TUM Campus Straubing for Biotechnology and Sustainability and Computomics recently collaborated on machine learning models for phenotype prediction. The joint effort was part of the CropML Grant funded by BMBF. As a result of the joint effort, Frontiers in Plant Science published the following research article:
A comparison of classical and machine learning-based phenotype prediction methods on simulated data and three plant species
The goal of the collaboration was to predict plant phenotypes from genotype data, and the paper demonstrates how this is achieved. The final, formatted version of the article can be found here.
TUM implemented a machine learning platform to systematically compare twelve machine learning algorithms on simulated data as well as on three real-world breeding datasets. Computomics contributed two of the three data sets. The outcome demonstrates which machine learning models work best for phenotype prediction, and paves the way for further work as part of this grant. We found that the machine learning model Elastic Net gave the best results compared to the other models. Further research is to follow.
This again demonstrates the power of machine learning to predict phenotypes. We are proud to collaborate with universities in developing machine learning models.
We presented more about the paper and about our machine learning-based solution xSeedScore at our presentation at EUCARPIA Biometrics in Plant Breeding Conference in Paris (21-23 September). Rupashree Dass presented a poster on how machine learning can be used for more accurate phenotype predictions on Wednesday, 21 September 2022, 18:00 - 20:00.
Feel free to contact Rupa directly at the event or via e-mail.
Photo by Meredith Petrick on Unsplash