Shujun He - Ribonanza: deep learning of RNA structure through dual crowdsourcing скачать в хорошем качестве

Shujun He - Ribonanza: deep learning of RNA structure through dual crowdsourcing

RNA CASP Special Interest Group (SIG) #15 Tuesday February 13 2024 Prediction of RNA structure from sequence remains an unsolved problem, and progress has been slowed by a paucity of experimental data. Here, we present Ribonanza, a dataset of chemical mapping measurements on two million diverse RNA sequences collected through Eterna and other crowdsourced initiatives. The Ribonanza measurements enabled solicitation, training, and prospective evaluation of diverse deep neural networks through a Kaggle challenge, followed by distillation into a single, self-contained model called RibonanzaNet. When fine tuned on previous datasets, RibonanzaNet achieves state-of-the-art performance in modeling RNA chemical stability and RNA secondary structure. The latter enables three-dimensional structure modeling of RNA-only targets from the recent CASP trials with accuracy approaching human expert groups.