人工智能在生物育种产品食用安全评价中的创新应用

doi:10.19586/j.2095-2341.2026.0032

生物技术进展 ›› 2026, Vol. 16 ›› Issue (4): 765-773.DOI: 10.19586/j.2095-2341.2026.0032

人工智能在生物育种产品食用安全评价中的创新应用

韩天意¹(), 高宇¹, 程楠¹, 贺晓云¹, 刘威辰², 付伟³, 黄昆仑¹(), 陈红³()

^1.中国农业大学食品科学与营养工程学院，北京 100083
^2.中国质量检验检测科学研究院，北京 100176
^3.农业农村部科技发展中心，北京 100176

收稿日期:2026-02-09 接受日期:2026-04-03 出版日期:2026-07-25 发布日期:2026-09-11
通信作者: 黄昆仑,陈红
作者简介:韩天意 E-mail: HanTianyi@cau.edu.cn
基金资助:
科技创新 2030—农业生物育种重大项目(2023ZD0406304)

Innovative Applications of Artificial Intelligence in Safety Assessment of Biological Breeding Products

Tianyi HAN¹(), Yu GAO¹, Nan CHENG¹, Xiaoyun HE¹, Weichen LIU², Wei FU³, Kunlun HUANG¹(), Hong CHEN³()

^1.College of Food Science and Nutritional Engineering，China Agricultural University，Beijing 100083，China
^2.Chinese Academy of Quality and Inspection & Testing，Beijing 100176，China
^3.Development Center for Science and Technology，Ministry of Agriculture and Rural Affairs，Beijing 100176，China

Received:2026-02-09 Accepted:2026-04-03 Online:2026-07-25 Published:2026-09-11
Contact: Kunlun HUANG,Hong CHEN

摘要/Abstract

摘要：

生物育种产品的食用安全评价对防范潜在的毒性及过敏风险、保障消费者食品安全至关重要，对建立消费者信心、推动产业可持续发展必不可少。人工智能（artificial intelligence, AI）的快速发展为生物育种产品的安全评价提供了全新的技术路径与方法支撑，不仅可以做到食品成分的无损检测，还能显著提升安全评价的精准度与效率。综述分别从多模态数据来源、致敏性与毒理学评价、营养学评价和非期望效应评价四个层面，系统梳理了人工智能在转基因、基因编辑等生物育种产品的食用安全评价应用进展，以期为人工智能更全面地应用于食用安全评价提供理论支撑。

关键词: 转基因食品, 生物育种, 人工智能, 生物技术, 安全评价

Abstract:

Edible safety assessment of biologically breeding products is critical for preventing potential toxic and allergenic risks and protecting consumer safety. It is also indispensable for building consumer confidence and advancing the sustainable development of the industry. The rapid development of artificial intelligence （AI） has provided a brand-new technical path and method of support for the safety evaluation of biological breeding products. It not only enables non-destructive food testing but also significantly improves the accuracy and efficiency of the evaluation process. This article systematically reviewed the application progress of artificial intelligence in the edible safety evaluation of biological breeding products， such as genetically modified and gene-edited foods， from four aspects： multi-modal data sources， allergenicity and toxicological evaluation， nutritional evaluation， and adverse effect evaluation. The aim was to provide theoretical support for the more comprehensive application of artificial intelligence in edible safety evaluation.

Key words: genetically modified foods, biological breeding, artificial intelligence, biotechnology, safety assessment

中图分类号:

韩天意, 高宇, 程楠, 贺晓云, 刘威辰, 付伟, 黄昆仑, 陈红. 人工智能在生物育种产品食用安全评价中的创新应用[J]. 生物技术进展, 2026, 16(4): 765-773.

Tianyi HAN, Yu GAO, Nan CHENG, Xiaoyun HE, Weichen LIU, Wei FU, Kunlun HUANG, Hong CHEN. Innovative Applications of Artificial Intelligence in Safety Assessment of Biological Breeding Products[J]. Current Biotechnology, 2026, 16(4): 765-773.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://swjsjz.magtechjournal.com/CN/10.19586/j.2095-2341.2026.0032

https://swjsjz.magtechjournal.com/CN/Y2026/V16/I4/765

图/表 5

参考文献 50

[1]	COHEN S N, CHANG A C, BOYER H W, et al.. Construction of biologically functional bacterial plasmids in vitro [J]. Proc. Natl. Acad. Sci. USA, 1973, 70(11): 3240-3244.
[2]	KIM Y G, CHA J, CHANDRASEGARAN S. Hybrid restriction enzymes: zinc finger fusions to Fok I cleavage domain[J]. Proc. Natl. Acad. Sci. USA, 1996, 93(3): 1156-1160.
[3]	SAMUEL A L. Some studies in machine learning using the game of checkers[J]. IBM J. Res. Dev., 1959, 3(3): 210-229.
[4]	HINTON G E, SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786): 504-507.
[5]	ROBERTS L G. Machine perception of three-dimensional solids[D]. Cambridge: Massachusetts Institute of Technology, 1963.
[6]	MCKUSICK V A, RUDDLE F H. A new discipline, a new name, a new journal[J]. Genomics, 1987, 1(1): 1-2.
[7]	GYGI S P, RIST B, GERBER S A, et al.. Quantitative analysis of complex protein mixtures using isotope-coded affinity tags[J]. Nat. Biotechnol., 1999, 17(10): 994-999.
[8]	FIEHN O. Metabolomics: the link between genotypes and phenotypes[M]// Functional Genomics. Dordrecht: Springer Netherlands, 2002: 155-171.
[9]	NGUYEN M N, KRUTZ N L, LIMVIPHUVADH V, et al.. AllerCatPro 2.0: a web server for predicting protein allergenicity potential[J/OL]. Nucleic Acids Res., 2022, 50(W1): 36-43.
[10]	WANG W, MIN W, LI T, et al.. A review on vision-based analysis for automatic dietary assessment[J]. Trends Food Sci. Technol., 2022, 122: 223-237.
[11]	LI G, AN L, YANG W, et al.. Integrated biotechnological and AI innovations for crop improvement[J]. Nature, 2025, 643(8073): 925-937.
[12]	WANG X, XU K, HUANG Z, et al.. Accelerating promoter identification and design by deep learning[J]. Trends Biotechnol., 2025, 43(12): 3071-3087.
[13]	CHUAI G, MA H, YAN J, et al.. DeepCRISPR: optimized CRISPR guide RNA design by deep learning[J/OL]. Genome Biol., 2018, 19(1): 80[2026-03-20]. .
[14]	GAO C. Genome engineering for crop improvement and future agriculture[J]. Cell, 2021, 184(6): 1621-1635.
[15]	YANG Y, WANG Z. China food composition tables standard edition[M]. Beijing: Peking Univ. Med. Press, 2018.
[16]	MCCANCE R A, WIDDOWSON E M. McCance and widdowson's the composition of foods[M]. Cambridge: Royal Society Chem., 2014.
[17]	ROTHWELL J A, URPI-SARDA M, BOTO-ORDOÑEZ M, et al.. Phenol-Explorer 2.0: a major update of the Phenol-Explorer database integrating data on polyphenol metabolism and pharmacokinetics in humans and experimental animals[J/OL]. Database, 2012, 201210.1093: database[2026-03-20]. .
[18]	JUMPER J, EVANS R, PRITZEL A, et al.. Highly accurate protein structure prediction with AlphaFold[J]. Nature, 2021, 596(7873): 583-589.
[19]	GOODMAN R E, EBISAWA M, FERREIRA F, et al.. AllergenOnline: a peer-reviewed, curated allergen database to assess novel food proteins for potential cross-reactivity[J]. Mol. Nutr. Food Res., 2016, 60(5): 1183-1198.
[20]	HORAI H, ARITA M, KANAYA S, et al.. MassBank: a public repository for sharing mass spectral data for life sciences[J]. J. Mass Spectrom., 2010, 45(7): 703-714.
[21]	SUD M, FAHY E, COTTER D, et al.. Metabolomics Workbench: an international repository for metabolomics data and metadata, metabolite standards, protocols, tutorials and training, and analysis tools[J]. Nucleic Acids Res., 2016, 44(D1): 463-470.
[22]	KANEHISA M, FURUMICHI M, SATO Y, et al.. KEGG for taxonomy-based analysis of pathways and genomes[J/OL]. Nucleic Acids Res., 2023, 51(D1): 587-592.
[23]	WISHART D S, GUO A, OLER E, et al.. HMDB 5.0: the human metabolome database for 2022[J/OL]. Nucleic Acids Res., 2022, 50(D1): 622-631.
[24]	WISHART D S. FooDB: The Food Database[DB/OL]. [2026-03-22]. .
[25]	BOWMAN C C, CLAIN S M. Biotechnology methods for assessing potential allergenicity of novel proteins according to global regulations[M]//MALEKI S J, BURKS A W, HELM R M. Food Allergy. Boca Raton: CRC Press, 2017: 295-316.
[26]	AHUJA V, QUATCHADZE M, AHUJA V, et al.. Evaluation of biotechnology-derived novel proteins for the risk of food-allergic potential: advances in the development of animal models and future challenges[J]. Arch. Toxicol., 2010, 84(12): 909-917.
[27]	KLEINSTREUER N, HARTUNG T. Artificial intelligence (AI): it's the end of the tox as we know it (and I feel fine)[J]. Arch. Toxicol., 2024, 98(3): 735-754.
[28]	LIU J, NEGI S S, YANG C, et al.. AllergenAI: a deep learning model predicting allergenicity based on protein sequence[J/OL]. BMC Bioinform., 2025, 26(1): 279[2026-03-20]. .
[29]	PALAZZOLO L, LAURENZI T, MARIEM OBEN, et al.. Development of in silico methodologies to predict the toxicity of novel proteins in the context of food and feed risk assessment[J/OL]. EFSA Support. Publ., 2024, 21(10): 9063E[2026-03-20]. .
[30]	MAYR A, KLAMBAUER G, UNTERTHINER T, et al.. DeepTox: toxicity prediction using deep learning[J/OL]. Front. Environ. Sci., 2016, 3: 80[2026-03-20]. .
[31]	XIONG G, WU Z, YI J, et al.. ADMETlab 2.0: an integrated online platform for accurate and comprehensive predictions of ADMET properties[J]. Nucleic Acids Res., 2021, 49(W1): 5-14.
[32]	BANERJEE P, ECKERT A O, SCHREY A K, et al.. ProTox-II: a webserver for the prediction of toxicity of chemicals[J]. Nucleic Acids Res., 2018, 46(W1): 257-263.
[33]	PALAZZOLO L, LAURENZI T, MARIEM OBEN, et al.. Development of in silico methodologies to predict the toxicity of novel proteins in the context of food and feed risk assessment[J/OL]. EFSA Support. Publ., 2024, 21(10): 9063E[2026-03-20]. .
[34]	NIELSEN S S. Food Analysis[M]. Chambridge: Springer International Publishing, 2017.
[35]	MÆHRE H, DALHEIM L, EDVINSEN G, et al.. Protein determination: method matters[J/OL]. Foods, 2018, 7(1): 5[2026-03-20]. .
[36]	SUBHI M A, ALI S H, MOHAMMED M A. Vision-based approaches for automatic food recognition and dietary assessment: a survey[J]. IEEE Access, 2019, 7: 35370-35381.
[37]	ASMARA N, SAPUTRO A H. Animal and plant-based milk identification system using hyperspectral imaging and convolutional neural network[J/OL]. Adv. Biolog. Sci. Res., 2021, 2022: 049[2026-03-20]. .
[38]	SHAO W, HOU S, JIA W, et al.. Rapid non-destructive analysis of food nutrient content using swin-nutrition[J/OL]. Foods, 2022, 11(21): 3429[2026-03-20]. .
[39]	PUROHIT P, RAWAT H, VERMA N, et al.. Analytical approach to assess anti-nutritional factors of grains and oilseeds: a comprehensive review[J/OL]. J. Agric. Food Res., 2023, 14: 100877[2026-03-20]. .
[40]	NAHA S, KAUR S, BHATTACHARYA R, et al.. ANPS: machine learning based server for identification of anti-nutritional proteins in plants[J/OL]. Funct. Integr. Genom., 2024, 24(6): 201[2026-03-20]. .
[41]	ZHAO D, WANG C, QU H, et al.. Glycomol: a pervasive tool for structure predication of natural saponin products basing on MS data[J/OL]. J. Pharm. Anal., 2024, 14(7): 100897.
[42]	US. FOOD AND DRUG ADMINISTRATION (FDA). Statement of policy: foods derived from new plant varieties[EB/OL]. 1992 [2026-03-22]. .
[43]	HSU C Y, ASKAR S, ALSHKARCHY S S, et al.. AI-driven multi-omics integration in precision oncology: bridging the data deluge to clinical decisions[J/OL]. Clin. Exp. Med., 2025, 26(1): 29[2026-03-20]. .
[44]	FERNANDEZ A, CLARE MILLS E N, KONING F, et al.. Allergenicity assessment of novel food proteins: what should be improved?[J]. Trends Biotechnol., 2021, 39(1): 4-8.
[45]	HE X, LIU X, ZUO F, et al.. Artificial intelligence-based multi-omics analysis fuels cancer precision medicine[J]. Semin. Cancer Biol., 2023, 88: 187-200.
[46]	YETGIN A. Revolutionizing multi-omics analysis with artificial intelligence and data processing[J/OL]. Quant. Biol., 2025, 13(3): e70002[2026-03-20]. .
[47]	ZHANG J, FENG X, JIN J, et al.. Concise cascade methods for transgenic rice seed discrimination using spectral phenotyping[J/OL]. Plant Phenomics, 2023, 5: 71[2026-03-20]. .
[48]	CICERO RIBEIRO M, CABRAL J, NICOLODELLI G, et al.. Discrimination of maize transgenic and non-transgenic varieties by laser induced spectroscopy (LIBS) and machine learning algorithms[J/OL]. Microchem. J., 2024, 203: 110898[2026-03-20]. .
[49]	JUHAS M, RODEKOHR B, BAUER-PANSKUS A, et al.. Combining AI and new genomic techniques to 'fine-tune' plants: challenges in risk assessment[J/OL]. Front. Plant Sci., 2025, 16: 1677066[2026-03-20]. .
[50]	EFSA Panel on Genetically Modified Organisms, MULLINS E, BRESSON J L, et al.. Criteria for risk assessment of plants produced by targeted mutagenesis, cisgenesis and intragenesis[J/OL]. EFSA J., 2022, 20(10): e07618[2026-03-20]. .

应用领域	数据库/工具名称	主要特征与应用价值
食品营养学评估	USDA Food Data Central	美国农业部发布，包含标准参考、品牌及实验食品数据，是科研分析的首选
	中国食物成分表	针对中国本土饮食习惯，记录特有食材营养数据，更具地域适用性
	McCance and Widdowson's	英国/欧洲代表性数据库，数据质量极高，常用于跨国营养研究
	Phenol-Explorer	专注植物生化物质（多酚），记录加工过程（煎炸煮等）对含量的影响
致敏与毒性评估	AlphaFold	预测蛋白质三维结构，弥补晶体结构数据缺失
	WHO/IUIS & AllergenOnline	权威过敏原数据库，用于序列比对和风险关联分析
	MassBank / Metabolomics Workbench	公共质谱数据库，用于复杂基质中未知代谢物、过敏原及污染物的鉴定
非期望效应	KEGG / HMDB / FooDB	生物代谢通路及化合物库，用于识别非自然变异范围的成分变化

应用领域	数据库/工具名称	主要特征与应用价值
食品营养学评估	USDA Food Data Central	美国农业部发布，包含标准参考、品牌及实验食品数据，是科研分析的首选
	中国食物成分表	针对中国本土饮食习惯，记录特有食材营养数据，更具地域适用性
	McCance and Widdowson's	英国/欧洲代表性数据库，数据质量极高，常用于跨国营养研究
	Phenol-Explorer	专注植物生化物质（多酚），记录加工过程（煎炸煮等）对含量的影响
致敏与毒性评估	AlphaFold	预测蛋白质三维结构，弥补晶体结构数据缺失
	WHO/IUIS & AllergenOnline	权威过敏原数据库，用于序列比对和风险关联分析
	MassBank / Metabolomics Workbench	公共质谱数据库，用于复杂基质中未知代谢物、过敏原及污染物的鉴定
非期望效应	KEGG / HMDB / FooDB	生物代谢通路及化合物库，用于识别非自然变异范围的成分变化

评价对象	模型/工具名称	核心算法/技术	优势与性能表现
大分子蛋白质	AllergenAI	CNN（局部特征）+Bi-LSTM（长程依赖）	仅依赖序列即可量化致敏潜力，无需经验数据支撑
	AllerCatPro 2.0	结构同源建模(SWISS-MODEL)+线性表位滑动窗口	整合10个权威库，高灵敏度的同时降低25%假阳性率
	ToxinPred2/Toxify	机器学习/深度学习	欧盟食品安全局关注的工具，准确度与适用性较好
小分子物质	DeepTox	深度多任务学习(deep learning）	Tox21挑战赛优胜者，擅长预测核受体激活及应激反应
	ADMETlab 2.0	多任务图注意力框架(graph attention）	高效预测吸收、分布、代谢、排泄及毒性等80多个指标
	ProTox-Ⅱ	分子相似性+药效团+机器学习	快速筛查LD₅₀、肝毒性、致癌性及致突变性

评价对象	模型/工具名称	核心算法/技术	优势与性能表现
大分子蛋白质	AllergenAI	CNN（局部特征）+Bi-LSTM（长程依赖）	仅依赖序列即可量化致敏潜力，无需经验数据支撑
	AllerCatPro 2.0	结构同源建模(SWISS-MODEL)+线性表位滑动窗口	整合10个权威库，高灵敏度的同时降低25%假阳性率
	ToxinPred2/Toxify	机器学习/深度学习	欧盟食品安全局关注的工具，准确度与适用性较好
小分子物质	DeepTox	深度多任务学习(deep learning）	Tox21挑战赛优胜者，擅长预测核受体激活及应激反应
	ADMETlab 2.0	多任务图注意力框架(graph attention）	高效预测吸收、分布、代谢、排泄及毒性等80多个指标
	ProTox-Ⅱ	分子相似性+药效团+机器学习	快速筛查LD₅₀、肝毒性、致癌性及致突变性

检测目标	模型/研究案例	关键技术	改进效果/主要发现
宏量营养素	动植物源奶识别	高光谱成像(HSI) + 卷积神经网络（CNN）	耗时仅78 s，准确率达99.9%，实现快速无损检测
宏量营养素	Swin-Nutrition	近红外光谱（NIR）+分层视觉变换器（SwinT）	利用注意力机制捕获长程波段关联；蛋白质RMSE降低23.6%，优于传统CNN
抗营养因子	ANPS模型	理化特征（AAC/DPC）+支持向量机（SVM）	识别准确率94.31%，解决了低同源性蛋白难识别的问题
抗营养因子	Glycomol	超高性能液相色谱/高分辨率质谱（UPLC/HRMS）+图神经网络(GNN）	模拟糖链裂解反推结构，Top-1准确率88.4%，将解析时间从数天缩短至秒级

人工智能在生物育种产品食用安全评价中的创新应用

Innovative Applications of Artificial Intelligence in Safety Assessment of Biological Breeding Products

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 5

参考文献 50

相关文章 15

编辑推荐

Metrics

本文评价

评价层面	整合数据/技术	评价逻辑与突破
宏观预测与设计	生成式AI+气象大数据+AlphaFold	模拟G×E互作：预测极端气候下蛋白质折叠与代谢流向，早期干预代谢瓶颈与过敏风险
微观表型检测	高光谱表型组学+级联筛选	构象偏离：低成本精准捕捉种子内部蛋白质与淀粉的微观构象变化
元素代谢指纹	激光诱导击穿光谱+随机森林	元素异常：通过Mg/Ca/K光谱指纹，揭示转基因操作对矿物质代谢的非期望干扰
基因调控与预警	生成式AI+可解释AI(XAI）	蓝图设计：识别关键调控元件（CREs/uORFs），强调利用XAI解析非期望效应背后的分子机制

[1]	高宇, 韩天意, 贺晓云, 程楠, 刘威辰, 付伟, 黄昆仑, 陈红. 多组学系统分析生物育种产品非期望效应的研究进展[J]. 生物技术进展, 2026, 16(4): 732-743.
[2]	贾孟瑶, 胡得琴, 李源, 王冬梅, 路伟, 潘洪生. 我国转基因棉花的研发进展及其环境安全评价[J]. 生物技术进展, 2026, 16(4): 774-785.
[3]	任翔, 杨明超, 章寅, 陈一帆, 晁盛茜, 陈胜, 姜玉萍, 吕贝贝. 基于植物微生物系统的转基因植物生态安全评价研究进展[J]. 生物技术进展, 2026, 16(4): 786-796.
[4]	魏攀, 曹培健, 陈千思. 基于文献计量学的烟草生物技术研究现状与热点分析[J]. 生物技术进展, 2026, 16(4): 941-951.
[5]	黎娟, 贺玉婷, 冶晓轩, 贾佳仪, 张耀辉, 张宏扬, 武陶. 微生物细胞工厂合成类胡萝卜素的代谢工程策略与应用进展[J]. 生物技术进展, 2026, 16(3): 523-534.
[6]	杜盈萱, 孙雨彤, 彭瑾瑜, 赵嘉昊, 董柯然, 高鑫源, 吴浩泽, 罗云耀, 穆小苹, 隋宏书. 非哺乳动物模型在预测毒理学中的应用：进展、挑战与展望[J]. 生物技术进展, 2026, 16(3): 610-617.
[7]	皇甫秉欣, 张瑞奇, 王腾, 吴广枫, 张领先, 梁晋刚, 贺晓云. 基于层次分析法的我国生物育种玉米产业化全链条风险评估[J]. 生物技术进展, 2025, 15(6): 992-1002.
[8]	徐振华, 高世伟, 高大伟, 于艳敏, 刘海英, 武洪涛, 张书利, 孙忠义, 王昕, 闫平. 生物育种科技创新助力种业发展及展望[J]. 生物技术进展, 2025, 15(4): 557-564.
[9]	高子雅, 皇甫秉欣, 李佳岢, 李昉倩, 马智昊, 庞洋, 梁晋刚, 贺晓云. 生物育种产业化风险识别及建议——基于种植者的实地调研[J]. 生物技术进展, 2025, 15(3): 456-465.
[10]	崔兆惠, 郭玲, 沈旭东, 林毅, 翟丽丽. 生物技术药物的免疫原性产生机制与控制策略[J]. 生物技术进展, 2025, 15(2): 212-219.
[11]	纪思佳, 张小雪, 翟丽丽, 董静, 李岩异, 常亮. 抗体药物筛选技术的研究进展[J]. 生物技术进展, 2025, 15(1): 43-49.
[12]	王晶, 关海涛, 张晓磊, 王堡槐, 刘宝海, 温洪涛. 农业基因编辑产品检测动态及发展趋势[J]. 生物技术进展, 2024, 14(5): 712-723.
[13]	张春雷. 病原微生物检测技术研究进展[J]. 生物技术进展, 2024, 14(2): 189-195.
[14]	曹克浩, 朱军利, 何华山, 徐慰倬. 专利法第四次修改对生物技术专利申请和产业发展的影响[J]. 生物技术进展, 2023, 13(5): 663-670.
[15]	雷豆豆, 马润然, 王伽伯, 孔维军. 食品中农药残留的新型生物检测技术研究进展[J]. 生物技术进展, 2023, 13(1): 1-10.