All news

Nature publishes open structRFM model for predicting RNA structure and function

Prepared by Sintelligent with the help of SI. Published in SI: . Updated: .

Open data and weights will let researchers apply one model to different RNA analysis tasks and test it on new families of molecules.

According to Nature's description, structRFM was trained on 21 million pairs of RNA sequences and their corresponding structures. Combining these representations improves secondary and tertiary structure prediction, functional inference, and generalization to previously unseen RNA families without additional training. The complete set of 21 million pairs, model weights, and code are open to researchers. Publication in Nature consolidates a result previously available as a preprint.

Sources 1