What does it take to learn the rules of RNA base pairing? Much less than you think

Machine Learning


  • Penic, RJ, Vlasic, T., Huber, RG, Wan, Y. & Sikic, M. RiNALMo: A general-purpose RNA language model can successfully generalize structure prediction tasks. nut. common. 165671 (2025).

  • Akiyama, M. & Sakebara, Y. Informative RNA base embeddings for RNA structure alignment and clustering by deep representation learning. NAR Genome. Bioinform. 44 (2022).

    Google Scholar

  • Lin, Z. et al. Predicting the evolutionary scale of atomic-level protein structures using language models. science 3791123–1130 (2023).

    Google Scholar

  • Gong, T. & Bu, D. Language models enable zero-shot prediction of RNA secondary structure, including pseudoknots. BioRxiv https://www.biorxiv.org/content/10.1101/2024.01.27.577533v1 (2024).

  • De Laharte, AA et al. Diverse databases and machine learning models to narrow the generalization gap in RNA structure prediction. Science. advanced. 12eadz4967 (2024).

  • Wang, N. et al. Versatile RNA language modeling with motif-aware pretraining and type-guided fine-tuning. nut. Mach. intelligence. 6548–557 (2024).

    Google Scholar

  • Yu, H., et al. Interpretable RNA-based models for exploring functional RNA motifs in plants. nut. Mach. intelligence. 61616–1625 (2024).

    Google Scholar

  • Zou, S. et al. Large-scale infrastructure models for predicting RNA function and structure. BioRxiv https://doi.org/10.1101/2024.11.28.625345 (2024).

  • Kazuya Sato, Masato Akiyama, Yuko Sakakibara RNA secondary structure prediction using deep learning with thermodynamic integration. nut. common. 12941 (2021).

    Google Scholar

  • Fu, L. et al. UFold: Fast and accurate RNA secondary structure prediction using deep learning. Nucl.Acid resolution. 50e14–e14 (2022).

    Google Scholar

  • Shin, J. et al. Improving prediction of RNA secondary structure and tertiary base pairing using evolutionary profiles, mutational coupling, and two-dimensional transfer learning. bioinformatics 372589–2600 (2021).

    Google Scholar

  • Sato, K. & Hamada, M. Recent trends in RNA informatics: An overview of machine learning and deep learning for RNA secondary structure prediction and RNA drug discovery. Easy. Bioinform. twenty fourbbad186 (2023).

    Google Scholar

  • Da Silva, PT et al. Nucleotide-dependent analysis of DNA language models reveals functional elements of the genome. nut. Genet. 572589–2602 (2025).

    Google Scholar

  • Rivas, E., Clements, J. & Eddy, SR Statistical tests of conserved RNA structure show a lack of evidence for the structure of lncRNAs. nut. method 1445–48 (2017).

    Google Scholar

  • Chomsky, N. Three models for describing language. IRE transformer. information theory 2113–124 (1956).

    Google Scholar

  • Dowell, RD & Eddy, SR Evaluation of several lightweight probabilistic context-free grammars for RNA secondary structure prediction. BMC Bioinform. 571 (2004).

    Google Scholar

  • Bradbury, J. et al. JAX: Composable transformation of Python+NumPy programs. http://github.com/google/jax (2018).

  • Rabiner, LR A tutorial on hidden Markov models and selected applications in speech recognition. procedure IEEE 77257–286 (1989).

    Google Scholar

  • Eddy, SR and Durbin, R. RNA-seq analysis using covariance models. Nucl.Acid resolution. twenty two2079–2088 (1994).

    Google Scholar

  • Rivas, E., Lang, R. & Eddy, SR A suite of complex probabilistic models for RNA secondary structure prediction, including the nearest neighbor model. RNA 18193–212 (2012).

    Google Scholar

  • Knudsen, B. & Hein, J. Pfold: RNA secondary structure prediction using probabilistic context-free grammars. Nucl.Acid resolution. 313423–3428 (2003).

    Google Scholar

  • Justyna, M., Antczak, M., Szachniuk, M. Machine learning for RNA 2D structure prediction benchmarked on experimental data. Easy. Bioinform. twenty four1–9 (2023).

    Google Scholar

  • Lari, K. & Young, SJ Estimating probabilistic context-free grammars using an inside-outside algorithm. Calculate. speech lang. 435–56 (1990).

    Google Scholar

  • Lari, K. & Young, SJ Applications of probabilistic context-free grammars using inside-outside algorithms. Calculate. speech lang. 5237–257 (1991).

    Google Scholar

  • McCaskill, JS Equilibrium partition functions and base pairing probabilities of RNA secondary structure. biopolymer 291105–19 (1990).

    Google Scholar

  • Zuker, M. Mfold Web Server for Hydrogenation Prediction of Nucleic Acid Folding. Nucleic acid research institute 313406–3415 (2003).

    Google Scholar

  • Lorenz, R. et al. ViennaRNA Package 2.0. Algorithm Mol. Biol. 61748–7188 (2011).

    Google Scholar

  • Reuter, JS & Mathews, DH RNA Structure: Software for the prediction and analysis of RNA secondary structure. BMC Bioinform. 1110 (2010).

    Google Scholar

  • Rivas, E. Four components of single-sequence RNA secondary structure prediction: A unifying perspective. RNA Biol. 101185–1196 (2013).

    Google Scholar

  • Eisner, J. The inside-outside and forward-backward algorithms are just backprops (tutorial paper). In the process. First Workshop on Structured Prediction for NLP (edited by Balsubramanian, VN et al.) 1–17 (Association for Computational Linguistics, 2016).

  • Matthies, MC, Krueger, R., Torda, AE & Ward, M. Differentiable partition function calculations for RNA. NAR 52e14 (2024).

    Google Scholar

  • Krueger, RK & Ward, M. JAX-RNAfold: Scalable differentiable folding. bioinformatics 41btaf203 (2025).

  • Krueger, RK, Aviran, S., Mathews, DH, Zuber, J. & Ward, M. Differentiable folding for nearest neighbor model optimization. arXiv preprint https://arxiv.org/abs/2503.09085 (2025).

  • Brown, JW Ribonuclease P Database. Nucl.Acid resolution. 27314 (1999).

    Google Scholar

  • Szikszai, M., Wise, M., Datta, A., Ward, M. & Mathews, DH Deep learning models for RNA secondary structure prediction (probably) do not generalize across families. bioinformatics 383892–3899 (2022).

    Google Scholar

  • Szikszai, M. et al. Deep learning for RNA secondary structure determination: assessing generalizability and expanding the scope of traditional methods. RNA 32428–442 (2026).

  • Cocco, S., Monasson, R. & Weigt, M. From principal components of protein coevolution to direct conjugation analysis: Structure prediction requires low eigenvalue modes. PLoS Computing. Biol. 9e1003176 (2013).

    Google Scholar

  • Gao, W., Yang, A. & Rivas, E. 13 questionable methods for detecting conserved structural RNAs. IUBMB Life 75471–492 (2022).

    Google Scholar

  • Rivas, E. RNA structure prediction using positive and negative evolutionary information. PLOS Computing. Biol. 16e1008387 (2020).

    Google Scholar

  • Karan, A. & Rivas, E. All-at-once RNA folding with evolution-informed 3D motif prediction. nut. method twenty two2094–2106 (2025).

    Google Scholar

  • Danay, P. et al. bpRNA: Large-scale automatic annotation and analysis of RNA secondary structure. NNucleic acid resolution 465381–5394 (2018).

    Google Scholar

  • Sloma, MF & Mathews, DH Precise calculation of loop formation probabilities identifies folding motifs in RNA secondary structure. RNA twenty two1808–1818 (2016).

    Google Scholar

  • Wheeler, TJ & Eddy, SR nmmer: DNA homology search using profile HMMs. bioinformatics 292487–2489 (2013).

    Google Scholar



  • Source link