# Structure-guided pre-training

structRFM writes base-pairing interactions directly into masked language modelling and pre-trains on 21 million RNA sequence–structure pairs, paired with MUSES, a multi-source secondary-structure ensemble, and a dynamic masking ratio. It ranks highly in zero-shot homology classification and sets new benchmarks for secondary-structure prediction; its tertiary-structure adaptation, Zfold, improves both three-dimensional predictions and the secondary structures extracted from them on CASP15-natural, CASP16 and RNA-Puzzles, with performance comparable to AlphaFold3.

structRFM writes base-pairing interactions directly into masked language modelling and pre-trains on 21 million RNA sequence–structure pairs, paired with MUSES, a multi-source secondary-structure ensemble, and a dynamic masking ratio. It ranks highly in zero-shot homology classification and sets new benchmarks for secondary-structure prediction; its tertiary-structure adaptation, Zfold, improves both three-dimensional predictions and the secondary structures extracted from them on CASP15-natural, CASP16 and RNA-Puzzles, with performance comparable to AlphaFold3.