SKM 2023 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

BP: Fachverband Biologische Physik

BP 11: Poster Session I

BP 11.76: Poster

Dienstag, 28. März 2023, 12:30–15:30, P1

RNA Contact Prediction by Data Efficient Deep Learning — •Oskar Taubert¹, Fabrice Lehr², Alina Bazarova^3,4, Christian Faber³, Philipp Knechtges², Marie Weiel^1,4, Charlotte Debus^1,4, Daniel Coquelin^1,4, Achim Basermann², Achim Streit¹, Stefan Kesselheim^3,4, Markus Götz^1,4, and Alexander Schug³ — ¹Karlsruhe Institute of Technology, Karlsruhe, Germany — ²Deutsches Zentrum für Luft- und Raumfahrt, Köln, Germany — ³Forschungszentrum Jülich, Jülich, Germany — ⁴Helmholtz AI

On the path to full understanding of the structure-function relationship or even design of RNA, structure prediction would offer an intriguing complement to experimental efforts. Any deep learning on RNA structure, however, is hampered by the sparsity of labeled training data. Utilizing the limited data available, we here focus on predicting spatial adjacencies (*contact maps*) as a proxy for 3D structure. We explore the space of self-supervised learning for RNA multiple sequence alignments and focus on downstream contact prediction from latent attention maps.

Boosted decision trees in particular prove an advancement in contact prediction quality that can be further enhanced by finetuning the pretrained backbone. We name our model BARNACLE. Our conceptional advance is reflected by a considerable increase of precision and other metrics for contact prediction, thus promising to decrease the sequence-structure gap for RNA.