bioinformatics

Background:

• Calculating optimal genetic contributions is crucial for maximizing genetic gain while managing inbreeding rates in livestock and aquaculture.
• Traditional methods struggle with very large numbers of selection candidates due to computational demands, particularly the inversion of large relationship matrices.

Purpose of the Study:

• To introduce and evaluate OCSELECT, a novel algorithm for computing optimal genetic contributions in large populations with a restricted inbreeding rate.
• To demonstrate the computational efficiency and accuracy of OCSELECT compared to existing software.

Main Methods:

• Developed the OCSELECT algorithm, which reformulates the inverse relationship matrix calculation to reduce computational complexity.
• The algorithm leverages matrix decomposition, inverting smaller parent-related matrices instead of the full candidate matrix.
• Compared OCSELECT's performance against the GENCONT software package using a large salmon dataset.

Main Results:

• OCSELECT successfully processed a dataset of 39,214 selection candidates and 45,846 total individuals, a size unmanageable by GENCONT without data splitting.
• Both OCSELECT and GENCONT (on subsets) yielded comparable results for genetic gain and selection numbers.
• The algorithm eliminated the need for preselection, allowing all candidates to be considered in the optimal contribution selection process.

Conclusions:

• OCSELECT provides a computationally efficient and accurate solution for optimal genetic contribution calculations in very large populations.
• The algorithm facilitates the selection of individuals that maximize genetic gain while effectively controlling inbreeding, even with extensive datasets.
• OCSELECT enhances the feasibility of advanced genetic selection strategies in large-scale breeding programs.