Explicit dynamic cross-strand interactions for DNA sequence language modelling

Machine Learning


  • Devlin, J., Chang, M.-W., Lee, K. & Toutanova, K. BERT: pre-training of deep bidirectional transformers for language understanding. In Proc. 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) (eds Burstein, J. et al.) 4171–4186 (Association for Computational Linguistics, 2019).

  • Achiam, J. et al. GPT-4 technical report. Preprint at https://arxiv.org/abs/2303.08774 (2023).

  • Avsec, Ž. et al. Advancing regulatory variant effect prediction with AlphaGenome. Nature 649, 1206–1218 (2026).

    Article 

    Google Scholar 

  • Brixi, G. et al. Genome modelling and design across all domains of life with Evo 2. Nature 652, 1349–1361 (2026).

  • Luo, X., Kang, X. & Schönhuth, A. Predicting the prevalence of complex genetic diseases from individual genotype profiles using capsule networks. Nat. Mach. Intell. 5, 114–125 (2023).

    Article 

    Google Scholar 

  • Benegas, G., Albors, C., Aw, A. J., Ye, C. & Song, Y. S. A DNA language model based on multispecies alignment predicts the effects of genome-wide variants. Nat. Biotechnol. 43, 1960–1965 (2025).

    Article 

    Google Scholar 

  • Xu, A. et al. SNPBag: a foundation model for multitask genome-scale SNP analysis. Preprint at Research Square https://doi.org/10.21203/rs.3.rs-7593414/v1 (2025).

  • Li, H. et al. BMFM-DNA: a SNP-aware DNA foundation model to capture variant effects. Preprint at https://arxiv.org/abs/2507.05265 (2025).

  • Gao, Z., Liu, Q., Zeng, W., Jiang, R. & Wong, W. H. EpiGePT: a pretrained transformer-based language model for context-specific human epigenomics. Genome Biol. 25, 310 (2024).

    Article 

    Google Scholar 

  • Zeng, W., Guo, H., Liu, Q. & Wong, W. H. Improving polygenic prediction from whole-genome sequencing data by leveraging predicted epigenomic features. Proc. Natl Acad. Sci. USA 122, e2419202122 (2025).

    Article 

    Google Scholar 

  • Zhou, H., Shrikumar, A. & Kundaje, A. Towards a better understanding of reverse-complement equivariance for deep learning models in genomics. In Proc. 16th Machine Learning in Computational Biology Meeting (eds Knowles, D. A. et al.) 1–33 (PMLR, 2022).

  • Ji, Y., Zhou, Z., Liu, H. & Davuluri, R. V. DNABERT: pre-trained bidirectional encoder representations from transformers model for DNA-language in genome. Bioinformatics 37, 2112–2120 (2021).

    Article 

    Google Scholar 

  • Zhou, Z. et al. DNABERT-2: efficient foundation model and benchmark for multi-species genomes. In Proc. International Conference on Learning Representations https://openreview.net/forum?id=oMLQB4EZE1 (OpenReview, 2024).

  • Mallet, V. & Vert, J.-P. Reverse-complement equivariant networks for DNA sequences. Adv. Neural Inf. Process. Syst. 34, 13511–13523 (2021).

    Google Scholar 

  • Schiff, Y. et al. Caduceus: bi-directional equivariant long-range DNA sequence modeling. In Proc. 41st International Conference on Machine Learning (eds Salakhutdinov, R. et al.) Vol. 235, 43632–43657 (PMLR, 2024).

  • Dalla-Torre, H. et al. Nucleotide transformer: building and evaluating robust foundation models for human genomics. Nat. Methods 22, 287–297 (2025).

    Article 

    Google Scholar 

  • Nguyen, E. et al. HyenaDNA: long-range genomic sequence modeling at single nucleotide resolution. Adv. Neural Inf. Process. Syst. 36, 43177–43201 (2023).

    Google Scholar 

  • Sanabria, M., Hirsch, J., Joubert, P. M. & Poetsch, A. R. DNA language model GROVER learns sequence context in the human genome. Nat. Mach. Intell. 6, 911–923 (2024).

    Article 

    Google Scholar 

  • Press, O., Smith, N. A. & Lewis, M. Train short, test long: attention with linear biases enables input length extrapolation. In Proc. International Conference on Learning Representations https://openreview.net/forum?id=R8sQPpGCv0 (OpenReview, 2022).

  • Dao, T. FlashAttention-2: faster attention with better parallelism and work partitioning. In Proc. International Conference on Learning Representations (Curran Associates, 2024).

  • Lee, N. K., Tang, Z., Toneyan, S. & Koo, P. K. EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations. Genome Biol. 24, 105 (2023).

    Article 

    Google Scholar 

  • Ma, M. Reverse-complement consistency for DNA language models. Preprint at https://arxiv.org/abs/2509.18529 (2025).

  • Duan, Q. et al. JanusDNA: a powerful bi-directional hybrid DNA foundation model. Adv. Neural Inf. Process. Syst. 38, 68791–68818 (2026).

    Google Scholar 

  • Shrikumar, A., Greenside, P. & Kundaje, A. Reverse-complement parameter sharing improves deep learning models for genomics. Preprint at bioRxiv https://doi.org/10.1101/103663 (2017).

  • Choi, C. H. et al. DNA dynamically directs its own transcription initiation. Nucleic Acids Res 32, 1584–1590 (2004).

    Article 

    Google Scholar 

  • Rohs, R. et al. The role of DNA shape in protein–DNA recognition. Nature 461, 1248–1253 (2009).

    Article 

    Google Scholar 

  • Kabir, A. et al. DNA breathing integration with deep learning foundational model advances genome-wide binding prediction of human transcription factors. Nucleic Acids Res. 52, e91 (2024).

  • Gordân, R. et al. Genomic regions flanking E-box binding sites influence DNA binding specificity of bHLH transcription factors through DNA shape. Cell Rep. 3, 1093–1104 (2013).

    Article 

    Google Scholar 

  • Chen, Y. et al. Structure of p53 binding to the BAX response element reveals DNA unwinding and compression to accommodate base-pair insertion. Nucleic Acids Res. 41, 8368–8376 (2013).

    Article 

    Google Scholar 

  • Zhou, T. et al. Quantitative modeling of transcription factor binding specificities using DNA shape. Proc. Natl Acad. Sci. USA 112, 4654–4659 (2015).

    Article 

    Google Scholar 

  • Mitra, R. et al. Geometric deep learning of protein–DNA binding specificity. Nat. Methods 21, 1674–1683 (2024).

    Article 

    Google Scholar 

  • Adam, S., Klingel, V., Radde, N. E., Bashtrykov, P. & Jeltsch, A. On the accuracy of the epigenetic copy machine: comprehensive specificity analysis of the DNMT1 DNA methyltransferase. Nucleic Acids Res. 51, 6622–6633 (2023).

    Article 

    Google Scholar 

  • Bashtrykov, P. et al. Specificity of Dnmt1 for methylation of hemimethylated CpG sites resides in its catalytic domain. Chem. Biol. 19, 572–578 (2012).

    Article 

    Google Scholar 

  • Senadeera, D. C. et al. Dual branch VideoMamba with gated class token fusion for violence detection. Preprint at https://arxiv.org/abs/2506.03162 (2025).

  • Yu, T., Cheng, L., Khalitov, R., Olsson, E. B. & Yang, Z. Self-distillation improves self-supervised learning for DNA sequence inference. Neural Netw. 183, 106978 (2025).

    Article 

    Google Scholar 

  • Yang, H. et al. HAD: hybrid architecture distillation outperforms teacher in genomic sequence modeling. Preprint at https://arxiv.org/abs/2505.20836 (2025).

  • Hu, J. et al. Comba: improving nonlinear RNNs with closed-loop control. Preprint at https://arxiv.org/abs/2505.12554 (2025).

  • Grešová, K., Martinek, V., Čechák, D., Šimeček, P. & Alexiou, P. Genomic Benchmarks: a collection of datasets for genomic sequence classification. BMC Genom. Data 24, 25 (2023).

  • Zhou, J. & Troyanskaya, O. G. Predicting effects of noncoding variants with deep learning-based sequence model. Nat. Methods 12, 931–934 (2015).

    Article 

    Google Scholar 

  • Li, J., Pu, Y., Tang, J., Zou, Q. & Guo, F. DeepATT: a hybrid category attention neural network for identifying functional effects of DNA sequences. Brief. Bioinform. 22, bbaa159 (2021).

    Article 

    Google Scholar 

  • Li, Z. et al. A novel interpretable deep learning-based computational framework designed synthetic enhancers with broad cross-species activity. Nucleic Acids Res. 52, 13447–13468 (2024).

    Article 

    Google Scholar 

  • Cheng, W. et al. DNALONGBENCH: a benchmark suite for long-range DNA prediction tasks. Nat. Commun. 16, 10108 (2025).

    Article 

    Google Scholar 

  • Avsec, Ž. et al. Effective gene expression prediction from sequence by integrating long-range interactions. Nat. Methods 18, 1196–1203 (2021).

    Article 

    Google Scholar 

  • de Almeida, B. P., Reiter, F., Pagani, M. & Stark, A. DeepSTARR predicts enhancer activity from DNA sequence and enables the de novo design of synthetic enhancers. Nat. Genet. 54, 613–624 (2022).

    Article 

    Google Scholar 

  • Gosai, S. J. et al. Machine-guided design of cell-type-targeting cis-regulatory elements. Nature 634, 1211–1220 (2024).

    Article 

    Google Scholar 

  • Fishman, V. et al. GENA-LM: a family of open-source foundational DNA language models for long sequences. Nucleic Acids Res. 53, gkae1310 (2025).

    Article 

    Google Scholar 

  • Linder, J., Srivastava, D., Yuan, H., Agarwal, V. & Kelley, D. R. Predicting RNA-seq coverage from DNA sequence as a unifying model of gene regulation. Nat. Genet. 57, 949–961 (2025).

    Article 

    Google Scholar 

  • Ahmed, F. S., Aly, S. & Liu, X. EPI-Trans: an effective transformer-based deep learning model for enhancer promoter interaction prediction. BMC Bioinform. 25, 216 (2024).

    Article 

    Google Scholar 

  • Feng, H. et al. Benchmarking DNA foundation models for genomic and genetic tasks. Nat. Commun. 16, 10780 (2025).

    Article 

    Google Scholar 

  • Van Der Harst, P. & Verweij, N. Identification of 64 novel genetic loci provides an expanded view on the genetic architecture of coronary artery disease. Circ. Res. 122, 433–443 (2018).

    Article 

    Google Scholar 

  • Medina, I. et al. Hck/Fgr kinase deficiency reduces plaque growth and stability by blunting monocyte recruitment and intraplaque motility. Circulation 132, 490–501 (2015).

    Article 

    Google Scholar 

  • Bobryshev, Y. V. Monocyte recruitment and foam cell formation in atherosclerosis. Micron 37, 208–222 (2006).

    Article 

    Google Scholar 

  • Li, J. et al. Novel insights: dynamic foam cells derived from the macrophage in atherosclerosis. J. Cell. Physiol. 236, 6154–6167 (2021).

    Article 

    Google Scholar 

  • Heinz, S. et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell 38, 576–589 (2010).

    Article 

    Google Scholar 

  • Beltagy, I., Peters, M. E. & Cohan, A. Longformer: the long-document transformer. Preprint at https://arxiv.org/abs/2004.05150 (2020).

  • Schneider, V. A. et al. Evaluation of GRCh38 and de novo haploid genome assemblies demonstrates the enduring quality of the reference assembly. Genome Res. 27, 849–864 (2017).

    Article 

    Google Scholar 

  • Nguyen, E. et al. Sequence modeling and design from molecular to genome scale with Evo. Science 386, eado9336 (2024).

    Article 

    Google Scholar 

  • Su, J. et al. Roformer: enhanced transformer with rotary position embedding. Neurocomputing 568, 127063 (2024).

    Article 

    Google Scholar 

  • Mou, L. et al. Natural language inference by tree-based convolution and heuristic matching. In Proc. 54th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) (eds Erk, K. et al.) 130–136 (Association for Computational Linguistics, 2016).

  • Upadhyaya, A., Nejdl, W. & Fisichella, M. Harnessing empathy and ethics for relevance detection and information categorization in climate and COVID-19 tweets. In Proc. 33rd ACM International Conference on Information and Knowledge Management 4091–4095 (ACM, 2024).

  • Zhang, K. et al. EATN: an efficient adaptive transfer network for aspect-level sentiment analysis. IEEE Trans. Knowl. Data Eng. 35, 377–389 (2021).

    Google Scholar 

  • Morales-Brotons, D., Vogels, T. & Hendrikx, H. Exponential moving average of weights in deep learning: dynamics and benefits. Trans. Mach. Learn. Res. 2024, 1–27 (2024).

    Google Scholar 

  • Yang, C. et al. CrossDNA: V1.1.0. Zenodo https://doi.org/10.5281/zenodo.19876436 (2026).



  • Source link