Advances in Artificial Intelligence for DNA Design

doi:10.13560/j.cnki.biotech.bull.1985.2025-0863

Abstract

Abstract:

DNA design, namely the targeted construction and optimization of genomic functional elements to meet specific performance requirements, has become a core enabling technology at the forefront of synthetic biology and precision breeding. Traditional design approaches are constrained by limited understanding of complex regulatory networks and the vastness of the sequence search space, making efficient and precise sequence innovation difficult. In recent years, artificial intelligence (AI), especially the integrated use of deep generative and predictive models, has been reshaping the theoretical foundations and technical paradigms of DNA design. By learning the "regulatory grammar" embedded in massive omics datasets, these methods enable high-resolution functional prediction, multimodal data integration, and condition-controlled sequence generation within ultra-long genomic contexts. This article systematically reviews cutting-edge advances in AI for DNA design, with an emphasis on key technical pathways and applications of deep generative and predictive models in the design of multi-level regulatory elements such as promoters and enhancers, sequence optimization, and crop breeding. By establishing an intelligent closed loop of “design–prediction–optimization–validation”, AI not only markedly improves the efficiency and accuracy of designing complex functional elements, but also gives rise to synthetic sequences that outperform their natural counterparts. Looking ahead, as AI further converges with synthetic biology and experimental automation, DNA design is poised to achieve a full pipeline from intelligent design to high-throughput experimental validation, thereby accelerating breakthroughs in basic life science research and modern agricultural breeding.

Key words: artificial intelligence, DNA design, generative model, predictive model

LIU Huan, GUO Fa-xu, ZHAO Xiao-yan, HUANG Long-yu, WANG Jian, ZHOU Guo-min, ZHANG Jian-hua. Advances in Artificial Intelligence for DNA Design[J]. Biotechnology Bulletin, 2026, 42(1): 51-66.

Figures/Tables 7

References 50

[1]	Eraslan G, Avsec Ž, Gagneur J, et al. Deep learning: new computational modelling techniques for genomics [J]. Nat Rev Genet, 2019, 20(7): 389-403.
[2]	Kelley DR, Reshef YA, Bileschi M, et al. Sequential regulatory activity prediction across chromosomes with convolutional neural networks [J]. Genome Res, 2018, 28(5): 739-750.
[3]	Agarwal V, Shendure J. Predicting mRNA abundance directly from genomic sequence using deep convolutional neural networks [J]. Cell Rep, 2020, 31(7): 107663.
[4]	Zhang PC, Wang HC, Xu HW, et al. Deep flanking sequence engineering for efficient promoter design using DeepSEED [J]. Nat Commun, 2023, 14: 6309.
[5]	Wu MR, Nissim L, Stupp D, et al. A high-throughput screening and computation platform for identifying synthetic promoters with enhanced cell-state specificity (SPECS) [J]. Nat Commun, 2019, 10: 2880.
[6]	Yu TC, Liu WL, Brinck MS, et al. Multiplexed characterization of rationally designed promoter architectures deconstructs combinatorial logic for IPTG-inducible systems [J]. Nat Commun, 2021, 12: 325.
[7]	Ji YR, Zhou ZH, Liu H, et al. DNABERT: pre-trained Bidirectional Encoder Representations from Transformers model for DNA-language in genome [J]. Bioinformatics, 2021, 37(15): 2112-2120.
[8]	Linder J, Bogard N, Rosenberg AB, et al. A generative neural network for maximizing fitness and diversity of synthetic DNA and protein sequences [J]. Cell Syst, 2020, 11(1): 49-62.e16.
[9]	Zhou ZH, Ji YR, Li WJ, et al. DNABERT-2: Efficient foundation model and benchmark for multi-species genomes [J]. arXiv.org. 2023. DOI: 10.48550/arxiv.2306.15006 .
[10]	Avsec Ž, Weilert M, Shrikumar A, et al. Base-resolution models of transcription-factor binding reveal soft motif syntax [J]. Nat Genet, 2021, 53(3): 354-366.
[11]	Wang Y, Wang HC, Wei L, et al. Synthetic promoter design in Escherichia coli based on a deep generative network [J]. Nucleic Acids Res, 2020, 48(12): 6403-6412.
[12]	Killoran N, Lee LJ, Delong A, et al. Generating and designing DNA with deep generative models [J]. arXiv.org. 2017. DOI: 10.48550/arxiv.1712.06148 .
[13]	Gupta A, Zou J. Feedback GAN for DNA optimizes protein functions [J]. Nat Mach Intell, 2019, 1(2): 105-111.
[14]	Wang XL, Xu KJ, Tan YM, et al. Deep learning-assisted design of novel promoters in Escherichia coli [J]. Adv Genet, 2023, 4(4): 2300184.
[15]	Alipanahi B, Delong A, Weirauch MT, et al. Predicting the sequence specificities of DNA-and RNA-binding proteins by deep learning [J]. Nat Biotechnol, 2015, 33(8): 831-838.
[16]	Wang XL, Xu KJ, Huang ZS, et al. Accelerating promoter identification and design by deep learning [J]. Trends Biotechnol, 2025.
[17]	Barbadilla-Martínez L, Klaassen N, van Steensel B, et al. Predicting gene expression from DNA sequence using deep learning models [J]. Nat Rev Genet, 2025, 26(10): 666-680.
[18]	Quang D, Xie XH. DanQ: a hybrid convolutional and recurrent deep neural network for quantifying the function of DNA sequences [J]. Nucleic Acids Res, 2016, 44(11): e107.
[19]	Jores T, Tonnies J, Wrightsman T, et al. Synthetic promoter designs enabled by a comprehensive analysis of plant core promoters [J]. Nat Plants, 2021, 7(6): 842-855.
[20]	Zhou J, Troyanskaya OG. Predicting effects of noncoding variants with deep learning-based sequence model [J]. Nat Meth, 2015, 12(10): 931-934.
[21]	Kelley DR, Snoek J, Rinn JL. Basset: learning the regulatory code of the accessible genome with deep convolutional neural networks [J]. Genome Res, 2016, 26(7): 990-999.
[22]	Li ZH, Zhang YY, Peng B, et al. A novel interpretable deep learning-based computational framework designed synthetic enhancers with broad cross-species activity [J]. Nucleic Acids Res, 2024, 52(21): 13447-13468.
[23]	Yin C, Castillo-Hair S, Byeon GW, et al. Iterative deep learning design of human enhancers exploits condensed sequence grammar to achieve cell-type specificity [J]. Cell Syst, 2025, 16(7): 101302.
[24]	Vaishnav ED, de Boer CG, Molinet J, et al. The evolution, evolvability and engineering of gene regulatory DNA [J]. Nature, 2022, 603(7901): 455-463.
[25]	Fu HG, Liang YB, Zhong XQ, et al. Codon optimization with deep learning to enhance protein expression [J]. Sci Rep, 2020, 10: 17617.
[26]	Yang G, Chen YJ, Guo QH, et al. Leveraging pre-trained AI models for robust promoter sequence design in synthetic biology [J]. Swwlxb, 2025, 11: 1.
[27]	Fallahpour A, Gureghian V, Filion GJ, et al. CodonTransformer: a multispecies codon optimizer using context-aware neural networks [J]. Nat Commun, 2025, 16: 3205.
[28]	Zrimec J, Fu XZ, Muhammad AS, et al. Controlling gene expression with deep generative design of regulatory DNA [J]. Nat Commun, 2022, 13: 5099.
[29]	de Almeida BP, Schaub C, Pagani M, et al. Targeted design of synthetic enhancers for selected tissues in the Drosophila embryo. [J]. Nature, 2024, 626(7997): 207-211.
[30]	Avdeyev P, Shi CL, Tan YH, et al. Dirichlet diffusion score model for biological sequence generation [J]. arXiv.org. 2023. DOI: 10.48550/arxiv.2305.10699 .
[31]	de Almeida BP, Reiter F, Pagani M, et al. DeepSTARR predicts enhancer activity from DNA sequence and enables the de novo design of synthetic enhancers [J]. Nat Genet, 2022, 54(5): 613-624.
[32]	Avsec Ž, Agarwal V, Visentin D, et al. Effective gene expression prediction from sequence by integrating long-range interactions [J]. Nat Meth, 2021, 18(10): 1196-1203.
[33]	Yang Y, Lee JH, Poindexter MR, et al. Rational design and testing of abiotic stress-inducible synthetic promoters from poplar cis-regulatory elements [J]. Plant Biotechnol J, 2021, 19(7): 1354-1369.
[34]	Jain R, Jain A, Mauro E, et al. ICOR: improving codon optimization with recurrent neural networks [J]. BMC Bioinform, 2023, 24(1): 132.
[35]	Lei X, Wang X, Chen GL, et al. Combining diffusion and transformer models for enhanced promoter synthesis and strength prediction in deep learning [J]. mSystems, 2025, 10(4)
[36]	Kelley DR. Cross-species regulatory sequence activity prediction [J]. PLoS Comput Biol, 2020, 16(7): e1008050.
[37]	Li JQ, Zhang PC, Xi X, et al. Modeling and designing enhancers by introducing and harnessing transcription factor binding units [J]. Nat Commun, 2025, 16: 1469.
[38]	Gasperini M, Tome JM, Shendure J. Towards a comprehensive catalogue of validated and target-linked human enhancers [J]. Nat Rev Genet, 2020, 21(5): 292-310.
[39]	Zhou J, Theesfeld CL, Yao K, et al. Deep learning sequence-based ab initio prediction of variant effects on expression and disease risk [J]. Nat Genet, 2018, 50(8): 1171-1179.
[40]	Friedman RZ, Ramu A, Lichtarge S, et al. Active learning of enhancers and silencers in the developing neural retina [J]. Cell Syst, 2025, 16(1): 101163.
[41]	Karbalayghareh A, Sahin M, Leslie CS. Chromatin interaction-aware gene regulatory modeling with graph attention networks [J]. Genome Res, 2022, 32(7): 1290-1304.
[42]	Jaganathan K, Kyriazopoulou Panagiotopoulou S, McRae JF, et al. Predicting splicing from primary sequence with deep learning [J]. Cell, 2019, 176(3): 535-548.e24.
[43]	Sample PJ, Wang B, Reid DW, et al. Human 5' UTR design and variant effect prediction from a massively parallel translation assay [J]. Nat Biotechnol, 2019, 37(7): 803-809.
[44]	Bogard N, Linder J, Rosenberg AB, et al. A deep neural network for predicting and engineering alternative polyadenylation [J]. Cell, 2019, 178(1): 91-106.e23.
[45]	Lee NK, Tang ZQ, Toneyan S, et al. EvoAug: improving generalization and interpretability of genomic deep neural networks with evolution-inspired data augmentations [J]. Genome Biol, 2023, 24(1): 105.
[46]	Cherednichenko O, Poptsova M. Data augmentation with generative models improves detection of Non-B DNA structures [J]. Comput Biol Med, 2025, 184: 109440.
[47]	Davidi D, et al. Regulatory DNA sequence design with reinforcement learning [DB/OL]. arXiv preprint: 2503.07981.
[48]	Jaganathan K, Ersaro N, Novakovsky G, et al. Predicting expression-altering promoter mutations with deep learning [J]. Science, 2025, 389(6760): eads7373.
[49]	Dalla-Torre H, Gonzalez L, Mendoza-Revilla J, et al. Nucleotide Transformer: building and evaluating robust foundation models for human genomics [J]. Nat Meth, 2025, 22(2): 287-297.
[50]	张冀东, 王志晗, 刘博, 等. 深度学习在生物序列分析领域的应用进展 [J]. 北京工业大学学报, 2022, 48(8): 878-887.
	Zhang JD, Wang ZH, Liu B, et al. Progress in the applications of deep learning in biological sequence analysis [J]. J Beijing Univ Technol, 2022, 48(8): 878-887.

模型 Model	架构类型 Architecture type	主要用途 Primary applications	特点 Features	优势 Advantages	局限 Limitations
Enformer	Transformer	功能预测	长程依赖	泛化性强	数据需求量大
DNABERT	Transformer	序列特征提取	k-mer嵌入	泛化能力好	解释性有限
Evo/Evo2	Transformer	序列设计/预测	跨模态泛化生	零样本迁移	复杂度较高
GAN	对抗生成网络	序列设计/预测	成创新序列	功能多样	训练稳定性差
VAE	自编码器	序列优化	潜在空间优化	控制性强	泛化较难
Diffusion	扩散模型	序列精准生成	稳定性强	可控性强	计算资源需求高

模型 Model	架构类型 Architecture type	主要用途 Primary applications	特点 Features	优势 Advantages	局限 Limitations
Enformer	Transformer	功能预测	长程依赖	泛化性强	数据需求量大
DNABERT	Transformer	序列特征提取	k-mer嵌入	泛化能力好	解释性有限
Evo/Evo2	Transformer	序列设计/预测	跨模态泛化生	零样本迁移	复杂度较高
GAN	对抗生成网络	序列设计/预测	成创新序列	功能多样	训练稳定性差
VAE	自编码器	序列优化	潜在空间优化	控制性强	泛化较难
Diffusion	扩散模型	序列精准生成	稳定性强	可控性强	计算资源需求高

数据库/资源 Database/Resource	内容简介 Description	类型/范围 Type/Scope	用途 Applications
GENCODE	人类注释基因组	注释序列	基因注释
INSD	核酸数据集	序列库	通用序列建模
EPDnew	启动子信息	启动子功能	功能预测
EnhancerAtlas	增强子注释	增强子库	功能研究
MethBank	甲基化数据	表观遗传	调控分析
UCSC Browser	多组学数据	组学集成	多组学分析
GEO	转录组表观组	组学数据库	转录组分析
ENCODE	功能元件注释数据	多组学	功能元件注释
Roadmap	表观遗传组	表观组	功能研究
GTEx	组织表达图谱	转录组	表达分析
Ensembl	多物种注释	基因组注释	跨物种建模
RNAcentral	ncRNA库	非编码RNA	序列分析
GreeNC	植物ncRNA	植物数据库	功能元件设计
DBHR	多组学整合	综合组学	功能研究
UniProt	蛋白功能库	蛋白注释	功能预测
MPRA	高通量实验	功能平台	功能筛选

数据库/资源 Database/Resource	内容简介 Description	类型/范围 Type/Scope	用途 Applications
GENCODE	人类注释基因组	注释序列	基因注释
INSD	核酸数据集	序列库	通用序列建模
EPDnew	启动子信息	启动子功能	功能预测
EnhancerAtlas	增强子注释	增强子库	功能研究
MethBank	甲基化数据	表观遗传	调控分析
UCSC Browser	多组学数据	组学集成	多组学分析
GEO	转录组表观组	组学数据库	转录组分析
ENCODE	功能元件注释数据	多组学	功能元件注释
Roadmap	表观遗传组	表观组	功能研究
GTEx	组织表达图谱	转录组	表达分析
Ensembl	多物种注释	基因组注释	跨物种建模
RNAcentral	ncRNA库	非编码RNA	序列分析
GreeNC	植物ncRNA	植物数据库	功能元件设计
DBHR	多组学整合	综合组学	功能研究
UniProt	蛋白功能库	蛋白注释	功能预测
MPRA	高通量实验	功能平台	功能筛选

位置 Position	A密度 A density	C密度 C density	G密度 G density	T密度 T density
1	1.00	0.00	0.00	0.00
2	0.50	0.00	0.50	0.00
3	0.33	0.00	0.33	0.33
4	0.25	0.25	0.25	0.25
5	0.20	0.20	0.20	0.10
6	0.17	0.17	0.33	0.33
7	0.14	0.29	0.29	0.29
8	0.25	0.25	0.25	0.25