Gast ::
| Hauptseite > Publikationsdatenbank > Contact Maps in RNA StructurePrediction |
| Hauptseite > Publikationsdatenbank > Contact Maps in RNA StructurePrediction |
| Poster (After Call) | FZJ-2024-07276 |
; ; ;
2024
Abstract: Predicting the spatial structure of non-coding RNA (ncRNA) is an important task forunderstanding fundamental processes in living nature. Physical force fields are used to infer thestructure from a sequence using simulations on high-performance computers. However, the bestresults are obtained by incorporating evolutionary data via a binary mapping of contacts. Thesame phenomenon can be seen in protein structure prediction, where the groundbreakingAlphaFold2 also incorporates this step. Much work has been done in the past to optimise thealgorithms for simulations, but what are good contacts and why are these contacts important inthe first place is an unsolved puzzle. To find answers, we tried different contact map topologieson a well-defined test set of ncRNAs. We also looked at using fewer, but wisely chosen contactsand how this can improve prediction. To obtain our results, we ran many simulations forcomparison on the high performance cluster JUWELS with the RNA folding software SimRNAand used convolutional neural networks (CNN) to select contacts. Our results suggest that it isimportant to pay more attention to the selection of contacts, especially when developing machinelearning algorithms. Furthermore, good contacts not only ensure faster folding in the simulation,they are actually essential for correct folding. It seems that it is the additional constraints thatbring the physical force field into the more correct form.
|
The record appears in these collections: |