Dr. Mehdi D. Davari
Department of Bioorganic Chemistry, Leibniz Institute of Plant Biochemistry, Weinberg 3,
06120 Halle, Germany, Email: mehdi.davari@ipb-halle.de
Abstract.
The ability to tailor protein function underpins progress in biotechnology, medicine, and sustainable biocatalytic processes. Yet the immense size of protein sequence space, combined with experimental limitations in screening capacity, makes the identification of functional variants highly challenging. Recent advances in structure prediction driven by deep learning, alongside increasingly accurate molecular simulation methods, have broadened our capacity to interpret protein structure, dynamics, and function at atomic resolution.
Nevertheless, translating this knowledge into systematic protein discovery and optimization of
new proteins remains a major bottleneck.
This lecture will highlight our efforts to apply physics-based modeling and data-driven learning to advance protein and enzyme engineering. Molecular simulations reveal mechanistic determinants of activity, stability, and environmental adaptation, enabling rational strategies for designing proteins with tailored properties, such as enhanced performance in nonconventional media or controlled interactions with materials and light. Complementarily, machine learning approaches allow rapid exploration of sequence landscapes by capturing relationships between sequence variation and functional outcomes from experimental and evolutionary information.
I will present our computational frameworks, including PyPEF and MERGE, which combine evolutionary statistics, biophysical modeling, and ML to prioritize promising variants even when training data are scarce. By combining mechanistic insight with predictive modeling, these approaches reduce experimental effort while expanding accessible regions of sequence space. The talk will conclude with perspectives on how use of simulations, curated datasets, and ML methods can accelerate the discovery of functional proteins and support sustainable innovation in protein engineering.