Robust Optimal Contribution Selection

TL;DR

This paper introduces a robust optimization approach to optimal contribution selection in breeding, accounting for data uncertainty, and provides open-source tools for practical implementation.

Contribution

It develops a novel robust OCS formulation using conic optimization and sequential quadratic programming, with an open-source Python package for real-world application.

Findings

Sequential quadratic programming with HiGHS performs well.

Robust OCS improves decision-making under data uncertainty.

Open-source 'robustocs' package facilitates adoption.

Abstract

Optimal contribution selection (OCS) is a selective breeding method that manages the conversion of genetic variation into genetic gain to facilitate short-term competitiveness and long-term sustainability in breeding programmes. Traditional approaches to OCS do not account for uncertainty in input data, which is always present and challenges optimization and practical decision making. Here we use concepts from robust optimization to derive a robust OCS problem and develop two ways to solve the problem using either conic optimization or sequential quadratic programming. We have developed the open-source Python package 'robustocs' that leverages the Gurobi and HiGHS solvers to carry out these methods. Our testing shows favourable performance when solving the robust OCS problem using sequential quadratic programming and the HiGHS solver.