Breeding Value Pattern Recognition in Hybrid Crops (BreedPatH)


In Europe rapeseed is the most important source of vegetable oil used as a renewable raw material for food and industry. Rapid breeding of high yielding rapeseed cultivars is however more and more depending on predictive selection tools that can cope with dynamic practice conditions. BreedPatH aims to develop a completely new paradigm for prediction of breeding values in hybrid crops. Advanced machine learning will be implemented to train innovative prediction models for pattern recognition in complex, heterogeneous “omics” datasets, towards an augmented association of abstract data patterns with hybrid performance. In parallel, we will apply novel breeding methodologies that facilitate a rapid, genomic-assisted separation of poorly differentiated materials into distinct heterotic pools for hybrid breeding. These approaches will initially be applied in an experimental breeding programme for winter oilseed rape, a crop with low genetic diversity in the primary gene pool and for which strongly divergent heterotic pools could not be established, meaning that the full yield potential of hybrid cultivars has yet to be achieved. In a second step, BreedPatH will provide a next-generation predictive breeding toolbox that helps accelerate the shift to hybrid breeding in other classical inbred crops.


Principal Investigators
Kloft, Marius Prof. Dr. (Details) (Mechanical Learning (J))

Duration of Project
Start date: 09/2016
End date: 08/2019

Research Areas
Interactive and Intelligent Systems, Image and Language Processing, Computer Graphics and Visualisation

Last updated on 2021-21-01 at 11:44