LMU Researchers Combine Machine Learning and Molecular Dynamics to Discover Novel RNA Delivery Materials.

A research team led by professor Olivia Merkel, Chair of Drug Delivery at LMU and co-spokesperson of the Cluster for Nucleic Acid Therapeutics Munich (CNATM) has developed the first integrated platform that combines molecular dynamics (MD) simulations and machine learning (ML) to identify new polymeric materials for therapeutic RNA delivery. The study, recently published in the Journal of the American Chemical Society, introduces a computational tool called Bits2Bonds that enables de novo design and optimization of polymer-based RNA carriers. This research was conducted within Olivia Merkel’s ERC Consolidator Grant “RatInhalRNA”, which focuses on the development of innovative RNA delivery systems for pulmonary administration.

While experimental screening of polymer libraries is time-consuming and costly, purely computational approaches have so far fallen short due to limited data availability and high computational demands. The Bits2Bonds platform bridges this gap by integrating coarse-grained MD simulations that mimic key biological challenges, such as siRNA binding and membrane interaction, with machine learning–driven molecular design. The approach allows rapid virtual screening of thousands of potential carrier molecules before experimental validation, dramatically accelerating the discovery of effective and safe RNA nanocarriers.

“Our work demonstrates for the first time that combining physics-based simulation with data-driven optimization can efficiently guide the discovery of entirely new materials for RNA therapeutics,” says Olivia Merkel. “This method paves the way for a more rational, high-throughput design of polymeric delivery systems, moving us closer to personalized RNA medicines.”

The team validated their computational predictions by synthesizing and experimentally testing several polymer candidates for siRNA delivery, confirming strong correlation between simulated performance and biological efficacy. The resulting pipeline is highly modular and can be adapted to other types of polymers or nucleic acid modalities, such as mRNA or CRISPR-based therapies.