Gesa Birkenbach (Gottfried Wilhelm Leibniz Universität Hannover)
Gesa Birkenbach's Master thesis was awarded with the Prize for the Best Plant Science Master Thesis, which was carried out at Gottfried Wilhelm Leibniz Universität Hannover in the year 2025 with the title:
Machine Learning Identification of Triterpene-Modifying Cytochrome P450 Monooxygenases
How machine learning can be used to facilitate the functional characterization of plant plant cytochrome P450 (CYP) enzymes by making predictions on the enzymes‘ substrate group.
Cytochrome P450 enzymes (CYPs) form one of the largest enzyme superfamilies with functions in many different metabolic pathways. They are especially important in the biosynthesis of plant triterpenoids, a large group of structurally and functionally diverse defense compounds. However, the identification of CYPs with activities in a specific metabolic pathway is often hindered by the multitude of CYP enzymes in the plant organism. To facilitate the selection of relevant CYP enzymes, a machine learning model was trained in this master’s thesis to predict whether a cytochrome P450 enzyme might be involved in the biosynthesis of plant triterpenoids based on the enzyme’s protein sequence. To understand the working mechanism behind the model, it was investigated which sequence properties were relevant for the model’s decision-making process. By applying the model to the protein sequences of a test plant, predictions could be made for CYP enzymes that might be involved in yet unknown metabolic pathways and could thus be helpful in their elucidation. An analysis of incorrect predictions furthermore provided valuable insights into the model’s weaknesses, which will guide future improvements to the model. The master’s thesis demonstrates the potential of interdisciplinary research approaches and the possible benefits of artificial intelligence for the plant sciences.
___
Gesa Birkenbach conducted this work at the Institute of Cell Biology and Biophysics and at the Institute of Botany in the working groups of Prof. Sophia Rudorf and Prof. Jakob Franke.