Operations Research Applied to Agricultural Genomics

Title
Operations Research Applied to Agricultural Genomics to Enable Better R&D Marker Picking Process

Presenters
Qinglin Duan

Description
Advances in agricultural genomics offer the techniques to speed up the process of developing crops with promising agronomic traits. Genetic markers have long been used for characterization of plant genetic diversity and exploitation in crop improvement. The DNA marker technology has enabled the breeding of Bayer Crop Science (BCS) elite lines with targeted selection of desirable gene or gene combinations in our breeding programs. It is an integral part of our fundamental genomics R&D pipeline since the phenotypic selection for complex agronomic traits is generally difficult, unpredictable and challenging.

The Marker Set Optimizer presented here involves selecting an optimal set of genetic markers to be used in the various molecular breeding workflows. By leveraging the rich crop data assets we collected over so many years, their genotype imputation and optimization modeling, BCS scientists can now make holistic, unified decisions by combined consideration of marker informativeness, coverage, marker availability, quality and total cost. The optimization model has been tested on our Marker-Assisted Selection (MAS) and Marker-Assisted BackCrossing (MABC) workflows for complex trait stacking. The results showed the new algorithm can handle those most complicated cases very well by removing the redundant markers through optimal utilization of marker-related materials and limiting unnecessary manual work.

The optimization model is implemented on the Kubeflow platform with our cloud-based optimization service powered by our own cloud team. The innovative Kubeflow platform provides a modular technology-agnostic solution as part of the future automated molecular testing workflow.

Qinglin Duan is a Senior Data Scientist at Bayer Crop Science (BCS), specializing in the application of Operations Research techniques for decision support across the enterprise. Qinglin holds a Ph.D. in Industrial Engineering from Louisiana State University and has 7+ years of industry experience in decision science. She finds her work the most exciting because it allows her interests in system engineering and applied mathematics to provide holistic solutions to complex business process modeling. Qinglin is also passionate about talent acquisition and engagement, actively involved in the Institute for Operations Research and the Management Sciences (INFORMS) in a variety of ways. She serves on the INFORMS committee for the Early Career Professionals’ network since 2017.

Our event will start with 30 minutes of networking in zoom breakout rooms. We will then have a 45-60 minute lecture followed by 15 minutes of questions. Attendees are welcome to stay after and chat.