Predicting RNA Folding Structures from Sequence Information
RNA is involved in virtually all cellular processes and RNA folding is an essential component in determining its function. Understanding this process and correctly predicting tertiary folding structures is an unsolved challenge, due to the complexity of the folding process and the scarcity of experimental data.
Deep Learning has proven to be a promising approach, with a recent breakthrough in predicting protein folding structures from sequence information. The ability to predict structures with almost experimental accuracy within hours or days has a profound impact on many scientific fields that are working with proteins. Due to the scarcity of available RNA only structures in the Protein Data Bank (PDB), similar results for RNA remain a challenge. This project is focused on the development of state-of-the-art Deep Learning methods, utilizing score-based generative modelling to estimate folding energy as well as advanced data augmentation and techniques from self-supervised learning to enable data efficient learning.
Starting date: 01.01.2022 / Doctoral Researcher
Thesis committee members: Markus Kollmann, tba