Julian M. Herold

This paper introduces RNA-MDLM, a model for generating and inpainting non-coding RNA sequences, showing it produces sequences with natural composition and folding statistics.

The paper uses high-resolution vascular data from mouse brains to simulate tumor growth, finding vessel density strongly influences growth rate and enabling brain-wide tumor predictions.

NucleicBERT is a self-supervised masked-language model for RNA that learns contextual representations without alignments or evolutionary data, outperforming existing models in structural and functional tasks.

NucleoSeeker curates high-quality RNA structure datasets by filtering the PDB database, improving dataset quality for training and evaluating RNA structure prediction methods.