Lucas Belz-Koeling

Lucas Belz-Koeling is a PhD student developing machine-learning models for the prediction of activation energy barriers in chemical reactions, with a focus drug-like molecules and pharmaceutical applications.

PhD Research

Started: June 2026

Lucas’ research focuses on investigating novel model architectures, datasets, and chemical descriptors to improve the accuracy of activation energy barrier prediction. Through this, he aims to develop accurate and transferable models to predict reaction barriers and transition state geometries for a broad set of pharmaceutically relevant reactions.

Research Approach

His work employs advanced computational methods including:

• Machine Learning Architectures: Applying and adapting state-of-the-art models to learn structure-reactivity relationships
• Quantum Chemistry: Using QC packages to generate training data and validate model predictions
• Reaction Network Exploration: Enabling efficient discovery of reaction pathways by coupling fast ML predictions with automated reaction network generation

Background

Lucas completed an integrated Master’s degree in Materials Science at St Catherine’s College, University of Oxford. His Master’s thesis focussed on using DFT methods to study the fluorescence and stability of Tin-vacancy defects in diamond, with the goal of investigating their viability in quantum computing applications. This research into defect-state transitions, combined with experience in DFT, provided him with a strong background in computational chemistry and transition state theory.

Research Impact

Reaction barrier prediction is a key part of synthesis pathway planning in drug discovery. To assess the viability of a candidate molecule, one has to be able to predict the complexity and yield of its synthesis. Lucas’ work aims to improve the accuracy and scope of these predictions, enabling faster and more efficient drug development.