{"version":1,"type":"story","url":"https://digestai.news/story/structrfm-model-trains-on-21-million-rna-pairs","json":"https://digestai.news/story/structrfm-model-trains-on-21-million-rna-pairs.json","markdown":"https://digestai.news/story/structrfm-model-trains-on-21-million-rna-pairs.md","slug":"structrfm-model-trains-on-21-million-rna-pairs","headline":"structRFM model trains on 21 million RNA pairs","summary":"structRFM is a new structure‑guided RNA foundation model trained on 21 million RNA sequence‑structure pairs. The authors add a multi‑source ensemble, MUSES, to reduce predictor bias and use a dynamic masking ratio that balances structure‑guided and nucleotide‑level masking. The model learns joint sequence and structure representations at the sequence‑, nucleotide‑ and pairwise levels.\n\nIn zero‑shot homology classification, structRFM ranks highly and sets new benchmarks for secondary‑structure prediction. Its tertiary‑structure adaptation, Zfold, improves three‑dimensional predictions and extracted secondary structures, achieving performance comparable to AlphaFold3 across CASP15‑natural, CASP16 and RNA‑Puzzles. The model also boosts function‑prediction tasks and generalises to newly released RNA families and long non‑coding RNAs.\n\nThe 21‑million‑pair dataset, model weights and source code are openly available. The pre‑training dataset can be downloaded from Zenodo (https://doi.org/10.5281/zenodo.18272040) and the model checkpoints from GitHub (https://github.com/heqin-zhu/structRFM).","keyPoints":["structRFM trained on 21 million RNA sequence‑structure pairs","MUSES ensemble reduces predictor bias and dynamic masking balances structure and nucleotide masking","Zfold matches AlphaFold3 performance on CASP15‑natural, CASP16 and RNA‑Puzzles"],"whyItMatters":"The model demonstrates that integrating structural information into RNA language models yields state‑of‑the‑art predictions, advancing both basic research and potential therapeutic design.","category":{"slug":"models","name":"Generative AI & Models","url":"https://digestai.news/category/models"},"entities":{"companies":["Zenodo","GitHub","Nature Machine Learning"],"models":["structRFM","Zfold","AlphaFold3"],"people":["Heqin Zhu"]},"firstPublishedAt":"2026-11-04T00:00:00Z","updatedAt":"2026-11-04T00:00:00Z","sourceCount":1,"hasPrimarySource":true,"sources":[{"outlet":"Nature Machine Learning","title":"An open structure-guided RNA foundation model for generalizable prediction","url":"https://nature.com/articles/s41586-026-11210-2","publishedAt":"2026-11-04T00:00:00Z","type":"primary","primary":true,"lead":true}],"sourceNotes":null,"discussions":[],"thread":null,"cite":{"text":"Digest AI, \"structRFM model trains on 21 million RNA pairs\", 4 November 2026, https://digestai.news/story/structrfm-model-trains-on-21-million-rna-pairs","publisher":"Digest AI","title":"structRFM model trains on 21 million RNA pairs","datePublished":"2026-11-04T00:00:00Z","url":"https://digestai.news/story/structrfm-model-trains-on-21-million-rna-pairs"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}