Parts | Days | Selection | Search | Updates | Downloads | Help
BP: Fachverband Biologische Physik
BP 27: Statistical Physics of Biological Systems 2 (joint session BP/DY)
BP 27.6: Talk
Thursday, September 8, 2022, 16:15–16:30, H16
The Influence of Contact Maps on RNA Structure Prediction — •Christian Faber1 and Alexander Schug1,2 — 1Jülich Supercomputing Centre, FZ Jülich — 2Steinbuch Centre for Computing, KIT
The 3d structure of Proteins and non coding RNA are essential for their function, but hard to determine via NMR or x-ray crystallography. Therefore an effective way of simulation with the knowledge of the sequence only would be a huge improvement. Impressive progress has been made in recent years, most notably AlphaFold2 for protein structure prediction using Machine Learning techniques. Such a break through is still missing for RNA.
For RNA, there are folding programs such as SimRNA, that simulate the structure with a physical force field [1]. The outcome can be improved by incorporating evolutionary data from homologous sequences. From the evolutionary data, we can make predictions about possible contacts in the form of contact maps [2].
We investigate how contact maps can influence prediction quality and what are particularly valuable contacts. From these insights we develop new measures for machine learning algorithms.
[1] Boniecki, M. J. et al. SimRNA: a coarse-grained method for RNA folding simulations and 3D structure prediction. Nucleic Acids Research 44, e63 (2016).
[2] Weigt, M., White, R. A., Szurmant, H., Hoch, J. A., Hwa, T. Identification of direct residue contacts in protein-protein interaction by message passing. PNAS 106, 67-72 (2009).